Home
MapQTL 5 Manual
Contents
1. 17 35 11 4 11 4 11 10 4 10 3 9 10 3 9 3 10 3 10 Index 57 results 4 10 results charts 4 11 session info 12 sessions 4 9 traits info 10 trait name 2 traits info 10 treeview 3 9 two lod support interval 14 unmapped 10 11 12 uppercase 39 use genotype probability approximation in rix 33 user interface 2 variable length 40 variance 15 whitespace 39
2. mu A mu B 2 when dominance could not be fitted it is given as 0 0 Expl the percentage of the variance explained for by the QTL 100 HO var var population_variance in which HO var is the residual variance under the current null hypothesis depends on cofactors used in MQM mapping GIC genotypic information coefficient see Genotypic information coefficient section of Mapping theory chapter p 33 GIC_1 genotypic information coefficient for the first parent GIC_2 genotypic information coefficient for the second parent GIC m mean of GIC_1 and GIC 2 Group the linkage group of the locus Iter the number of iterations needed to reach the tolerance criterium when this number is followed by an asterisk the maximum number of iterations was reached without satisfying the tolerance Locus the name of the locus at the current position LOD the LOD score Nr sequential number of the row Position the current position on the map mu_A the estimated mean of the distribution of the quantitative trait associated with the a genotype mu_B idem for the b genotype mu_H idem for the h genotype when no dominance was fitted mu H mu A mu B 2 mu_A 0 the mean associated with the a genotype with phase type 0 or with the b genotype with phase type 1 mu_B 0 idem for the b genotype with phase type 0 or for the a genotype with phase type 1 Pa idem for the ac genotype with phase type ce mu_ad
3. under the Windows versions ME NT 4 0 and XP and is further expected to run flawlessly under all current PC Windows platforms starting from 95 and above It comes with an InstallShield installation program that does most of the installation work Start the SETUP EXE program from the set of installation files e g by double clicking on it from within Windows Explorer or My Computer Choose the settings you are prompted for and let SETUP EXE finish After this process the license file MAPQTL LIC will be present in the program directory typically C Program Files MapQTL5 This is the evaluation license file which allows you to use the software with your own and demonstration data under certain limitations there are maxima of two populations two numerical traits per population and two linkage groups per map while printing copying to the clipboard and exporting to file are not available A purchased copy of MapQTL comes with your individual license file which usually resides in the Licenses directory of the product CD Replace the evaluation license file with your individual license file and make sure it gets the name MAPQTL LIC in the MapQTL Help menu there is an Install License function that can assist you with this Successful installation of the individual 2 Introduction license removes all above mentioned limitations and gives unrestricted access to the program the About box will show the name of the licensed organisation MapQTL 5 st
4. when plotting the results of an unmapped group of loci these will always show symbols and there will be no line connecting the points Clicking the Reset button on the charts Control page restores all chart options to the project default values The tabsheet on display or a selection of it can be exported to file Ea printed and copied to the clipboard using the corresponding File or Edit menu options or tool bar buttons File export and copying to clipboard are useful for taking the data or charts to for instance MS Excel or MS PowerPoint Charts are exported in the Enhanced Windows Meta File emf format which as an MS Windows standard can be used in many other applications When one or more rows in a table are selected not necessarily a contiguous set of rows or when there is a text selection in a plain text view selection is done in the regular MS Windows fashion the print export and copy functions are performed on the selection only Selections can also be dragged with the mouse and subsequently dropped into other accepting applications such as MS Excel or MS Word Prior to printing a preview of the print out can be obtained through the Print Preview option of the File menu or the tool bar button El From within the Print Preview and from the File menu the Page Setup and the Print Setup can be modified The Print Preview also allows the selection of pages for printing AS a nice navigation feature the selection of a
5. 00 idem for the ad genotype with phase type 00 mu_bc 00 idem for the bc genotype with phase type 00 mu_bd 00 idem for the bd genotype with phase type 00 Var the residual variance after fitting the QTL MQM mapping The MQM mapping method based on multiple QTL models was developed by Jansen 1993 1994 and Jansen amp Stam 1994 Although the definition of MQM mapping is very wide the current implementation in MapQTL is limited to using markers as cofactors in an approximate multiple QTL model with additive and dominant gene actions only Other uses of marker cofactors such as with gene by environment Jansen et al 1995 or gene by gene Fijneman et al 1996 interactions and the inclusion of the experimental design are quite difficult to implement in an easy user friendly manner in a general purpose mapping program Using a true multiple QTL model to detect and map QTLs would mean a multi dimensional search over the linkage groups At present this is computationally not really feasible The suggested approach is to first look for putative QTLs either by multiple regression preferably using backward elimination or by using interval mapping i e a single QTL model Care must be taken not to pick up so called ghost QTLs Martinez amp Curnow 1992 Next close to detected QTLs markers are selected as cofactors to take over the role of the nearby QTLs in the approximate multiple QTL models used in the Using MapQTL 17 subse
6. Co Inc Lancaster PA Chpt 7 pp 59 74 Van Ooijen J W 1999 LOD significance thresholds for QTL analysis in experimental populations of diploid species Heredity 83 613 624 Van Ooijen J W amp C Maliepaard 1996 MapQTL tm Version 3 0 Software for the calculation of QTL positions on genetic maps Plant Research International Wageningen the Netherlands Van Ooijen J W amp R E Voorrips 2001 JoinMap 3 0 Software for the calculation of genetic linkage maps Plant Research International Wageningen the Netherlands 54 Lists and references Van Ooijen J W M P Boer R C Jansen amp C Maliepaard 2002 MapQTL 4 0 Software for the calculation of QTL positions on genetic maps Plant Research International Wageningen the Netherlands Index 55 Index 14 default file name extensions 50 emf 11 degrees of freedom 13 18 mqd 3 8 demonstration data 3 mqp 3 8 deviance 15 18 32 acs 13 18 dominance 14 17 all markers mapping 33 enhanced windows meta file 4 11 analysis abbreviations 13 environment options 8 analysis options 8 example data files 8 analysis selector 11 execute analysis 8 automatic cofactor selection 18 exit program 8 calculate 3 12 expected segregation ratio 35 47 case sensitivity 39 export 4 11 chart 10 file charts 4 11 cof 39 classification type 41 45 cofactors 39 50 classification type codes 47 loc 39 40 cofactor 4 10 locus genotype 39 40 46 50 cofactor monitor 17 map 39
7. Example 3 Quantitative data file This file holds the data of the quantitative traits of all individuals It has a sequential structure The header of the file contains three instructions on the contents of the data body followed by the names of the traits The data body contains the actual information for each trait and for all individuals The three instructions define the numbers of traits and individuals and the text that indicates a missing value These instructions can be given in any order The syntax of the three instructions is ntrt NTRT nind NIND miss MISS where NTRT and NIND are the numbers of traits and individuals respectively and MISS is the missing value indicator i e a text string that is used to indicate missing values cannot contain spaces maximum length is 20 There are no maximum values for NTRT and NIND of course reading the file becomes time consuming at extreme values of NTRT but NIND must be equal to or larger than selective genotyping see below the value of NIND in the corresponding loc file These instructions must be followed by the names of all the traits after which the data body must be given It is allowed and even advisable see below to include non numerical traits After loading the data into a MapQTL project the non numerical traits will show up in green in the Populations tabsheet and no analysis can be done on such a trait The names of the traits may be up to 9 characters long canno
8. I Show Symbols Symbol Size p Figure 4 Results Charts tabsheet with subordinate Control tabsheet visible here for the simulated DemoF2 data set In the permutation test the significance threshold is determined on the actual data each iteration the quantitative trait values are permuted over the individuals thereby releasing any possible association with the markers Subsequently the permuted data are analysed by interval mapping and the maximum LOD scores are recorded By doing this repeatedly preferably at least 10 000 times because the results are quite variable the frequency distribution of the LOD is determined based on the actual data of which we are certain that there is not any association between any segregating QTL and a marker due to the permutations Try the permutation test but because it takes so long for all computations first set the number of permutations for this time to the low value of 100 e Use the Analysis Options of the Options menu e Set the number of permutations to 100 and click OK e Select the Permutation Test in the Analysis selector e Verify that the trait gtrait and the ten linkage groups are still specially selected if not restore this and subsequently click on the Calculate button Notice that new session nodes are created The calculations will take some time to complete e Inspect the results in the Results tabsheet Look up the interval value in the group GW at the relative cumulative v
9. Jansen 1994 Lander amp Kruglyak 1995 Kruglyak amp Lander 1995 Doerge amp Rebai 1996 For MapQTL extensive computer simulations were carried out to come towards a convenient method of calculating the appropriate threshold for the genome size and population type under study this research was published by Van Ooijen 1999 MapQTL also offers the permutation test for interval mapping with which the significance threshold can be determined based on the actual data rather than on assumed normally distributed data Data files 39 Data files General MapQTL uses plain text files to load the data that must be analysed A plain text file can be made with any text editor program MapQTL uses several types of data files each containing different kinds of information Besides the actual data the files contain instructions that guide the program through the information First there is the locus genotype file also called loc file which contains the genotype codes for the loci of a single segregating population Then there is the map file containing the map positions of all loci Thirdly the quantitative data file also called qua file which as the name suggests holds the data of the quantitative traits of all individuals And finally there is the cofactors file also called cof file containing the names of the markers that should be used as cofactors in MQM mapping analysis The loc file and the map file have the same formats as are
10. Menu and tool bar Bopulations wraps Sessions Genotypes Info Map Ipfo ession Mme Results Charts Results Project Info Project Notes Population NS Traits Info start a ew project and load data into it or ope an existing project Navigation panel Contents and results panel Status bar Figure 1 User interface Introduction 3 The QTL analysis is organised into so called MapQTL projects A project consists of a project file and a project directory both are and must be in the same directory The project directory will contain all files used internally by the program You can view these plain text files but it is strongly advised not to edit remove or rename them because that may damage the project so that it cannot be handled by MapQTL anymore copying is fine When creating a new project which is done using New Project function of the File menu or the New Project button LI you are prompted for a project file name with a standard save file dialog window a project file with the extension mqp and the corresponding project data directory of the same name with the extension mqd will be created Once a new project is opened you load data into the project This must be done with the Load Data function of the File menu or with a tool bar button W Data must be loaded from three separate files 1 the set of locus genotypes of a population 2 the set of quantitative trait da
11. as so called treeviews like the Folders panel in the Windows Explorer Populations their traits and genotypes and maps with their linkage groups are shown hierarchically as nodes in a tree The Populations tabsheet also has a Common traits node which will show all traits as its child nodes that are common to all loaded populations NB Traits within the quantitative trait data set that contain some non numerical data will show up as nodes with a green font and icon i e different from completely numerical traits as they cannot be used for analysis The nodes in the treeviews can be selected by clicking on them The names of the population map and session currently selected are shown in the three small status bars at the bottom of the navigation panel The selection of a node enables the inspection of its data in the corresponding tabsheet of the contents and results panel Selecting a linkage group child node under a map or session node puts the focus on the corresponding table position in the Map Info or Results tabsheets of the contents and results panel respectively The trait nodes and the linkage group nodes can also be specially selected by right clicking or by pressing the space bar when the node is selected i e usually blue This is a special type of selection which applies only to the analysis that is going to be performed on these traits and linkage groups As a result the nodes will show up in red or magenta for the selecte
12. become a different HO model At the start of each linkage group the HO model is always re calculated The set of cofactors for the selected traits must be chosen using the Map Info tabsheet see the Contents and results panel section p 10 In contrast to interval mapping in MQM mapping dominance is always fitted also for cofactors for an F2 population whereas it is never fitted for an RIx population an F2 may be analysed as an RI2 when no dominance is required Unmapped loci cannot be used in the analysis Selective genotyping is not possible MQM mapping output The output is similar to that of interval mapping so please see the Interval mapping output section p 15 Of course some additional information is given Loci used as cofactor are indicated with an X in the additinal Cofactor column on the Results tabsheet while they are listed with group and position information on the Session Info tabsheet This tabsheet also gives the name of the so called cofactor monitor output file this plain text file resides in the project directory The file lists at each calculated map position the estimated values of the regressors for each cofactor and of the means associated with the QTL genotypes at the map positions where the HO is calculated the output lists the estimates of the cofactor regressors and the overall mean see the MQM mapping section p 35 of the Mapping theory chapter for details The locus names are printed at corresp
13. data after the last individual and will issue a warning If NTRT is incorrect then for most individuals the wrong trait will be read this may remain undetected unless there is a trait that consists of text instead of numerical values in which case the unsuccessful attempt to interpret text as a numerical value will lead to an error message So you might consider including a text trait just for the sake of error detection It is important to note that it is absolutely essential that the order of the individuals is identical to the order in the loc file i e the X th individual in the loc file must be the same as the X th individual in the qua file In the case of selective genotyping i e of only a part of the population of which of all individuals a quantitative trait was determined the genotypes of the markers were determined the first individuals must correspond to the individuals in the loc file while the not genotyped individuals are appended Example 4 demonstrates a small qua file 50 Data files Cofactors file This file is used to feed MapQTL with the names of the loci to be used as cofactors in the MQM analysis It is line structured The header consists of one instruction only It is ncof NCOF where NCOF is the number of loci in the file This instruction is followed by the names of the loci each on a separate line Example 5 demonstrates a small cof file Default file name extensions For ease of use we have introd
14. in the cofactor monitor output file The presence of a single segregating QTL is tested by comparing the model with QTL to the nearly identical model but without QTL both models having the same cofactors The LOD score or deviance is used for this purpose The problem of incomplete genotypes flanking the QTL in the single QTL model of interval mapping of course also occurs for the single fitted QTL in the approximate multiple QTL model and is solved in the same fashion using the markers beyond the QTL flanking markers A new difficulty arises however when cofactor genotypes are incomplete This is solved by taking into account all possible genotypes given the incomplete genotype and calculating the probabilities of these so called complete genotypes using the markers linked to the cofactor If necessary this can be done for more than just one cofactor Subsequently the mixture model is extended into more components one for each combination of fitted QTL genotype and possible complete cofactor genotype The means of the component distributions are the separate 44 s adjusted with the genetic effects associated with the complete cofactor genotypes The component probabilities are determined by calculating the joint probabilities of the possible complete cofactor genotypes combined with the QTL genotypes For instance if in a model with one cofactor an F2 individual has a dominant cofactor genotype c then there are two possible genotypes for the
15. make your own notes about the project and which will be stored with the project In order to perform an analysis the trait or traits must be selected from the Populations tabsheet by right clicking their nodes or by pressing the space bar when the node is selected i e usually blue as a result the node will show up magenta or red This selection is a toggle i e right clicking again will deselect the node Selection of child nodes under the Common traits node automatically selects the trait within all populations In a similar fashion the linkage groups that must be analysed for the selected trait s must be selected by right clicking their nodes on the Maps tabsheet Once this is done and an analysis is selected on the tool bar the Calculate function will be enabled available both 4 Introduction as a menu option and as a tool bar button E and can be chosen or clicked respectively This will start a so called calculation session with a corresponding node in the Sessions tabsheet on the navigation panel Nodes will be created in the Sessions tabsheet for each analysed trait and linkage group with the appropriate hierarchy of the various nodes in the sessions treeview If the analysis requires marker cofactors they can be selected by checking their Cofactor checkboxes on the Map Info tabsheet of course the appropriate map must be selected in the Maps tabsheet The Cofactors Tool can be very helpful with this it is available from the Edi
16. probabilities 7 for each of the Q components are calculated for each individual in the population from the marker genotypes and the linkage map If for instance the current QTL position on the genome coincides with a marker and the marker genotype of the individual is completely known then one of the g component probabilities 7 equals 1 while the others are 0 If as an instance of another extreme the current position is in between markers while the genotypes of these two so called flanking markers of the individual are not completely known such as c in an F2 or hk of segregation type lt hkxhk gt ina CP then the genotypes of the linked markers beyond the flanking markers and their map positions are used to obtain the probabilities Thus genetic information from markers surrounding the current assumed QTL map position is used to calculate the most accurate values of the component probabilities In the mixture model the mixture density f x for individual n 1 N is the sum of the products of the component densities f x with their probabilities 7 Q FADET n 32 Mapping theory thus the likelihood for the population under the hypothesis that a QTL is segregating L is N N Q L f T X rf n 1 q 1 With the EM algorithm the likelihood or actually its logarithm is maximised and the parameters 4 and are estimated Dempster et al 1977 The EM algorithm is an iterative procedure in which at each iteration the l
17. scored none is missing the scores are 100 correct the map is accurate and there are 6 segregating QTLs with quite large effects In real life you will have to do with marker scores that contain unknown errors you will have missing marker scores and as a consequence some uncertainty or even errors in your linkage map The result will be that the QTL analysis will not be as straightforward as in this tutorial With MapQTL 5 you have quite a powerful tool to analyse the data that you have obtained from your experiments the software cannot however improve the quality of its input data that area remains your responsibility Mapping theory 31 Mapping theory Interval mapping The implemented QTL mapping procedure is a maximum likelihood approach to the segregation of a mixture of probability distributions cf Titterington et al 1985 McLachlan amp Basford 1988 Under the hypothesis that a single QTL is segregating thus a single QTL model the mixture consists of Q distributions components one for each QTL genotype g 1 Q with Q depending on the type of population CP Q 4 F2 RIx Q 3 the other types Q 2 The component distributions are assumed to be normal with means u and common variance o The mapping function of Haldane is assumed which means that recombination events are mutually independent Map distances between markers are taken as fixed For a given assumed position on the genome of a segregating QTL the
18. that go with the population type For population type CP it is more interesting to look at the two parents separately thus you have the first Mapping theory 35 and second parent G C s the program output also gives the mean of the two The first and second parent G C s have the same formula for populations with two genotypes above but here the QTL genotype probabilities are sums of two underlying genotype probabilities for the first parent GIC m T T ct Tp and 7 1 1 M and for the second parent GIC 1 1 T ct Te and MEN T pt Tp The resulting formula s are identical to those given by Knott et al 1997 though at first sight they may appear different Finally when there is selective genotyping see below the standard Mendelian expectations are used as the QTL genotype probabilities for the ungenotyped individuals this results in a maximum GIC value equal to the fraction of genotyped individuals Selective genotyping When the marker genotypes are determined of only a selected part of the segregating population e g the 20 individuals with the highest and the 20 with the smallest values of the quantitative trait this is called selective genotyping Generally it is considered to be a way of enhancing the power of QTL detection and mapping and also a cost reduction method However when there are several QTLs the employment of a multiple QTL model becomes problematic it is supposed that the genetic effects of the distinct
19. the null hypothesis i e there is no segregating QTL or perhaps better the segregating QTL has no effect the Kruskal Wallis statistic is distributed approximately as a chi square distribution with the number of genotype classes minus one as degrees of freedom e g 1 degree for a backcross 2 degrees for an F2 Since the test will generally be performed on many linked and unlinked loci it is prudent to use a stringent significance level P value for the individual tests in order to obtain an overall significance level of about 0 05 we suggest a level of at least 0 005 The linkage group with a segregating QTL must reveal a gradient in the test statistic towards the locus with the closest linkage to the QTL The power of the test depends on the degrees of freedom So for instance when codominant loci are combined with dominant loci the latter may show a smaller significance level even if they are more closely linked The power also depends on the number of individuals in the test Because the analysis can only be done on individuals for which both marker genotype and quantitative trait value are known differences between markers in numbers of individuals in the test will affect the gradient in the test statistic over the linkage group Nonparametric mapping output A summary of the parameters input and the data read from the files are given on the Session Info tabsheet The Results tabsheet lists the results under column headers with the fol
20. trait with the name nr this was entered in the original quantitative data file as the individual number of which the data happen to coincide here with the row number Nr e Any table in MapQTL can be sorted by clicking on the column header Try this by clicking on the gtrait column header click again and notice that the rows are sorted in the opposite direction notice that the largest gtrait value is 5 744531 To return to the original order click on the Nr column header Tutorial 23 e Select the Genotypes Info tabsheet This tabsheet shows a copy of the loaded loc file with the marker names and genotype scores for all individuals and markers e Select the Map Info tabsheet This tabsheet displays the details of the loaded map file the locus marker name its linkage group name usually a number and its map position As in any table in MapQTL there is a column Nr with the original row number Notice there is a column with check boxes this will be used later for indicating which markers are to be used as cofactors Also notice a column filled with X s under the header of the population name DemoF2 an X indicates if a marker of the map is present in the genotypes of the population when loading more populations this is especially useful to see in which populations each marker is determined e Select the Maps tabsheet on the navigation panel Click on the Group 4 node in the treeview and observe that on the Map Info tabsheet the row pointer
21. 1 1 1 ac ad bc and bd ee ef eg fg 1 1 1 1 ee ef eg and fg hh k 1 3 hh and k hk and kk will be included in class k h kk 3 1 h and kk hh and hk will be included in class h hh hk kk 1 2 1 hh hk and kk 11 1m 1 1 11 and 1m nn np 1 1 nn and np for RIx the ratios are adjusted according to the generation number x Table 9 Default and optional classification types Pop type Seg type Default Optional A classification type is NOT ALLOWED in the data file BC1 a h or h b none DH a b none DH1 a b none HAP a b none HAP1 a b none CP lt abxcd gt ac ad bc bd none lt efxeg gt ee ef eg fg none lt lmxll gt 11 1m none lt nnxnp gt nn np none Classification types are ALLOWED in the data file F2 a h b a c or b d RIx a b a h b a c or b d CP lt hkxhk gt hh hk kk h kk or hh k automatically determined 48 Data files Example 3 A map file the file is completely line structured group a j lt locus gt lt map position gt rapd0o2 0 0 rapd86 11 1 rapd0s t532 rapd22 17 3 group b rapd54 0 0 rapd66 15 2 rapd18 22 3 per line It is not required to start at map position 0 0 A following linkage group must start again with the group instruction Next to the group instruction MapQTL attempts to read a group name of up to twenty characters no spaces which if available will be used in the output A small map file is demonstrated in
22. 2 What happens if NIND or NLOC are incorrect If NIND is incorrect then MapQTL will try to interpret part of a locus name as a genotype code which in general will lead to an error message If NLOC is larger than the actual number of loci in the file then MapQTL will try to read beyond the end of the file which will also lead to an error message If NLOC is smaller than the actual number then it will issue a warning that there are more data in the file You might want to exploit this feature to park loci that you do not want to be used The data body contains the information for all loci and individuals grouped per locus The data group for a locus consists of the name of the locus followed by the genotype codes of all individuals In between the locus name and the genotypes there can optionally be up to three additional instructions depending on the type of population MapQTL is indifferent to the order of these instructions The instructions are concerned with the type of segregation of the locus SEG for population type CP the linkage phases of the locus PHASE for population types CP DH and HAP and the type of classification for the locus CLAS In short the syntax of a data group for a locus is optional is indicated with lt locus name gt SEG PHASE CLAS lt NIND genotypes gt It is important to note that it is absolutely essential that the order of the individuals is identical over all loci in the file The geno
23. 38 871 881 Jansen R C amp P Stam 1994 High resolution of quantitative traits into multiple loci via interval mapping Genetics 136 1447 1455 Jansen R C J W Van Ooijen P Stam C Lister amp C Dean 1995 Genotype by environment interaction in genetic mapping of multiple quantitative trait loci Theor Appl Genet 91 33 37 JoinMap http www joinmap nl Knott S A D B Neale M M Sewell amp C S Haley 1997 Multiple marker mapping of quantitative trait loci in an outbred pedigree of loblolly pine Theor Appl Genet 94 810 820 Kruglyak L amp E S Lander 1995 A nonparametric approach for mapping quantitative trait loci Genetics 139 1421 1428 Lander E S amp D Botstein 1989 Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps Genetics 121 185 199 Lists and references 53 Lander E S amp L Kruglyak 1995 Genetic dissection of complex traits guidelines for interpreting and reporting linkage results Nature Genetics 11 241 247 Lehmann E L 1975 Nonparametrics McGraw Hill New York Maliepaard C amp J W Van Ooijen 1994 QTL mapping in a full sib family of an outcrossing species In Van Ooijen J W amp J Jansen Eds Biometrics in Plant Breeding Applications of Molecular Markers Proceedings of the Ninth Meeting of the EUCARPIA Section Biometrics in Plant Breeding 6 8 July 1994 Wageningen the Netherlands pp 140 146 MapQTL http www mapg
24. 46 48 50 cofactors 16 output data 50 cofactors file 19 39 50 project 7 8 50 cofactors tool 4 10 qua 39 cof file 39 quantitative data 39 48 49 50 comment line 40 file name extensions 50 common traits 3 9 find tool 11 component 31 fixed length 40 contents and results panel 2 7 9 10 functional tolerance value 15 32 control 4 11 genome wide 12 copy to clipboard 4 11 genotype codes 39 41 43 45 ctrl a 4 genotypes 12 data file characteristics 39 genotypes info 10 data files 39 genotypic information coefficient 33 default 11 gic 33 default classification type 45 group name 2 48 default classification types 47 group instruction 48 56 HO 14 17 H1 14 17 header 40 help menu 4 8 im 13 14 installation 1 interval mapping 14 31 key combinations 7 keyboard shortcuts 7 kruskal wallis 12 kurtosis 15 kw 12 13 layout 40 length of names 2 40 license file 1 likelihood 14 likelihood ratio test statistic 32 line structured 40 linkage group 11 linkage groups 9 linkage phase 41 44 46 load data 3 8 loc file 39 40 locus genotype file 39 40 46 50 locus name 2 lod 14 32 lod significance threshold 38 lowercase 39 map file 39 46 48 50 map info 4 10 mapping function 31 mapping step size 15 mapqtl project 3 8 mapdtl lic 1 mapqtl5 exe 7 maps 3 9 maximum length of names 2 40 maximum number of iterations 15 32 maximum number of neighbouring markers used 15 33 memory 2 miss instruction 48 Index modifying mixin
25. Loaded locus genc date time added under por original pop r population type nr of loci nr of individu G3 Group 10 Loaded quantitati Loaded quantitati Figure 2 Populations tabsheet with the DemoF2 population Figure 3 Maps tabsheet with the DemoF2 map e Select the Maps tabsheet click on the Load Data button and select the map file DemoF 2 map the navigation panel will resemble Figure 3 notice there are ten linkage groups The project tutorial now has the basic data loaded Let s have a look at the contents and results panel and see what the contents of the data are e First click on the DemoF2 nodes in the Populations panel and the Maps panel just to make sure the population node and map node are selected in the treeviews e Select the Project Info tabsheet Here you can see when the project was created and what data sets were loaded including a summary of the data sets e Select the Project Notes tabsheet This tabsheet is empty when you click in it you can start entering notes these will be stored with the project e Select the Population Info tabsheet This tabsheet shows a summary of the population currently selected i e DemoF2 e Select the Traits Info tabsheet This tabsheet has a table with all numerical traits data of the currently selected population each trait in its own column The table includes a first column Nr for the original row number in the table There is also a numerical
26. MapQTL 5 Software for the mapping of quantitative trait loci in experimental populations J W van Ooijen Wageningen February 2004 MapQTL is developed in collaboration with statistical geneticists of Biometris of Wageningen UR www biometris nl The sales and support are taken care of by Kyazma B V Copyright 1996 2004 Plant Research International B V and Kyazma B V All rights reserved Unauthorized reproduction and distribution prohibited MapQTL and JoinMap are a trademarks of Plant Research International B V and Kyazma B V registered in the Benelux and the U S A Other brand and product names are registered trademarks of their respective holders Kyazma B V P O Box 182 6700 AD Wageningen support kyazma nl Netherlands www kyazma nl Contents Introduction 1 Installation 1 Overview 2 Final remarks 4 How to cite MapQTL 5 5 Acknowledgement 5 Using MapQTL 7 Controlling the program 7 The MapQTL project 8 Navigation panel 9 Contents and results panel 10 Starting an analysis 11 Nonparametric mapping Kruskal Wallis analysis 12 Nonparametric mapping output 13 Interval mapping 14 Interval mapping output 15 MQM mapping 16 MQM mapping output 17 Automatic selection of cofactors 18 Permutation test 19 Permutation test output 20 Tutorial 21 Mapping theory 31 Interval mapping 31 Genotypic information coefficient 33 Selective genotyping 35 MQM mapping 35 LOD significance threshold 38 Data files 39 Gener
27. V Wageningen Netherlands Acknowledgement All new versions of software programs build on their predecessors MapQTL 5 is no exception the main contributors to version 4 0 are gratefully acknowledged for their input Martin Boer Ritsert Jansen and Chris Maliepaard To the present version several people of Wageningen University and Research Centre contributed with wish lists remarks positive criticism software testing and alike Richard Finkers Sjaak van Heusden Hans Jansen Piet Stam Roeland Voorrips their assistance is greatly appreciated Introduction Using MapQTL 7 Using MapQTL The program can be started in the various ways of MS Windows by using the Start menu by double clicking on the MapQTL5 exe file from within Windows Explorer or My Computer or by double clicking on a project file The latter way is established only after running the program a first time When the program runs you will see a window that is divided into several main parts on the top the menu and the tool bar with buttons and a selector on the left hand side the navigation panel on the right hand side the contents and results panel and on the bottom the status bar Figure 1 Once a project is created and data are loaded the navigation panel will show the populations with their traits and genotypes and the maps with their linkage groups The contents and results panel contains a set of tabbed pages tabsheets in which contents of data sets and resu
28. al 39 Data file characteristics 39 Locus genotype file 40 Map file 46 Quantitative data file 48 Cofactors file 50 Default file name extensions 50 Lists and references 51 List of figures 51 List of tables 51 List of examples 51 References 52 Index 55 Introduction 1 Introduction MapQTL is a computer program for the calculation of QTL positions on genetic linkage maps in experimental populations of diploid species The present version 5 is based on its predecessor version 4 0 Van Ooijen et al 2002 of which the user interface is completely revised giving more ease of use better QTL charts and improved exportability Multiple populations and maps can now be loaded into a project thereby allowing an easier comparison of results across related populations or using different maps Results of analyses are stored in so called sessions within the project these sessions can be inspected immediately after the computations and stay available for later re inspection A very important enhancement is the creation of QTL charts by the program in which charts of all or a selection of linkage groups can be combined on a single page and many options are easily controlled The results and their charts can be exported to files copied to other MS Windows programs like MS Word or MS Excel and printed and there is also a preview prior to printing Installation MapQTL is a program for the MS Windows platform on the PC It was tested to run
29. alue Rel cum count column of 0 9500 or a value close See the Permutation test section p 19 for details on how to deal with these results Redo this permutation test a few times and make notes of each estimated threshold 26 Tutorial value You will notice that it varies around the 3 8 value calculated above which is a correct value because the DemoF2 data are based upon a simulated normal distribution and thus agrees with the required assumption of normality We decide we want to use the 3 8 LOD as a genome wide significance threshold Now let s go back to the interval mapping results charts e Select the Control tabsheet On the Options 1 tabsheet enter the above threshold value 3 8 in the field Show Horizontal Dotted Line at Left Y axis Value e Check the Show Loci option on the Options 1 tabsheet Select the Charts tabsheet again and notice that it is easy to spot the regions with a significant LOD score groups 1 3 4 and 5 have significant scores group 2 is just below the threshold on the Results tabsheet significant it shows a maximum LOD of 3 43 at locus m34 e Find the markers closest to the highest LOD scores on these groups 1 3 4 and 5 and verify this using the Results tabsheets These are m9 m54 m75 and m97 Also take notice of the explained variance Expl at these markers In interval mapping the association at a certain map position is tested against the residual variance the larger the genetic effect assoc
30. and further only m33 m53 m75 and m96 e Set the Analysis selector on Automatic Cofactor Selection and click on the Calculate button Automatic cofactor selection uses a backward elimination procedure to see which markers show a significant association and which do not all not significant markers are removed so you end up with only significant cofactor markers When you inspect the results you will see that on group 1 of all markers the two markers m9 and m13 remain as being significant while the cofactor markers on the other linkage groups also remain as significant Now you would like to redo the MQM mapping analysis but with the Group 1 40 307 20 Figure 5 LOD profile on linkage group 1 10 with LOD values above the significance Ea E Re cpestecins sti threshold 3 8 some distance away from 0 4 the cofactor marker m9 28 Tutorial present set of cofactors e Make a note from the Results tabsheet of the file name under which the selected set of cofactors is stored it should be something like Session 5 ACS DemoF2_qtrait cof and it should reside in the project directory Click on the Cofactors Tool button M Select the action Load cofactor setting from file and click on the Do it button You are asked whether you wish to clear the currently checked loci choose Yes Subsequently you will get a dialog in which you are prompted for a cofactors file point the dialog to the p
31. aper give you the value of 3 8 as the LOD significance threshold This means more formally the probability that the LOD score is above this threshold value just by chance rather than by a segregating QTL anywhere on the genome is 5 More practically this means that you will conclude that a QTL is present when the LOD is above the value 3 8 Another way of getting the significance threshold is to do the permutation test This is often thought of to be more correct because the method in the paper by Van Ooijen 1999 is based on the assumption that the trait is distributed according to the normal distribution and this might not be true for the data you are analysing however it is true Tutorial 25 Project Info Project Notes Population Info Traits Info Genotypes Info Map Info Session Info Results Charts Results Groups if Let Y axis Right Y axis Control tab LOD Lob lter lter Charts tab mu_A mu_A mu_H mu_H Groups to plot muB mu_B Variance _ Variance Data to plot against left Y axis TPP Expl Additive Data to plot against right Y axis V Gring Dnrinanre 7 Daminance fd Options 1 Options 2 Options 3 I Show Loci M Show Legend Reset F Show Intervals I Black amp White Show Horizontal Dashed Line at Left Y axis Value Be I Show Horizontal Dotted Line at Right Y axis Value
32. ata the contents and results are shown as plain text or as a table on the Results tabsheet Results shown as a table can also be viewed as a chart see below Of any table the view can temporarily be changed columns can be moved by dragging the header with the mouse column widths can be resized by dragging the splitter between column headers rows can be sorted by clicking on the column header to use as sorting key click twice for sorting in the opposite direction You can revert to the original row order by sorting on the first column always labelled Nr Some functions within the program will automatically revert corresponding tables to the original row order The changes in the view are not stored so closing and reopening the project results in the original views of the tables Using MapQTL 11 The Results Charts tabsheet contains a set of two subordinate tabsheets one for the control of the charts and one for the actual charts There are many features of the charts that can be handled using this subordinate Control tabsheet On this tabsheet there are splitters that can be dragged with the mouse to divide the space between the checklists for Groups and Left and Right Y axis data and also between the upper part with its checklists and the lower part with checkboxes and fields for various chart options Most features are self explaining just two need some description when plotting cofactors these will always be plotted as symbols on the X axis
33. c 3 it contains data on three loci nind 7 and seven plants RFLP21 lt efxeg gt 01 marker RFLP21 segregates with three alleles genotypes of the seven plants classify into h and kk the seven genotypes in identical order as for RFLP21 the linkage phase at this seg type defines it only for the second parent the autoradiogram was unclear for plantnr 5 ef ee eg fg fg ef eg RAPD17 lt hkxhk gt h kk 00 h h kk h kk kk h RFLP34 lt nnxnp gt 1 nn np np np nn np Map file The map file contains the map positions of all loci The nonparametric mapping of MapQTL examines the loci one by one the map positions are only used to sort the loci For interval and MQM mapping the map positions are used to calculate recombination frequencies necessary for the calculation of the likelihood The map file is strictly line structured and there is no header Linkage groups must be started with the instruction group or chrom on a separate line On the subsequent lines the loci with their map positions must be given in ascending order one locus with its position Data files 47 Table 8 Classification type codes Ratio is the expected segregation ratio Code Ratio Classification into genotype classes a b 1 1 a and b a h 1 1 a and h a c 1 3 a and c h and b will be included in class c h b 1 1 h and b b d 1 3 b and d a and h will be included in class d a h b 1 2 1 a h and b ac ad bc bd 1
34. cofactor h and b while there Mapping theory 37 are three possible QTL genotypes Thus for this individual the mixture will consist of six 2x3 components With and 6 defined as the additive and dominance genetic effects respectively associated with the cofactor the means of the three distributions that correspond to the cofactor possibility h will be Hyg H while those corresponding to possibility b will be Hig H a With the probabilities for cofactor possibilities h and b based on the surrounding markers defined as 7 and 7 respectively the component probabilities that correspond to the cofactor possibility h will be due to independence of recombination events within and between linkage groups Ti T T and those corresponding to possibility b will be Ty 1 1 This simple relation doesn t hold if the cofactor is on the same chromosome as the QTL and the markers in between QTL and cofactor do not provide complete information In such cases joint probabilities of cofactor genotype and QTL genotype must be calculated as multipoint genotype probabilities In fact independence is only employed in these calculations between chromosomes in all other cases e g with more cofactors with possibly unknown genotypes on a single chromosome multipoint genotype probabilities are determined If the cofactor genotype in the example would be unknown u then the mixture would consist of nine compo
35. d b s Table 4 Genotype codes for population types DH1 and HAP1 Code Description a homozygote or haploid as the one parent b homozygote or haploid as the other parent genotype unknown genotype unknown u genotype unknown Table 5 Genotype codes for population types DH and HAP Code Description a the one genotype b the other genotype genotype unknown genotype unknown u genotype unknown 44 Data files For population type CP the type of segregation may vary across the loci Up to four different alleles may be segregating Therefore a code indicating the segregation type must be given in between the locus name and the genotypes The segregation type codes are shown in Table 6 The two characters left of the x in these codes represent the alleles of the first parent the two on the right represent those of the second parent each distinct allele is represented with a different character The genotypes for a CP population must be coded with two characters representing the two alleles per individual The coding depends on the segregation type and is shown in Table 7 MapQTL is indifferent to the order of the alleles so ac is equivalent to ca In all cases the the and the u are treated as equivalent so h and hu are both equivalent to h Although not required it is recommended as a good measure against errors to separate the genotype codes of individuals with a space The two character codes thems
36. d node that should be blue as well Right clicking any trait node under the Common traits node will apply this special selection to this trait under all populations Right clicking a population node specially selects all numerical traits of that population Right clicking a map node specially selects all linkage groups of that map This special selection is a toggle i e when right clicking again the nodes are deselected Only when 1 one or more traits and 2 one or more linkage groups are specially selected 3 the corresponding population s has the genotypes loaded and 4 an analysis is selected on the tool bar only then the Calculate function is enabled and it becomes possible to activate the Calculate button or Calculate function of the Calculate menu Once calculations have been performed a corresponding session node is created in 10 Using MapQTL the session treeview with subordinate nodes for each trait and linkage group all shown hierarchically It is possible to remove complete maps i e all linkage groups complete populations i e all traits and genotypes complete sets of traits of populations the set of genotypes of populations and complete sessions from the project This can be done by selecting blue the relevant node and pressing shift Del or applying the Delete Node function from the File menu There is a splitter between the navigation panel and the contents and results panel allowing you to reassign the space ava
37. e the genotype codes of individuals from one population are all the same size two characters for cross pollinators CP and one for other population types and may be given without spacing though this will result in poor readability The names of loci linkage groups traits and populations and also the missing value indicator may be up to twenty characters long Names cannot include spaces The full path names of files may be up to 255 characters long Lines may be up to 1000 characters wide this only applies to line structured data Locus genotype file The locus genotype file oc file contains the information of the loci for a single segregating population It has a sequential structure The header of the file contains four instructions on the contents of the data body The data body contains the actual genotype Data files 41 information for each locus and for all individuals The four instructions define the name of the population which is for administrative use only the type of the population the number of loci and the number of individuals These instructions can be given in any order within the header The syntax of the four instructions is name NAME popt POPT nloc NLOC nind NIND where NLOC and NIND are the numbers of loci and individuals respectively NAME is the name of the population which cannot contain spaces and POPT is the code for the population type which must be one of the codes given in Table
38. e extensions List of examples Example Example 1 A locus genotype file for an F2 population 2 A locus genotype file for a CP type population 51 22 22 25 27 13 42 43 43 43 45 45 47 47 50 46 46 52 Lists and references Example 3 A map file 48 Example 4 A quantitative data file 49 Example 5 A cofactors file 50 References Barnard G A 1949 Statistical inference J R Statist Soc Ser B 11 115 139 Churchill G A amp R W Doerge 1994 Empirical threshold values for quatitative trait mapping Genetics 138 963 971 Dempster A P N M Laird amp D B Rubin 1977 Maximum likelihood from incomplete data via the EM algorithm J R Statist Soc Ser B 39 1 38 Doerge R W amp A Rebai 1996 Significance thresholds for QTL interval mapping tests Heredity 77 459 464 Feingold E P O Brown amp D Siegmund 1993 Gaussian models for genetic linkage analysis using complete high resolution maps of identity by descent Am J Hum Genet 53 234 251 Fijneman R J A S S De Vries R C Jansen amp P Demant 1996 Complex interactions of new quantitative trait loci Slucl Sluc2 Sluc3 and Sluc4 that influence the susceptibility to lung cancer in the mouse Nature Genetics 14 465 467 Jansen R C 1993 Interval mapping of multiple quantitative trait loci Genetics 135 205 211 Jansen R C 1994 Controlling the type I and type II errors in mapping quantitative trait loci Genetics 1
39. e locus are modelled and tested The process stops when the change in likelihood is significant according to the P value for the test can be modified with the Analysis Options of the Options menu or when there is no more cofactor locus remaining in the set The test statistic used for the comparison of subset models with the full model is the deviance see the Mapping theory chapter p 32 The deviance is assumed to follow a chi square distribution The degrees of freedom is the number of regressors per cofactor which is one for population types BC1 HAP1 HAP DH1 DH and RIx two for F2 and three for CP However when the number of parameters in the model one for the overall mean plus one BCI etc two F2 or three CP for each cofactor locus is large the estimate of the residual variance will be biased and as a result the assumption of the deviance following a chi square distribution will be violated Therefore the number of parameters in the model should not be too large preferably less then twice the square root of the number of individuals in the population Jansen amp Stam 1994 A warning is issued by the software when this is not the case Often though this warning can be ignored because in such situations the difference in the likelihood of the full model with that of the subset model can be so small that even with a violation of the chi square assumption the test will most probably be not significant When the number of
40. e to missing genetic information missing marker scores dominant scores low information segregation types and usually limited population size there are often insufficient degrees of freedom and memory RAM to accommodate for all or many loci in the starting set of the automatic selection procedure this can be circumvented by doing linkage groups one by one and fixing the detected results Let s try this approach with the current DemoF2 data set 1 Verify that the trait gtrait and the ten linkage groups are still specially selected and set the Analysis selector on Automatic Cofactor Selection 2 Click on the Cofactors Tool button M 3 Select the action Clear all loci and click on the Do it button 4 Check Group 1 in the list Act on checked groups select the action Check all loci on indicated groups click on the Do it button and close the Cofactors Tool Now you have checked all loci on group 1 and no other loci verify this Remark In the DemoF2 population the information is sufficient to do automatic cofactor selection with all loci of a group as a starting set Whenever your own data set have insufficient information you will receive error messages about insufficient memory or the calculations proceed extremely slow In such cases you should reduce the number of selected loci for the starting set by picking loci say every 20 cM If subsequently a significant locus is detected you may try to improve the result by using a starting se
41. ed to calculate the component probabilities The way this is done varies slightly over the population types For all population types except CP markers with unknown genotype u are completely ignored markers with unknown genotype are sometimes referred to as missing markers If for a certain individual a QTL flanking marker has an unknown genotype then the first not unknown marker beyond this missing marker will be treated as the flanking marker of course taking its distance to the current QTL position into account For the F2 and RIx population types marker genotypes can also be dominant which means that the genotype is just partially unknown or partially known MapQTL tries to resolve these so called incomplete genotypes as much as possible by taking into Mapping theory 33 account the markers beyond it on the map However to prevent endless calculations in case neighbouring markers also have incomplete genotypes a limit is set upon the number of markers used beyond the flanking markers The corresponding parameter is called the maximum number of neighbouring markers used when set to 0 only the genotypes of the markers flanking the current QTL position are used Missing markers are ignored in the counting of these neighbouring markers used In the hopefully rare case that all of the linked marker genotypes are unknown and cannot be used to calculate the component probabilities then their average values over the population are used the
42. elves may not be separated with whitespace The CP coding scheme is enhanced from the previous version 3 0 and from JoinMap 2 0 but this older format is interpreted correctly by the present version Analogous to the population types DH and HAP MapQTL requires known linkage phases of the alleles of the loci for interval and MQM mapping in a CP type population The genotype coding scheme is based on the alleles on the same position within the segregation type codes to be in coupling in the parent i e the a e h and 1 alleles from the first parent come from the same one grandparent the b k and m alleles from the first parent from the other grandparent However to allow for linkage phase differences a linkage phase indicator is used similar to DH and HAP but here we need a two digit phase type of which the first relates to the one parent and the second to the other The phase type must be one of the next two letter codes between curly brackets for the seg type lt 1mx11 gt o or 1 for the seg type lt nnxnp gt o or 1 for the other seg types oo 01 10 or 11 Locus pairs with the same digit in the first position of their phase types are assumed to be in coupling in the first parent and in repulsion in the first parent otherwise for the second position the relation is likewise about the second parent For instance if a locus L is of type lt hkxhk gt 00 and another locus M is lt abxcd gt 01 this means t
43. ent to the two sided Wilcoxon rank sum test The test is performed on each locus separately no use is being made of the linkage map other than for sorting the loci An application is described by Van Ooijen et al 1993 The Kruskal Wallis test can be regarded as the nonparametric equivalent of the one way analysis of variance The test ranks all individuals according to the quantitative trait while it classifies them according to their marker genotype A segregating QTL with big effect linked closely to the tested marker will result in large differences in average rank of the marker genotype classes A test statistic based on the ranks in the genotype classes is calculated For individuals in ties i e several individuals have equal values of the quantitative trait the average rank midrank is used while for the test the statistic Using MapQTL 13 Table 1 Analysis abbreviations Code Description KW nonparametric mapping or Kruskal Wallis analysis IM interval mapping MQM MQM mapping rMQM restricted MQM mapping ACS automatic cofactor selection PT permutation test adjusted for ties is used indicated by K Lehmann 1975 eqn 5 11 For the genotype classification the usual genotype classes are used when you wish to classify in another way i e when there is dominance or for recombinant inbreds it is possible to indicate another classification by adding a code to the locus in the loc file see Table 9 p 47 Under
44. es This method was called all markers mapping by Maliepaard amp Van Ooijen 1994 In contrast to F2 and RIx the completely unknown markers uu are not completely ignored in the implementation for CP they are included in the counting of the number of neighbouring markers used although of course they do not contribute to the resolution of partially unknown genotypes Genotypic information coefficient The power of interval mapping lies in the possibility to make use of genetic information of markers in the region of the QTL When markers lie further away when many marker scores are missing or when markers are scored dominantly the power reduces because the genotypic information is smaller In order to quantify the genotypic information across the genome the genotypic information coefficient GIC is defined analogous to 34 Mapping theory the marker information content defined by Knott et al 1997 and Reyes Vald s amp Williams 2002 The GIC can have a value in the range from 0 to 1 0 meaning there is no marker information at all 1 meaning that there is complete or maximum marker information the marker lies on top of the putative QTL while all individuals in the population have a codominant score The derivation of the GIC is as follows A QTL generates an amount of genotypic variance V This variance can be partitioned into the variance to be explained by genotypic marker information V and the variance remaining due to uncer
45. g a so called QTL likelihood map is calculated i e for each position on the genome say every centiMorgan the likelihood for the presence of a segregating QTL is determined the likelihood under the alternative hypothesis H1 At the same time the genetic effects of the QTL and the residual variance are calculated This likelihood under H1 is compared to the likelihood for the situation when a locus with zero genetic effect would segregate i e there is no segregating QTL the likelihood under the null hypothesis HO This comparison is done with a likelihood ratio statistic called the LOD or LOD score which is the 10 base logarithm of the quotient of the two respective likelihoods When the LOD score exceeds the predefined significance threshold somewhere on a linkage group a segregating QTL is detected the position with the largest LOD on the linkage group is the estimated position of the QTL on the map To obtain a roughly 95 confidence interval around this point estimate a so called two LOD support interval must be constructed by taking the two positions left and right of the point estimate that have a LOD value of two less than the maximum In the Mapping theory chapter p 31 the choice of the significance threshold and other more technical details are discussed For the F2 and RIx population types MapQTL allows the fit of dominance of the QTL but this may also be restricted so that the heterozygous QTL genotype is strictly i
46. g proportions mqm mqm mapping name instruction navigation panel ncof instruction new project nind instruction nloc instruction non numerical data nonparametric nr ntrt instruction number of permutations open project optional classification types output data file page setup permutation test phase type phase type codes plain text popt instruction population info population name population type population type codes populations preset default print print preview print setup program directory program settings directory project project backup project directory project file project info project notes pt p value 35 13 16 16 35 41 2 3 7 9 50 3 8 41 48 41 3 9 49 12 10 48 19 8 47 50 4 11 19 38 41 42 44 42 44 qua file quantitative data file ram remove reset reset to default restricted mqm mapping results results charts right clicking rmqm save as default segregation ratio segregation type segregation type codes selective genotyping 15 selector sequential session session info sessions settings directory setup exe shift del significance level significance threshold skewness sort table special keys specially selected splitter table tabsheet charts control genotypes info map info maps population info populations project info project notes 39 48 39 49 50 2 10 8 11 8 17 4 10 4 11 9 13 17 8 35 47 41 44 45
47. genotyped part of the population in case of selective genotyping It must be noted that in the case of analysing unmapped loci no use can be made of linked markers so that for incompletely known genotypes here no further information can be obtained In the calculations of the component probabilities for an RIx population use is made of exact composite three locus genotype probabilities i e for the QTL in between two flanking markers This is necessary because recombination events are not completely statistically independent except in the generation of RI2 These computations involve matrix multiplications for every generation after the RI2 and are somewhat heavy By ignoring the dependence the composite three locus genotype probabilities can be approximated by multiplying a single locus probability and two transition probabilities for each transition to a neighbouring locus as is done for all other population types where there is independence and it is therefore correct This approach generally leads to very similar outcomes but the computations are faster It can be chosen with the Use genotype probability approximation in RIx parameter In a CP population markers of various segregation type can be used Here like dominant genotypes in an F2 several genotypes of a number of segregation types provide partially incomplete information Similar to the approach for F2 and RIx neighbouring markers are used to calculate the component probabiliti
48. hat in the first parent the h allele of L and the a allele of M are in coupling and thus also their k and b alleles and that in the second parent the h allele of L is in repulsion with the c allele of M and thus in coupling with the d allele of M The phase type must be given in between the locus name and the genotypes Here too the linkage phases can be obtained automatically with JoinMap 3 0 For the nonparametric mapping the program classifies the genotypes according to the Data files 45 Table 6 Segregation type codes for population type CP Code Description lt abxcd gt locus heterozygous in both parents four alleles lt efxeg gt locus heterozygous in both parents three alleles lt hkxhk gt locus heterozygous in both parents two alleles lt lmxll gt locus heterozygous in one parent lt nnxnp gt locus heterozygous in other parent Table 7 Genotype codes for a CP population depending on the locus segregation type Seg type Possible genotypes lt abxcd gt ac ad bc bd no dominance allowed lt efxeg gt ee ef eg fg no dominance allowed lt hkxhk gt hh hk kk h k lt lmxll gt 1l lm lt nnxnp gt nn np Remarks 1 each character a to p represents a distinct allele means unknown allele 2 h and k are dominant genotypes h means hh or hk and k means kk or hk 3 and u are treated equivalent to usual genotype classes However you may wish to cla
49. he MQM mapping and inspect the results You will see that on none of the groups 7 9 and 10 significant LOD scores are computed so the loci can indeed be regarded as false positives 11 Another fact that you should observe is the significant region some distance away from cofactor m49 on group 3 Similarly to what we did to tackle the comparable problem of Figure 5 select all markers on this group and do automatic cofactor selection The result will be that m49 will be swapped for m53 12 Now we should return to step 7 and redo groups 1 to 5 to try to see if the results improve or change Actually the previous step 11 may well have been a part of this In fact groups 6 to 10 should also be redone If you do this you will notice that several false positives will emerge which may be unmasked with MQM mapping as in step 10 The final result should have the following set of cofactor markers m9 m13 m33 instead of m32 m53 m75 m96 the MQM results with this set you have already obtained above This result is identical to what we have found earlier above but now with a more systematic approach If you open the DemoF2 loc file in the DemoData directory with Notepad you will see at what positions QTLs were located in the simulation in all cases the detected cofactor markers are one of the markers flanking the QTL which is the best we could have found As a final remark the DemoF2 data set is a nice simulated data set where all markers are
50. he analysis has increased by taking the closest markers as cofactors with the MQM mapping analysis Tutorial 27 Subsequently you can try to improve the results Examine all five linkage groups and find the marker closest to the LOD peak and modify the set of cofactors correspondingly The set of cofactor markers should now be m9 m33 m53 m75 and m96 Redo the MQM mapping analysis When you inspect the results you will notice that the chosen cofactor markers are now still closest to the current LOD peaks that is what you would want to have as a final result however There is however one important aspect that you must see on linkage group 1 there are LOD values above the significance threshold some distance away from the cofactor marker m9 Figure 5 This is an indication that the cofactor m9 is a distance away from the real QTL position or there may be more QTLs on this group In order to study how many and where these QTLs are you can use the automatic cofactor selection procedure in combination with the cofactor markers already determined on groups 2 to 5 e Select the Map Info tabsheet e Verify that only m9 m33 m53 m75 and m96 are checked in the Cofactor column e Click on the Cofactors Tool button M e Check Group 1 in the list Act on checked groups select the action Check all loci on indicated groups click on the Do it button and close the Cofactors Tool e Verify that now all loci of group 1 are checked as cofactor
51. iated with a position is in relation to the residual variance the more significant is the test However when several QTL are segregating some of the residual variance will be determined by the other segregating QTLs If we could take these QTLs into account while testing for a QTL at a certain position then the residual variance would be reduced and as a consequence the test would become more powerful This is achieved by taking the markers that we think are associated with a QTL as cofactors in the so called MQM Mapping analysis also called composite interval mapping e Select the Map Info tabsheet e Check the boxes in the Cofactor column for markers m9 m54 m75 and m97 e Set the Analysis selector on MQM Mapping e Verify that the trait gtrait and the ten linkage groups are still specially selected if not restore this and subsequently click on the Calculate button Now inspect the results of MQM mapping using the charts the LOD significance threshold can be taken as the same value 3 8 use this value for the horizontal dashed line in the charts check Cofactors in the Left Y axis list on the Control tabsheet Find the linkage groups with a value above this threshold These should be groups 1 2 3 4 and 5 group 2 is now also included with a very significant LOD score as well Notice that all LODs are much larger values than with interval mapping because there apparently are QTLs with a larger explained variance the power of t
52. ilable to the panels Contents and results panel The contents and results panel contains a set of tabsheets that will display the data of the population map and session nodes In addition to the tabsheets for loaded data i e Population Info Traits Info Genotypes Info and Map Info tabsheets and the results of analyses the contents and results panel also has a Project Info tabsheet with an overview of all actions done within the project and a Project Notes tabsheet on which you can make your own notes about the project and which will be stored with the project The Map Info tabsheet does not only show the linkage groups and positions of loci it also has a column with checkboxes for indicating whether loci should be used as cofactors in the analysis to be performed The Cofactors Tool can be very helpful with this it is available from the Edit menu and from a tool bar button It is a floating tool allowing you to change the tabsheets while the tool remains available The Map Info tabsheet will indicate for each locus its presence within each loaded population by the character X in the column for that population Any locus present within a loaded population but not present in the loaded map file will be added to the map on the Map Info tabsheet as an unmapped locus without a map position A node named Unmapped will be created in the Maps navigation treeview that will correspond to all unmapped loci Depending on the type of d
53. ill be prompted for the name of the population that it should be stored under just click OK for the default DemoF72 Notice that on the Populations tabsheet a DemoF 2 population node is created with a child node Genotypes To load the quantitative data click on the Load Data tool bar button 1 and select the file DemoF2 qua in the DemoData directory Next you will be prompted for the name of the population that it should be stored under just click OK for the default DemoF 2 the navigation panel will resemble Figure 2 notice the population DemoF2 with two numerical traits nr and qtrait and with Genotypes also notice the common traits node which is only useful when there are more populations in a project 22 BA MapQTL 5 tutorial File Edit Calculate Options Help Dio Analysis Ctrl M MapQTL 5 tutorial File Edit Calculate Options Help Di Analysis Ctrl yl Populations Maps Sessions amp C Common traits M nr qtrait P DemoF2 nr T qtrait Genotypes Genotypes Info Project Info Project tutorial created Fri Loaded locus genc date time added under por original pop r population type nr of loci nr of individ i DemoF2 Ei Group 1 EH Group 2 EH Group 3 it Group 4 Ei Group 5 3 Group 6 i Group 7 Ei Group 8 i Group 9 Genotypes Info Project Info Project tutorial created Fri
54. inting a preview of the print out can be obtained through the Print Preview option of the File menu or the tool bar button El From within the Print Preview and from the File menu the Page Setup and the Print Setup can be modified This user manual is accessible as an Adobe pdf document though the Help menu Final remarks With MapQTL 5 you have quite a powerful tool to analyse the data that you have obtained from your experiments It is important to realize that the quality of your data is crucial to the possibility to discover real QTLs many missing marker observations reduce the power of the analyses erroneous marker scores and an incorrect linkage map usually the product of missing and erroneous marker observations both may generate inconsistent results It does not lie within the power of a software tool to compensate for Introduction 5 missed out quality of its input data But even with good quality data the detection of QTLs is only possible if QTLs do segregate in the population under study and if their genetic effects are sufficiently large in relation to the residual variance and the size of the experiment And above all you have to keep in mind that MapQTL is a statistical tool and that the results point you to statistical conclusions with a definite amount of uncertainty How to cite MapQTL 5 Van Ooijen J W 2004 MapQTL 5 Software for the mapping of quantitative trait loci in experimental populations Kyazma B
55. linkage group node in the Maps tabsheet will select the first locus of that group in the Map Info tabsheet Similarly the selection of a linkage group node in the Sessions tabsheet will select the first locus of that group in the Results tabsheet Finally when you are looking for certain text on a plain text or table tabsheet in the contents and results panel you can make use of the Find tool available from the Edit menu It is a floating tool allowing you to change the tabsheets while the tool remains available Starting an analysis On the tool bar there is a selector for the analysis In order to start an analysis you must 12 Using MapQTL first choose the analysis itself from this selector Secondly the traits that must be analysed must be specially selected by right clicking see the Navigation panel section p 9 Even when traits are to be analysed separately multiple traits can be selected to be analysed in one go Traits that have non numerical data cannot be selected for analysis they are shown in green When a population has no subordinate Genotypes node i e its locus genotypes are not loaded into the project the traits for that population cannot be analysed Thirdly the linkage groups that must be analysed must be specially selected on the Maps tabsheet The group of unmapped loci represented by the node Unmapped can only be used for separate nonparametric and interval mapping analyses Only if all above conditions are
56. loci will be estimated with a large negative covariance Therefore selective genotyping is implemented only for interval mapping in MapQTL not for MQM mapping The analysis of selective genotyping with MapQTL is easily arranged by putting the data of the genotyped individuals in the top of the quantitative data file and those of the not genotyped ones below The top individuals should correspond to the individuals in the loc file In the calculation of the likelihood the not genotyped individuals obtain probabilities for the possible QTL genotypes based upon the usual expected segregation ratio see Table 8 p 47 corrected for the segregation ratio acquired with the genotyped individuals However when this correction leads to illegal values negative or larger than 1 the uncorrected expected segregation ratio will be used When this happens the program reports this as modifying mixing proportions of ungenotyped individuals MQM mapping The approximate multiple QTL model used in MQM mapping is an extension of the single QTL model presented in the Interval mapping section In this model selected cofactors take over the role of nearby QTLs this forms the approximation aspect of the model In the model a single segregating QTL is fitted in a background of cofactors 36 Mapping theory Genetic effects of the separate QTLs i e the single fitted QTL plus the others as represented by cofactors are modelled as additive fixed effects there is
57. lowing meaning in alphabetical order lt class gt the genotype class for which the details are given Df the degrees of freedom 14 Using MapQTL Group the linkage group of the locus K the Kruskal Wallis test statistic K Locus the name of the locus at the current position Mean lt class gt the arithmetic mean of the class Meanrank lt class gt the mean rank of the class Nr sequential number of the row Nr lt class gt the number of individuals in the class Nr inf the number of informative individuals i e the individuals with a genotype within the current classification and with a known quantitative trait value Position the current position on the map Signif the significance level in asterisks details are given on Session Info tabsheet Sporadically it may occur that there is just a single tie in the quantitative data of course you can conclude here that there is no genetic effect in this case it is impossible to calculate the statistic and the relevant cells of the table will stay empty When a genotype class is empty its mean and mean rank cannot be calculated relevant cells will stay blank When one or more genotypes are detected outside the current classification a warning is printed in the Signif column in the form of Interval mapping The interval mapping method was developed by Lander amp Botstein 1989 The method is more extensively described in a paper by Van Ooijen 1992 In interval mappin
58. lts of analyses are displayed for the population map and analysis session selected i e blue in the navigation panel Controlling the program Because MapQTL is an MS Windows program you can expect the many features to be controlled in the normal MS Windows fashion with the mouse and the keyboard Below is a summary of some normal and special keys and key combinations alttkey key being any underlined character shown in the program as usual go to the associated part of the window or perform the associated action ctrl A select all in selected tabsheet of the contents and results panel ctrl C copy the selected tabsheet of the contents and results panel to clipboard or its selection ctrl F open the Find dialog ctrl H show the Charts page within the Results Charts page ctrl O show the Control page within the Results Charts page ctrl P print the selected tabsheet of the contents and results panel or its selection ctrl Y go to the Analysis selector shifttDel delete the selected node in the visible tree view Tab rotate focus through all visual elements Esc close the Cofactors tool close the Print Preview tool cancel the options 8 Using MapQTL dialogs cancel the calculations Break cancel the calculations Fl view the manual as Adobe pdf document F4 load data into the project F9 execute the selected analysis on the selected traits and selected linkage groups alt F4 exit the program The Environment Options of the Optio
59. moves to the beginning of this group in the map table This navigation feature works for each group node of the treeview Select one by one the remaining three tabsheets and observe that they are empty these are for showing the results of analyses which you haven t done yet After having looked for a little while at the user interface of MapQTL you may wish to modify the fonts your copy of MapQTL is using This is possible by selecting the Environments Options of the Options menu Do this and pick the fonts and font sizes of your preference that will be used for the various elements of the user interface By clicking the OK button the current choice is saved with this project If you wish to use this choice for any future MapQTL project press the Save as default button You are now ready to start doing analyses Let s do interval mapping for the trait gtrait on all linkage groups e Click in the Analysis selector and pick Interval Mapping e Select the Maps tabsheet on the navigation panel and right click on the map node DemoF 72 The result will be that all linkage group nodes will be highlighted with a red background This is a toggle right clicking again removes the red highlighting This special selection may also be done by pressing the keyboard space bar try this The specially selected linkage groups will be used in the analysis e Select the Populations tabsheet on the navigation panel and right click on the trait node qtrait You ma
60. n all many internal data files After the data files i e the locus genotype file the map file and the quantitative data file are loaded into the project the original files are not needed by MapQTL When backing up a MapQTL project always take the project file as well as the project directory with all its files Once a project is opened you can load data into the project This must be done with the Load Data function of the File menu or with a tool bar button W Data must be loaded from three separate files 1 the set of locus genotypes of a population 2 the set of quantitative trait data of a population and 3 the map data the location and name of the files can be chosen with a standard save file dialog window The formats of data files used by MapQTL are described thoroughly in the Data files chapter p 39 Some example data files are present in the DemoData subdirectory of the program directory Using MapQTL 9 typically C Program Files MapQTL5 More than just one population and more than a single map can be loaded into a project Navigation panel The navigation panel has three tabsheets and at the bottom three small status bars The Populations tabsheet will show the populations with their traits and genotypes loaded into the project The Maps tabsheet will display the maps with their linkage groups The Sessions tabsheet will show the calculation sessions that have been performed These tabsheets show their contents
61. nents And when the model contains more cofactors and the individual also has a dominant genotype on a second cofactor then the number of components is multiplied by two Thus the number of components in the mixture is multiplied by the number of possible genotypes for each cofactor The number of parameters that have to be estimated doesn t change with incomplete cofactor genotypes The number of components and thus the number of probabilities that have to be calculated however grows in a multiplicative fashion with cofactor incompleteness It is important to realise this because choosing several cofactors that have high amounts of incompleteness e g RAPDs in an F2 can dramatically reduce the speed of the program A not very obvious phenomenon with MQM mapping is that the estimated genetic effect and the percentage of explained variance are rather biased when the fitted QTL lies not 38 Mapping theory very far from a linked cofactor marker that is included in the model to absorb the effects of another QTL This is due to the fact that under the null hypothesis the linked cofactor marker will absorb part of the same genetic effect that is being fitted to a QTL under the alternative hypothesis LOD significance threshold Several papers address the subject of what significance threshold to use for the LOD score in a QTL mapping experiment e g Lander amp Botstein 1989 Van Ooijen 1992 Feingold et al 1993 Rebai et al 1994
62. no QTL by QTL interaction epistasis Just as in the single QTL mixture model the mixture density for an individual is the sum of the products of the Q component densities with their probabilities The important distinction is that each individual now has a separate set of component densities These component densities are normal densities with common variance O but the component means are based on both QTL genotype and cofactor genotype the means 4 of the Q distributions of individual n consist of the means of the distributions associated with the QTL genotypes u that are adjusted with the genetic effects associated with the genotypes of its cofactors Depending on the population type these genetic effects are the additive and or dominance effects or for a CP population the deviations from the ac genotype in phase type 00 all effects as defined in the Interval mapping output section of the Using MapQTL chapter p 15 For instance when we have an F2 individual modelled with two cofactor markers on which it has genotypes b and h respectively the three means 4 are modelled as LH H a 3 in which is the additive genetic effect associated with the first cofactor and is the dominance effect of the second As in the single QTL model the parameters 4 o and the genetic effects associated with each cofactor are estimated with the EM algorithm The estimates of the cofactor genetic effects the regressors are printed
63. ns menu allow the setting of the fonts for the various elements of the user interface and the various chart options The Analysis Options allow the setting of the various calculation parameters Clicking on the Preset default button on these options dialogs changes all values to the internal program values of all parameters Clicking on the Save as default button stores all current values to the program settings directory My Documents MapQTL5 which will be used as starting values for each new project and can be loaded into an opened project by clicking on the Reset to default button Clicking on the OK button applies all fonts settings immediately to the current project all chart and analysis options will be applied to all new calculation sessions The new chart options can be applied to existing charts by clicking on the Reset button on the charts Control page after selecting the particular chart The MapQTL project In MapQTL your work is organised into a project You create a new project D or open an existing project a using the File menu or tool bar buttons the location and name of the project can be chosen with a standard save file dialog window The whole of a MapQTL project consists physically of a the project file with extension mqp and b the project data directory with the same name as the project file but with the extension mqd The project data directory resides in the same directory as the project file it will contai
64. ntermediate In advanced RIx generations the fit of dominance can be impossible due to Using MapQTL 15 a complete lack of heterozygous marker genotypes For an F2 the default is to fit dominance for an RIx population it is no dominance The analysis of a selectively genotyped population see the Mapping theory chapter p 35 is easily arranged by putting the data of the genotyped individuals in the top of the quantitative data file and those of the not genotyped ones below The top individuals should correspond to the individuals in the loc file The calculation of the maximum likelihood is implemented in MapQTL as an iterative EM procedure The iterations stop when the relative change in the logarithm of the likelihood is less than the so called functional tolerance value or when the maximum number of iterations is reached These and some other options may be set in MapQTL It is possible to print the test statistic as a deviance instead of a LOD score see the Mapping theory chapter p 32 for a definition of deviance Further MapQTL uses the so called mapping step size parameter to go from one position on the map to the next in between loci for the positions for which the LOD or deviance must be calculated choose a large value if you only want computations on locus positions and not in between For population type RIx it is possible to speed up computations using a QTL genotype probability approximation instead of using correct three p
65. ogarithm of the likelihood the log likelihood function is calculated The iterations stop once the relative change in the log likelihood function has become smaller than the so called functional tolerance value which means that the algorithm has converged To prevent endless iterations in the case of non convergence the iterations stop when the number of iterations has reached the maximum number of iterations The functional tolerance value and the maximum number of iterations are parameters that can be set although their default values usually suffice The L likelihood is compared to the likelihood under the null hypothesis L which is similar to L except that there is just a single component in the mixture Q 1 i e no QTL is assumed to be segregating The comparison is done using the so called LOD score LOD log of odds Barnard 1949 as a test statistic LOD log L L Some prefer the equivalent test statistic called the deviance which uses the natural instead of the 10 base logarithm the deviance is also called the likelihood ratio test Statistic D 2 log L L 4 605 LOD LOD 0 21715 The test statistic can be compared to a significance threshold to decide upon presence or absence of a QTL see the LOD significance threshold section p 38 If the markers directly flanking the QTL do not provide complete information genetic information from markers surrounding the assumed QTL map position is us
66. oint genotype probabilities details are described in the Mapping theory chapter p 33 Finally for population types F2 RIx and CP the parameter called maximum number of neighbouring markers used can be modified This parameter is important in the calculation of the QTL genotype probabilities based on the marker genotypes when the markers have a dominant genotype or a not fully informative segregation type details are described in the Mapping theory chapter p 31 These are all parameters that can be set with the Analysis Options of the Options menu Interval mapping output A summary of the parameters input and the data read from the files are given on the Session Info tabsheet The presented population variance is the usual ML estimate to get the unbiased variance this is multiplied by n n 1 with n being the number of individuals The population skewness and kurtosis are the coefficients of skewness and kurtosis denoted as g1 and g2 respectively by Snedecor amp Cochran 1980 sec 5 13 5 14 The Results tabsheet lists the results for all fitted QTL positions under the following headers in alphabetical order see the Data files chapter p 39 for the genotype codes Additive the estimated additive effect BC1 mu A mu H or mu_H mu B F2 RIx HAP1 or DH1 mu_A mu_B 2 HAP or DH mu_A 0 mu B 0 2 Deviance the deviance Dominance the estimated dominance effect F2 or RIx 16 Using MapQTL mu H
67. onding map positions at the end of the line and cofactors are indicated with an exclamation point On the Session Info tabsheet the information is given for the 18 Using MapQTL successive HO models that were fitted a the locus name at the position for which the HO applies b the In likelihood In log c the number of iterations d the residual variance after fitting the overall mean plus the cofactors and e the variance explained with this HO model Automatic selection of cofactors MapQTL offers the possibility for automatic selection of cofactors The analysis is based on backward elimination Starting with the set of cofactor loci that are selected by the user for instance four on each linkage group a standard MQM model is fitted that includes these cofactors and excludes the QTL the starting set of cofactors for the selected traits must be chosen from the Map Info tabsheet As such it is equal to a null hypothesis model in MQM mapping Subsequently by leaving out one cofactor locus at a time subsets of loci are created for which the corresponding models are fitted The likelihoods of each of these subset models there are as many as there are loci in the starting set are compared to the likelihood of the full model with all cofactor loci The subset of which the model caused the smallest change in likelihood is chosen as the starting set for a subsequent round of elimination in which new subsets containing all but on
68. ores its various program settings in the directory MapQTL5 which is created in the My Documents directory when running the program Apart from the length of names maximum of twenty characters for population locus trait and linkage group names there are no limits built into the software memory for storing data is allocated dynamically only for the amount needed Thus project size is limited only by the amount of RAM memory in the PC for which a size of 256 MB is recommended for reasonably sized projects Overview Start the program by using the Windows Start menu When the program runs you will see a window that is divided into several main parts on the top the menu and the tool bar with buttons and a selector on the left hand side the navigation panel on the right hand side the contents and results panel and on the bottom the status bar Figure 1 The navigation panel contains three tabbed pages tabsheets which will show bookmarks for the data loaded into a mapping project the traits and the marker genotypes of populations the maps and for the analysis sessions performed These bookmarks are given as nodes in treeviews like the Folders panel in the Windows Explorer The contents and results panel also contains several tabbed pages which as the name suggests display the contents or results of the bookmark selected in the navigation panel MapQTL 5 olx File Edit Calculate Options Help neja Pras en A Ele eea
69. ously heterozygous and homozygous diploid parents linkage phases originally possibly unknown For interval and MQM mapping in a DH or HAP population MapQTL requires known linkage phases of the alleles of the loci The nonparametric mapping does not need linkage phases because it analyses the loci one by one These linkage phases can be obtained automatically with JoinMap 3 0 The genotype coding scheme is based on the loci to be in coupling in the parent i e the a s come from the same one grandparent the b s from the other grandparent However to allow for linkage phase differences a linkage phase indicator is used a phase type Such a phase type must be one of the following single letter codes between curly brackets o or 1 For a locus with a phase type 1 the grandparental origin is switched i e the a s originate from the other grandparent the b s from the one grandparent Locus pairs with the same phase code are assumed to be in coupling in the parent and in repulsion otherwise Data files 43 Table 3 Genotype codes for population types F2 BC1 and RIx Code Description a homozygote as the one parent b homozygote as the other parent h heterozygote as the F1 c not genotype a dominant b allele not possible for a BC1 d not genotype b dominant a allele not possible for a BC1 genotype unknown genotype unknown u genotype unknown Remark A BC1 must be coded either with a s and h s or with h s an
70. parameters is so large that it leaves twenty or less degrees of freedom for the estimation of the residual variance then the automatic selection algorithm is not executed Using MapQTL 19 The Results tabsheet presents the entire procedure of backward elimination The final selection of cofactors is saved in a cofactors file in the project data directory under a name given at the end of the Results tabsheet This final set of cofactors can be used in subsequent MQM analysis and can be loaded from this cofactors file using the Cofactors tool In MQM analysis the QTL likelihood map is studied in which it may be seen that sometimes QTLs are fitted somewhat distant to the closest cofactor in the set Then it can be a good idea to modify the set of cofactor loci replace the cofactor in the set with one that is closer to the maximum in the QTL likelihood map and redo the MQM analysis Possibly this needs to be done a few times In the end you would like to finish with a set of cofactor loci that are the loci closest the significant maxima in the QTL likelihood map In many cases there are insufficient degrees of freedom and or RAM memory to accommodate all or many loci in the starting set of the automatic selection procedure this is generally due to a larger number of missing observations dominance or less informative segregation types in CP populations One can think of several approaches where the procedure can still be used in an adapted way For in
71. quent MQM mapping With this MQM mapping a one dimensional search over the genome is done by testing for a single segregating QTL as in interval mapping while simultaneously fitting the selected cofactors both under HO and under H1 Thus the cofactors will reduce the residual variance If a QTL explains a large proportion of the total variance then the use of a linked marker as cofactor in subsequent MQM mapping will importantly enhance the power in the search for other segregating QTLs After the first attempt of MQM mapping it is possible that the most likely positions of some QTLs are different from those in the cofactor selection phase after all the power is enhanced In such cases one should adjust the selection of cofactors and redo the MQM mapping Sometimes even more of these rounds will be necessary to obtain the best possible final solution MapQTL offers two options in MQM mapping The first called restricted MQM mapping is to use all cofactor markers except the ones on the linkage group the QTL is fitted on The second option just called MQM mapping is to use all indicated cofactor markers in this method a cofactor is temporarily excluded from the HO and H1 models when it is one of the flanking markers of the interval on which the QTL is fitted This means that in moving through the map the set of cofactors included in the model will change and hence the HO needs to be recalculated on change of the set of cofactors because it has
72. roject directory pick the proper cofactors file click on Open and close the Cofactors Tool e Verify that the boxes for m9 m13 m33 m53 m75 and m96 are checked this can be done easily after sorting the Cofactors column click on the column header and all checked loci will be together on top or bottom e Set the Analysis selector on MQM Mapping and click on the Calculate button When you inspect the new results you will notice that nowhere outside the intervals flanking the chosen cofactor markers are LODs significant and that the cofactor markers are at the highest LOD positions On group 1 you have detected two significant QTLs about 20 cM apart This was only due to the fact that we discovered significant LODs outside the region where another QTL was detected and we decided to try automatic cofactor selection starting with all markers on group 1 These high LODs just outside a QTL region are caused by the fact that the second QTL has a large genetic effect while the preselected cofactor m9 was a fair distance away In other circumstances we might have missed out on the second QTL using the approach we have taken here For instance if we would have used marker m11 instead of m9 we would not have observed the significant LODs outside the region of m11 This would be an example of a ghost QTL one non existing QTL would be detected in the middle of the two real QTLs both having genetic effects acting in the same direction often called QTL
73. rts and another for the charts themselves they can be selected with keyboard combinations ctrl O and ctrl H respectively try this e Select the subordinate Charts tabsheet Each linkage group has a separate chart in which the LOD score is plotted against the map By default the axes are rounded upward to natural values depends on font and screen size and resolution but the data are plotted upto their largest values e Select the subordinate Control tabsheet There are many options to control what is plotted and how it is plotted In the upper part there are three checklists one labelled Groups to choose which linkage groups must be plotted two labelled Left Y axis and Right Y axis respectively to control which data should be plotted against the corresponding Y axis In the lower part there are three tabsheets full with options that allow you to set the charts to your preference Before you go on you will need to know what LOD value is significant You can use the formula and tables in the paper by Van Ooijen 1999 For this you need the following the current map DemoF2 has ten linkage groups n 10 and an average chromosome map length of 98 6 cM the population type is an F2 you used the MapQTL option to have unrestricted dominance see the Analysis Options under the Options menu therefore you will need Table 2 in the above paper Verify that for the standard genome wide significance of 5 0 05 the formula and tables of the p
74. s in coupling phase The approach we have followed up to the automatic cofactor selection can be seen as a forward selection procedure do interval mapping and fix significant regions using markers as cofactors As we have seen a drawback of the approach is the possibility to obtain ghost QTLs but it also has the possibility to miss out on linked QTLs with counteracting genetic effects i e QTLs in repulsion phase not present in the DemoF2 example Contrarily the automatic cofactor selection procedure with its backward elimination does have the potential to discover linked QTLs in coupling and in repulsion Therefore a more systematic approach employing automatic cofactor selection is to be recommended in order not to make any mistakes with linked QTLs if these happen to be present Tutorial 29 One type of a more systematic approach could be to do forward selection with interval mapping followed by automatic cofactor selection starting with the cofactor markers at the detected significant LOD peaks as a fixed set extended with all markers on a single linkage group or a subset at a certain distance and do this for each linkage group When nothing new is detected you are finished but when new linked QTLs are detected you should start over again because you have increased in power due to the newly detected QTL Another type of a more systematic approach would be to start off right from the beginning with automatic cofactor selection Du
75. satisfied will the Calculate function be enabled When all set the analysis can be started by pressing the Calculate button E by pressing the F9 function key or by using the Calculate menu When everything is OK the progress of the calculations is shown on the right hand side of the status bar a set of nodes is created on the Sessions tabsheet and when done the results are shown on the Results tabsheet The parent of the set of nodes is the main session node child nodes represent the individual populations from which the analysed traits are selected the grandchild nodes represent the traits themselves and the great grandchild nodes represent the linkage groups that are analysed For the permutation test there will also be a Genome wide node as a sibling to the linkage group nodes representing the genome wide results of the test The main session node has a serial number and between brackets the abbreviation of the analysis Table 1 All session details like parameters and population summary data are listed in the Session Info tabsheet Nonparametric mapping Kruskal Wallis analysis Nonparametric means that no assumptions are being made for the probability distribution s of the quantitative trait after fitting the QTL genotype For the nonparametric mapping method MapQTL uses the rank sum test of Kruskal Wallis see e g Lehmann 1975 ch 5 when a locus segregates in only two genotype classes such as in a backcross this test is equival
76. ssify in another way e g when there is dominance A classification type can optionally be given in the loc file in between the locus name and the genotypes to force a certain classification The classification type codes are given in Table 8 The classification type must only be given when a classification other than the default is desired In fact this is only necessary when there is dominance or in the case of population type RIx MapQTL does not allow classification types other than the default and optional types for the population and or segregation type If there is only the default classification type then a classification type need not and cannot be given The defaults and the options are shown in Table 9 Examples 1 and 2 are demonstrations of a locus genotype file 46 Data files Example 1 A locus genotype file for an F2 population 12 March 1995 this is a ridiculously small data file but it serves only as an example name some_demo popt F2 these data are from an F2 population nloc 2 the file contains data on two loci nind 6 and six plants RFLPOS5 this is a locus name aahba b these are the genotypes of the six plants RFLP67 a c classify this locus into a and c accac a Example 2 A locus genotype file for a CP type population 12 March 1995 this is another ridiculously small data file again just an example name what_a_ demo popt CP it is a CP type of population nlo
77. stance one could start with interval mapping to find areas with higher LOD scores and use a subset of loci in those regions as starting set Another option possibly combined with the previous is to start of with one linkage group select a locus every 10 or 20 cM for the starting set perform the automatic cofactor selection and then go forward to the next linkage group while keeping the resulting final selection of the previous linkage group easily done with the cofactors file and the Cofactors tool and so on until all linkage groups have been done Permutation test In order to determine the significance threshold of the LOD score or the deviance it is possible to use the permutation test This is a resampling method to obtain empirical significance threshold values Churchill amp Doerge 1994 MapQTL offers this method for interval mapping Over a set of iterations in this case permutations the frequency distribution of the maximum LOD score or deviance is determined In each iteration the quantitative trait data are permuted i e sampled without replacement over the individuals while the marker data remain fixed Subsequently interval mapping is done on the thus obtained data set The maximum LOD score or deviance over each linkage group as well as over all linkage groups the genome is observed in each iteration After a large set of iterations at least 1 000 but preferably 10 000 or more an estimate of the frequency distrib
78. t contain spaces but only the first twenty will be used Data Data files 49 Example 4 A quantitative data file tiny experiment tomato mildew 94 t 3 three traits nind 5 five genotypes s missing values are indicated with a nr indiv number non numerical trait not to analyse for QTLs length length to the 5th leaf averaged over 3 cuttings mildew mildew is the disease score averaged over 3 cuttings 94 1 69 3 a 94_2 75 4 5 94_3 54 aod 94 4 66 94 5 71 1 1 fields of non numerical traits may also be up to 99 characters long Because the trait name is used as part of the output file name a trait name should only contain legal file naming characters this is operating system dependent In the data body the values of all the traits are given grouped per individual Contrarily in the loc file the data are grouped per locus Although the layout is completely sequential for explanatory reasons it is best to look at the data body as a matrix In this matrix the columns represent the traits while the rows represent the individuals What happens if NTRT or NIND are incorrect If NIND or NTRT are larger than the actual numbers MapQTL will try to read beyond the end of the file which will lead to an error message If NIND is smaller it is not allowed to be smaller then the NIND value in the loc file then the traits will be read correctly for all NIND individuals but the program will find more
79. t menu and from a tool bar button Wl The results of a calculation session can be inspected by selecting the requested trait node in the sessions treeview the results will be shown on the Results tabsheet as a table and as a chart or set of charts on the Results Charts tabsheet An exception here are the results of automatic cofactor selection which are shown as plain text on the Results tabsheet and are not shown graphically The Results Charts tabsheet contains a set of two subordinate tabsheets one for the control of the charts and one for the actual charts There are many features of the charts that can be handled using this subordinate Control tabsheet The current view of the contents and results panel except the chart control tabsheet can be printed exported to file and copied to the MS Windows clipboard to enable the pasting into for instance an MS Word document This can be done using the Print option of the File menu and the Export To File and Copy To Clipboard options of the Edit menu The tool bar has buttons to perform these functions 5 5 a respectively When one or more rows in a table are selected or when there is a text selection in a plain text view the print export and copy functions are performed on the selection only pressing ctrl A will select all of the current view Charts are exported in the Enhanced Windows Meta File format which as an MS Windows standard can be used in many other applications Prior to pr
80. t of just a few loci neighbouring the significant locus 5 Click on the Calculate button 6 When finished load the resulting set of cofactors from the file using the Cofactors Tool button while clearing the currently checked loci prior to the loading 7 Repeat the steps 4 to 6 each time going to the next linkage group until all groups are done This way all groups are searched one by one while retaining the significant loci When significant loci are detected the analysis gains in power along the way Therefore theoretically you should go back to group 1 again and keep on repeating steps 4 to 6 until nothing changes anymore 8 However it is advisable to do a round of MQM mapping first because that will reveal some false positive cofactors since the P value is set to a non stringent value 0 02 as not to miss out on linked QTLs The set of cofactors that you should have found at this point is m9 m13 m32 m49 instead of m53 m75 m96 these are nearly all the same as found in the forward selection approach above and some additional loci on 30 Tutorial groups 7 9 and 10 m145 m146 m172 m173 m180 m183 m193 9 Load this set of cofactors and do MQM mapping Plot the cofactors and draw the significance line at 3 8 LOD and inspect the results charts 10 The LOD scores on groups 7 9 and 10 are all below the significance threshold therefore we can remove their cofactor loci as being false positives Do this redo t
81. ta of a population and 3 the map data The formats of data files used by MapQTL are described thoroughly in the Data files chapter p 39 Data for demonstration are available in the DemoData subdirectory of the program directory typically C Program Files MapQTL5 There may be more than just one population and more than just a single map loaded into a project After successful loading of data into a project the Populations tabsheet on the navigation panel will show the populations with their traits and genotypes as nodes in a treeview the Maps tabsheet will also show the maps and their linkage groups as nodes in a treeview In addition the Populations tabsheet has a node called Common traits with as child nodes the traits available in all loaded populations NB Traits within the quantitative trait data set that contain some non numerical data will show up as nodes with a green font and icon these cannot be used for analysis The selection of a node within the navigation panel enables the inspection of its data in the corresponding tabsheet of the contents and results panel The names of the selected nodes are given in the three stacked status bars at the bottom of the navigation panel In addition to the tabsheets for the loaded data and the results of analyses the contents and results panel also has a Project Info tabsheet with an overview of all actions done within the project and a Project Notes tabsheet on which you can
82. tainty in the genotypic marker information V From this partitioning the GIC is defined as GIC Vy 1 Vo Vo Ve 1 Vo For each population type the GIC has specific formula s For an F2 we use a QTL with the heterozygote strictly intermediate expected means of the A H and B genotypes of the QTL are 1 0 and 1 respectively From the neighbouring markers the probabilities for the QTL genotypes can be calculated 7 7 m Using these probabilities expected means and the standard formula for the variance V is the following sum over N individuals N Ve gt r 2 2 0 2 1 x m 047 D n Il ji r r 7 75 mz ll us n It can be seen that this variance is due to uncertainty in the QTL genotype when there is no marker information the probabilities 7 7 2 are the Mendelian expectations 0 25 0 5 and 0 25 respectively resulting in V V 0 5N whereas when the marker is on top of the QTL the probabilities for any individual are either 7 1 7 0 2 0 or 7 0 7 1 1 0 or 7 0 2 0 2 1 resulting in V 0 i e there is no variation due to uncertainty Applying the above definition of G C the formula for the F2 is GIC 1 2 r 2 n 1 IN n l This formula also applies to population type RIx For the population types with two genotypes BC1 DH DH1 HAP HAP1 the following formula applies N GIC 1 4 x 2 N n l with X and Y being the two genotypes
83. the parameters input and the data read from the files are given on the Session Info tabsheet The Results tabsheet lists the frequency distributions of the LOD score or deviance for all analysed selected linkage groups as well as the genome wide frequency distribution The distributions are given for intervals of 0 1 LOD units or 0 5 deviance units in size starting from 0 0 upto the value to accommodate for largest value that came about The total number of permutations is corrected for the numbers of cases where singularity or perfect fit occurred The frequencies are presented as interval counts cumulative counts relative counts and relative cumulative counts under the following headers in logical order Group the linkage group if this shows GW these are the genome wide results Interval the upper exclusive boundary of the interval into which a single permutation result can be classified the lower inclusive boundary is given by the previous value in the table e g the interval labelled 1 1 ranges from 1 0 inclusive to 1 1 exclusive Count the absolute count the number of permutations that had a result i e a maximum LOD score or deviance in the interval as defined above Cum count the cumulative count the number of permutations that had a result in the interval or in any lower value interval Rel count the relative count the absolute count divided by the total number of permutations Rel cum count the relative cum
84. tion of errors But occasionally some data groups may be so large that they don t fit on a single line Line structured means that data belonging together have to reside on the same single line For instance in the map file the locus name and its map position must be on a single line Sequential means that the data are read from left to right from top to bottom and there is no requirement to group data on a single line For instance in the locus genotype file the genotype codes belonging to a single locus determined in a large population may not fit on a single line and often have to be continued over several lines Of course it is a good measure to obtain proper readability by suitable spacing Some data files contain in the top of the file instructions regarding the contents of the data file e g the number of individuals and the number of loci This part of the file is called the header The program is indifferent to the order in which the various instructions in the header are given The header always has a sequential structure Some data elements are of fixed length while others are of variable length For instance locus names may be up to twenty characters long but they may also be shorter In order to read variable length data fields they must be separated from other data fields by whitespace On the other hand fixed length data fields need not be separated by whitespace although it is allowed and often to be recommended For instanc
85. tlL nl Martinez O amp R N Curnow 1992 Estimating the locations and the sizes of the effects of quantitative trait loci using flanking markers Theor Appl Genet 85 480 488 McLachlan G J amp K E Basford 1988 Mixture models inference and applications to clustering Marcel Dekker New York Rebai A B Goffinet amp B Mangin 1994 Approximate thresholds of interval mapping tests for QTL detection Genetics 138 235 240 Reyes Vald s M H amp C G Williams 2002 A haplotype approach to founder origin probabilities and outbred QTL analysis Genet Res Camb 80 231 236 Snedecor G W amp W G Cochran 1980 Statistical Methods Seventh edition Iowa State University Press Ames Iowa Stam P amp J W Van Ooijen 1995 JoinMap tm version 2 0 Software for the calculation of genetic linkage maps Plant Research International Wageningen the Netherlands Titterington D M A F M Smith amp U E Makov 1985 Statistical analysis of finite mixture distributions Wiley New York Van Ooijen J W 1992 Accuracy of mapping quantitative trait loci in autogamous species Theor Appl Genet 84 803 811 Van Ooijen J H Sandbrink C Purimahua R Vrielink R Verkerk P Zabel amp P Lindhout 1993 Mapping quantitative genes involved in a trait assessed on an ordinal scale A case study with bacterial canker in Lycopersicon peruvianum In J I Yoder Ed Molecular Biology of Tomato Technomic Publishing
86. type codes for population types F2 BC1 and RIx are given in Table 3 Those for population types DH1 and HAP are identical to these albeit that the heterozygous and dominant genotypes are excluded Table 4 The genotype codes for a DH or HAP population are identical to those for DH1 and HAP1 but have a slightly different meaning since the parentage of the alleles is not relevant Table 5 42 Data files Table 2 Population type codes Type Description F2 an F2 population the result of selfing the F1 of a cross between two fully homozygous diploid parents BC1 a first generation backcross population the result of crossing the F1 of a cross between two fully homozygous diploid parents to one of the parents RIX a population of recombinant inbred lines in the x th generation the result of inbreeding an F2 by single seed descent RI2 is equivalent to an F2 DH a doubled haploid population the result of doubling the gametes of one heterozygous diploid individual linkage phases originally possibly unknown DH1 a doubled haploid population produced from the gametes of the F1 of a cross between two homozygous diploid parents HAP a haploid population the gametes or derived individuals of one heterozygous diploid individual linkage phases originally possibly unknown HAP1 a haploid population derived from the F1 of a cross between two fully homozygous diploid parents CP a population resulting from a cross between two heterogene
87. uced default file name extensions for the various files The default extensions are given in Table 10 Example 5 A cofactors file ncof 3 three cofactor markers 7 group 2 rapd23 rapd19 7 group 5 rflp11 Table 10 Default file name extensions File Extension cofactors file cof cofactor monitor file Cmo locus genotype file loc map file map output data file mqo project directory mqd project file mqp quantitative data file qua Lists and references Lists and references List of figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 User interface Populations tabsheet with the DemoF2 population Maps tabsheet with the DemoF2 map Results Charts tabsheet with subordinate Control tabsheet visible LOD profile on linkage group 1 with LOD values above the significance threshold 3 8 some distance away from the cofactor marker m9 List of tables Table 1 Table 2 Table 3 Table 4 Table 5 Table 6 Table 7 Table 8 Table 9 Table 10 Analysis abbreviations Population type codes Genotype codes for population types F2 BC1 and RIx Genotype codes for population types DH1 and HAP1 Genotype codes for population types DH and HAP Segregation type codes for population type CP Genotype codes for a CP population depending on the locus segregation type Classification type codes Ratio is the expected segregation ratio Default and optional classification types Default file nam
88. ulative count the cumulative count divided by the total number of permutations Tutorial 21 Tutorial In this tutorial you will be taken through the most important steps of a QTL mapping project using a simulated data set of an F2 population that has several segregating QTLs including two linked QTLs The first thing to do after starting MapQTL is to create a new project Use the New Project function from the File menu You will get a dialog in which you are prompted for a file name under which to save the project this file name is also used for the project subdirectory name if necessary change the directory where the dialog is pointing to and enter tutorial in the dialog s File name field Click on the Save button this will create your project file tutorial mqp and in addition the project directory tutorial mqd which will contain all internal files of MapQTL for this project a new project is just a new workspace to store results Now you have a new project you can load the basic data the marker scores the map and the quantitative data Make sure the Populations tabsheet is shown in the navigation panel and use the Load Data function from the File menu A dialog will prompt you for a quantitative data or locus genotype file go to the DemoData directory which is a subdirectory of the program directory typically C Program Files MapQTL5S and find the locus genotype file DemoF2 loc and click on the Open button Next you w
89. used for JoinMap 3 0 Van Ooijen amp Voorrips 2001 although the format for population type CP is different from JoinMap 2 0 Stam amp Van Ooijen 1995 MapQTL also reads and interprets JoinMap 2 0 CP type files correctly Data file characteristics Here we give some important general features with respect to the data files for MapQTL The various data files themselves will be described in detail in subsequent sections For the sake of readability the data files may contain extra so called whitespace wherever found appropriate this is not allowed however within the various instructions indicators locus and file names etc Whitespace is a sequence of one or more of the next characters space tab newline linefeed carriage return vertical tab and formfeed The software is indifferent to the use of lower or uppercase both in the instructions and in the actual information It is possible and good practice as well to put relevant comment 40 Data files mon in a data file To make a comment line place a semicolon at the beginning of the line to put comment somewhere in a line place whitespace followed by a semicolon Anything on the line behind the semicolon will be ignored by MapQTL The layout of the various files is either line structured or sequential The choice for a particular layout has to do with readability by eye and the amount of data that belongs together Good readability is a proper measure for the preven
90. ution of the maximum test statistic LOD or deviance under the null hypothesis no QTL is obtained The number of permutations can be set with the Analysis Options of the Options menu The results are presented as frequency tables 20 Using MapQTL absolute relative and cumulative in the Results tabsheet the frequency tables are given per linkage group and genome wide genome wide meaning over the set of analysed i e selected groups as a whole To determine the significance threshold one first has to decide upon what P value to use and whether or not to use individual thresholds per linkage group or to use the genome wide threshold Next do the permutation test for the required linkage groups In general for standard applications the genome wide including all groups threshold with a P value 0 05 or 5 is required This means that we have to find the interval in the results of the permutation test where the relative cumulative count is 1 0 05 0 95 and take its upper boundary which is given as the value under the header nterval as the significance threshold value to use As the exact relative cumulative count 0 95 is not always present the first higher value that is realised must be taken merely to be on the safe side An alternative method of getting the significance threshold is described by Van Ooijen 1999 often this method gives very similar answers as to which threshold value to use Permutation test output A summary of
91. y also click in the treeview area use the keyboard arrow keys to navigate to the gtrait node and then press the space bar to specially select this node try this e Ifthe trait and the linkage groups are specially selected i e highlighted in red and interval mapping is selected in the analysis selector the calculate function is enabled notice that the Calculate button IE and the Calculate option in the Calculate menu are enabled i e not greyed out e Click on the Calculate button 24 Tutorial e Observe that the Sessions tabsheet is automatically selected and gets filled with various session nodes and that the progress bar on the status bar gradually proceeds while the calculations are being performed e Once the calculations are finished the Results tabsheet is automatically selected and filled with the outcomes of the analysis Inspect these results Similar to the Map Info tabsheet you can navigate through the results using the linkage group nodes but here using those in the Sessions treeview If the group nodes are not visible click on the symbol before the gtrait node in the Sessions treeview Try this e Take a look at the Session Info tabsheet to see what parameter settings and data were used in the current calculation session e Studying the results as charts is possible with the Results Charts tabsheet select this tabsheet Figure 4 Notice that it consists of two subordinate tabsheets one for the control of the cha
Download Pdf Manuals
Related Search
Related Contents
DITORIAL Soundmaster TR 100 PDF user`s guide HMP2030 Manual DE_EN.indd sterilux USER ` S MANUAL Nikon Coolpix P90 12.1MP Digital Camera with 24x Wide Angle Optical Vibration Reduction VR Zoom and 3 inch Tilt LCD Copyright © All rights reserved.
Failed to retrieve file