Home

BScEindverslag - Repository TU Delft

1. AnnotationtParser addRow AnnotationProbe figure 16 UML of the alignment tool HParse in file String lt unspecified gt ClinicalParser Parse in file String lt unspecified gt AnnotationProbe JFileChooser ClinicalPanel alignp AlignmentPanel FileChooser DataFileChooser AttributeNames list string selected bool addAttribute in atiname string in track Track getTrack in attname string Track removeAttribute in attname string addProbe write in attBoolean in upperT in lowerT void probes list ValueProbe write in name string in upperT int in lowerT int void add in probe ValueProbe contains in chr int in start int in end int um in other Track devide in devider int ValueProbe p Prol value float write in upperT int in lowerT int void 1 probe Probe values list float add in value float void getProbe ValueProbe 1 be ProbeRow 1 Al be robe Fehr int Parse in file String lt unspecified gt name string sta or 11 Appendix Pathway class diagram In this appendix part is shown complete UML class diagram of this pathway part Details explanation about the processes of this class diagram is elaborated in chapter 3 MultipleTest int numberOfProbesets Multiple Test int noProbes numberOfDigits int Main csv ImportCSV
2. An annotation file provides info about a specific probe set Annotation files are always linked to a specific data file and most often be provided alongside one Just as in the data file the rows represent probe sets and the first column provides the probe set ID The columns however represent properties of the probe set such as the start and end location on the genome or the chromosome on which it is located Probeset start end chromosome probe1 50000 50200 1 probe2 60000 60200 1 These columns can be in any order but they need to have specific names to be recognized The Alignment tool requires the chromosome number start and end position of the probe set It will search for columns containing chr start or end for the chromosome number Alternatively a single column named Alignment can be provided This column must contain the chromosome number start and end for each probe set in the following format Chromosome start end For example 2 50000 50200 Clinical file The clinical file is used to provide info about specific sample patients Each row represents a sample with the first column giving the sample ID The folowing rows provide info about the patient For example sample id gender age sample1 male 48 sample2 female 36 Part 2 Correlation The user interface The user interface consists of 3 parts 1 Loading the annotation and data files 2 Loading of the clinical data file 3 Selecting genes
3. Suppose that there are 20000 number of probesets for every probeset that is probeset probeset probeset a t test is being applied 20000 p values from these t tests will be found This t test process could be found in TtestTwoGroups java For Example in Table 1 four samples are found that are positive to AMLgroup1 and in Table 2 six samples are found from AMLgroup2 Probeset Sample Sample Sample Samples Probeset X XXXXXX X XXXXXX X XXXXXX X XXXXXX Probeset X XXXXXX X XXXXXX X XXXXXX X XXXXXX Probesetzo000 X XXXXXX X XXXXXX X XXXXXX X XXXXXX Table 1 Samples from AML group 1 Probeset Sample Samples Samples Samples Samplez Sample Probeset X XXXXXX X XXXXXX X XXXXXX X XXXXXX X XXXXXX X XXXXXX Probeset X XXXXXX X XXXXXX X XXXXXX X XXXXXX X XXXXXX X XXXXXX Probesetzo000 X XXXXXX X XXXXXX X XXXXXX X XXXXXX X XXXXXX X XXXXXX Table 2 Samples from AML group 2 Next step these 20000 p values might have to be corrected using FWER multiple test depends on what the user choses Class MultipleTest java will handle this multiple test These 20000 found p values would then have to be filtered or thresholded using alpha that is filled in by the user a standard alpha of 0 05 would be displayed in the field but users can fill another alpha value Suppose that 100 probesets are found from this thresholding process then this pat
4. a P Type Error B P Type II error FWER or Family Wise Error Rate is then the probability of at least one type under the global null hypothesis FWER P V gt 1 Back to the example above suppose that p values are approximately around 10 then the end result of p values after being corrected by 10 will be 10 Pa P N FWER Last step of this process is to generate config files that will be used as input for circos given genesets from tests mentioned above 3 4 Implementation Main ListOfAllSamples GroupAndAnnoCrossCheck PathwayGUI ListOfPositiveSamples A AMLDualList MultipleTestGUI ThresholdPValues AMLGroups DataStore MultipleTest importGeneSets ProbesetAfterThreshold GSEA CircosinputFile Figure 12 Simplifed UML diagram of pathway part Figure 12 displays global UML diagram of pathway expansion with its main classes important classes For more details of this UML please see appendix at the end of this report 3 4 1 Reading and Importing Files Data in CSV file format will be read and imported using H2 Database engine H2 is a relational database management system written in Java and is available as open source software This database can be embedded in Java applications or in client server mode It is possible to create both in memory tables as well as disc based tables and these tables can be temporary tables or even permanent H2 It is decided to use the disc based tables in
5. two AML groups and their significantly related pathways according to the the Molecular Signatures Database MsigDB e Alignment The final expansion is a tool that can load the data files and visualize them in Integrative Genomics Viewer IGV This is a genome browser where the data is shown on horizontal tracks aligned based on their location on the genome This software users manual describes how to use these three expansions to HeatMapper Part 1 Input This first part of the manual deals with the input files used in all three tools The different input files used can be divided in three catagories data files e annotation files e clinical files However they all have the following characteristics They are comma seperated value tables csv where each element in the table is seperated by a Numeric elements should use a as a decimal seperator Data file Data files provide the values of sample and probe set pairs The correlation and pathways tools use GEP and DMP data files The alignment tool uses any data file that follows the format outlined here In a data file the first row provides a list of sample names Each row after that denotes a probe set The first column provides the probe set id for each probe set and further columns provide the value that the probe set has in that corresponding sample An example is given below Probeset sample1 sample2 probe1 2 2 1 2 probe2 2 4 0 8 Annotation file
6. uFrame Navigator X a 5 S a LI Ml El FormCorrelatonPratotype et B 212209200 Ml 41 E Other Components Select al clot Set gt E o E prnl r Default Cursor z Selected genes Intersection of al genes Genes within selected chromosomes ca ER Delmiterts Correlate e iria RA A ne E Tasks x a G Description Fie Location al Pjr CorrelationPrototype java n src correlation main vew CorrelationPrototype java 5 5 P l TODO 1 n all opened projects A Figure 5 NetBeans IDE for creating user interfaces In Figure 5 the NetBeans IDE is shown On the right hand side are components which can be dragged into the interface in the middle It is also possible to create custom components these should be dragged into the interface in the middle from the source tree on the left 2 6 3 Back end The implementation of the back end consists of three main parts 1 Parsing input data 2 Manipulating the input data 3 Creating a visualization of the input data based on the input 2 6 3 1 Parsing the input data In total there are five different input files to be parsed all in CSV format e GEP Affymetrix annotation file o information of each probe set GEP data file o measurements for multiple samples for each probe set DMP annotation file o information of multiple probes in each probe set DMP data file o measurements for multiple samples for each probe set e Clinical data file o information about each sample The attrib
7. 4 Datta Mile tac 4 ANNO lia AAA AA 4 Clinical a ad 5 Rart 2 Correlatio A A A A A a 6 The user nta dr 6 A Ut eSEE E E E AA E TAA A T E E 7 Filtering on Clinical Data anainn as a ai dd 8 C rr lation OPTIONS ata SA A AA ee 9 Par PA Wii A a aan 11 File OPUta da a a LA a A OA EA AA a 11 AME groups selection nnna n e aeaa Ee cnssulavdesenentesidesiesdidadevesbeatescvexss 12 Multiple T est FWER dira ld E RA aa ves 13 Gene Set Collections Chosen misiista eaae aaa aa aan 14 Displaying Result Using CirCOS oooocooccnnnnnnccnnnnnnnnnnananrnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn ono nnnnnnnnnnnonanenos 14 Part Ay ALi SPIO ri tie si asa cd 15 LOANS MATUS ia i A elisa sate AA eel eee 15 Data tV Pe OPIO oaidcinai cs 16 Selecting SAM PIES ii AA Ai latas 17 Merging samples taa AS a AE 17 The clinical data searches add 17 Rorre Na AAEN A E Sees estan a A MA a ee RAR es 18 Introduction Currently HeatMapper draws heatmaps and displays clinical data next to the heatmap but it is not clear enough for the researchers what each matrix element represents in human genome Therefor this expansion to HeatMapper is made This expansion provides three manners of visualizing DNA data e Correlation This expansion is an environment in which the user can select genes and samples to correlate and consequently visualize the results in a circular diagram e Pathways This second expansion provides another view also in a circular diagram between
8. Exit button will exit the application e Pressing Next button will go on further to the next screen Gene Set Collections Chooser S Choose a GSEA file Please choose one gene set collection 8 c1 positional gene sets c2 curated gene sets c3 motif gene sets c4 computational gene sets c5 gene ontology GO gene sets Exit Next figure 4 Five gene set collections that one of them has to be chosen There are totally five gene set collections that are one at a time selectable namely e c1 positional gene sets e c2 curated gene sets e c3 motif gene sets e c4 computational gene sets c5 gene ontology GO gene sets Please choose one of these five collections and press Next button to go on further with the process or Exit button to quit the application Note It might take a little while to complete this step Displaying Result Using Circos Note It might take a little while to complete this last step Part 4 Alignment This final part describes the functions of the alignment tool The alignment tool is created to load genomic data into IGV from the format used in the other parts of this project On top of that the user can manipulate the data to influence what is visualized and how To ensure the correct functioning of the program make sure IGV is started up before opening the tool In IGV open the preference window under view and go to the advanced tab as shown
9. If the search value is empty the parser will print a list of all unique values in the search column This to allows the user to make a selection Once the user has successfully searched for a list of samples three buttons allow him to immediately select deselect or merge those samples The implementation for this is simple The clinical data searches simply copies the list into the select or merge field of the main gui and calls the doClick method of the corresponding button Visualizing the output When the user is done he can press a button to visualize the data The program will produce an output file for each Track in the bedGraph format In the bedGraph format the first line of the file gives some basic information about the track such as the name that has to be displayed The following lines provide the probe information with the chromosome number start end and value separated by spaces This output file is created using write methods in Probe Track Sample and Alignment These method must also check to ensure that only the required data is written The write method in Alignment will check what samples are selected It will then call the write method of the selected samples with the boolean and threshold lists as parameters The write method of the samples will then check the boolean list to see what Tracks must be written The write method of the Tracks receive the thresholds that apply to them These two integers are then pass
10. String str numberOfSamples int numberOfProbesets int getValu get GEPAnnotation DataStore ds String geneSymbol String probesetID GEPAnnotation getOnlySeaid ArrayList lt String gt getGeneSymbol ArrayList lt String gt removeDuplicates removeNan removeDash GroupAndAnnoCrossCheck ArrayList lt String gt groupA DataStore ds GroupAndAnnoCrossCheck doCrossCheck ListOfAllSamples all ListOfPositiveSamples pos getValueAfterCrossCheck double ProbesetAfterThreshold DataStore ds ProbesetAfterThreshold getProbesetNames String table int index ArrayList lt String gt crossCheck ArrayList lt integer gt DataStore Connection conn Statement stmt ResultSet rs DataStore importCSV String f String type void getGroupResult String groupType ArrayList lt String gt getProbeset ArrayList lt String gt getSampleResult String tableType ArrayList lt String gt getValues ArrayList lt String gt sampleNr String tableName ArrayList lt double gt GSEAChooserGUI closeConnection void ArrayList lt String gt geneSymbols String group1 String group2 GSEAChooserGUI ArrayList lt String gt genes String gr1 String gr2 A actionPerformed ActionEvent ae void DataStore ds run void ArrayList lt String gt afterCrossCheck String tableName GSEA String table ArrayList lt String gt afferX joinGroupNames String gr1 String gr2 String getAliCategories Ar
11. TTestTwoDoubles TTestTwoDoubles multiTTest double tenToThePower double ThresholdPValues couble pValues ThresholdPValues double pVal newLength double alpha int doThreshold fdouble alpha void actionPerformed ActionEvent ae void ImportCSV DataStore ds dolmportDMP String fileName void dolmportGEP String fileName void dolmportAnnotation String fileName void dolmportControlGroup String fileName void AMLDualList String table Type AMLDualList String type actionPerformed ActionEvent ae void MultipleTestGUI int numberOfProbesets MultipleTestGUI String type String gr1 String gr2 run void actionPerformed ActionEvent ae void dolmportGSEA String fileName void DMPAnnotation dolmportDMPAnnotation String fileName void ListOfAllSamples DataStore ds dolmportGEPAnnotation String fileName void DataStore ds String geneSymbol String SEQID DMPAnnotation getOnlySeaid ArrayList lt String gt getGeneSymbol ArrayList lt String gt removeDuplicates importRoutines String fileName String s void smg tableType ListOfAllSamples String str numberOfSamples int numberOfProbesets int getValue int post String getSampleResult String s ArrayList lt String gt ___AMLGroups DataStore ds ArrayList lt String gt amis AMLGroups castToArray String DataStore ds String tableType ListOfAllSamples
12. These sample names are then used for the select and deselect buttons The alignment class will then search for those samples and set a boolean called selected to true or false The merge function works similar A large text field will allow the user to enter multiple sample names A merge button will then call the the merge method of Alignment The merge method of Alignment will accept an array of sample names It will then look up the corresponding samples It will then retrieve all the tracks from the samples and call a helper method attributeMerge Track tracks for each attribute This helper method will return a new Track by averaging out all the probes of the input tracks The Track class has a sum Track other and divide int n method for this purpose The merge method can then use the newly created tracks to create a new Sample This Sample is then added to the ArrayList Clinical Data Searcher The clinical data searcher is a simple gui where a file location can be given with the help of a file chooser as well as a search column and search value A search button will create a clinicalparser object which will read the selected file searching the search column for the requested search value This parser will then print the corresponding sample names in a results area If the search column is empty the parser instead merely provides the first line of the csv file This gives a list of column names from which the user can select
13. a previously loaded file also generated a sample with the same sample id the track is added to that sample object instead of generating a new one Because of the high performance requirements for all these processes the entire loading process is done on a background thread Data manipulation Once the data is stored it can be manipulated The Alignment class contains a list of selected samples as well as a list of attribute objects These objects contain a boolean to see if it is enabled or disabled as well as the lower and upper thresholds When the write method of Alignment is called the attribute list will be used to determine what tracks and Probes are displayed This will be explained further in visualizing the input Each attribute object in the attribute list can be modified via the GUI A dropdown box will allow the user to select the attribute he wants to modify A button can set the enabled boolean to true or false Another button and two text fields will allow the user to set the thresholds of the attribute Lastly the user can delete the attribute completely The GUI also enables the user to modify the selected samples This is done by a large text field and four buttons The four buttons are select deselect select all and deselect all In the text field the user can enter multiple sample names separated by semicolons The smart thing about this is that these names can be copied easily from the input CSV files
14. defined this part of the expansion in more detail But while the application is functioning properly when used as explained in the manual it misses error messages and bug reporting functionality at some points it s not clear for the user whether an error has occurred in the background or that the program is still running Pathways This program part is able to produce a visualization in a circular diagram but there are still plenty of improvements need to be considered First of all much faster processing is inquired it takes at least 5 minutes from starting the program part until an image shows up Optimization is highly recommended regarding this pathways part Stability is also an issue there are not enough warning showed during process Program part will stop working when an error occurs without showing any warning it also has to do with a better user friendly user interface As mentioned in part 3 this program part imports data into a database Database file can be temporary stored in memory or in the disc Perhaps another approach of importing data might have worked better more stable and even more robust Alignments While not all requested features where able to be implemented the program can perform most of the required functions For other features such as changing the color of tracks or being able to highlight certain probes other output file formats or genome browsers would need to be considered The greatest problem with
15. specific sample respectively AnnotationProbes are used during the loading process and ValueProbes are finally stored until the program terminates The ValueProbes are stored in Track objects Asample object contains multiple of these tracks representing each attribute Finally an ArrayList of Sample objects is stored in the Alignment class which will perform most operations This can be seen in the UML Once activated the AlignmentParser will first call upon AnnotationParser to provide a full list of annotation probes containing the name chromosome number start and stop for every probe set but no values This data is obtained from the annotation file The AlignmentParser will then read each row of the data file each row corresponding to a probe set For each row it will search for the corresponding AnnotationProbe It will then create a ProbeRow objects containing the Probe object contained in the AnnotationProbe as well as a list of values for each sample column stored as a float A list of these ProbeRows is then loaded into the Allignment class The Alignment class also receives a list of sample names from the parser which matches the order of the ValueProbes Using this list and the list of ProbeRows ValueProbes for each sample and probe set pair are created These ValueProbes are stored in separate tracks for each samples The Alignment class will then generate a sample object to attach the track to If the sample already exist because
16. symbols will be displayed in the respective lists Selected genes Intersection of all genes Genes within selected chromosomes Delimiter s Correlate Enter genes of GEP data set GEP data set genes DMP data set genes Enter genes of DMP data set AR CEBPA Select all genes in GEP data set IV Select all genes in DMP data set The user can choose to put genes of interest in the text areas and there s a quick checkbox for selecting all genes of either the GEP or DMP data set Intersection of all genes This mode only has a Correlate button as it computes the intersection of gene symbols of both the GEP data and the DMP data Press the correlate button to start Selected genes Intersection of all genes Genes within selected chromosomes Genes within selected chromosomes After parsing has finished chromosomes which were parsed will be displayed in the select chromosomes list After 1 or more chromosomes have been selected the correlate button can be pressed and progress will be displayed in the progress bar Part 3 Pathways This part 2 describes step by step how to operate this pathways expansion in order to obtain an end visualization in the form of a circular diagram visualized using Circos File Input To enter this part just click on the Pathways Button on HeatMapper initial screen This pathway home screen will then appear j Pathway Please choose a csv DMP or GEP file t
17. the program is its performance In hindsight a better implementation would not have stored all the values in memory Instead the program should only read values when it needs to For example if the user wants to load in a data file select a sample and visualize only that sample a better implementation would first only read the sample names then allow the user to select the sample and then read the corresponding values to generate the output file that must be loaded into IGV Bibliography Grewal SI Rice JC Grewal SI Rice JC Regulation of heterochromatin by histone methylation and small RNAs 2004 Circos Martin I Krzywinski Jacqueline E Schein Inanc Birol Joseph Connors Randy Gascoyne Doug Horsman Steven J Jones and Marco A Marra Circos An information aesthetic for comparative genomics 2009 MolDevices Molecular Devices Inc Pathway Analysis 2011 http www moleculardevices com Applications Drug Discovery Pathway Analysis html GSEA BROAD Institute Gene Set Enrichment Analysis 2011 http www broadinstitute org gsea index jsp TTest Richard Lowry Concepts and Applications of Inferential Statistics 1999 http faculty vassar edu lowry ch11pt1 html FWER Mark J van der Laan et al Multiple Testing Part III Procedures for Control of the Generalized Family Wise Error Rate and Proportion of False Positives 2004 H2 H2 H2 Database Engine Version 1 3 164 2012 http www h2database com html main html JDBC Ora
18. ChangeListener o A listener which listens to which values at which attributes are selected correlation components annotationdataselectionpane correlation components annotationdataselectionpanel view GEPAnnotationDataSelectionPanel AnnotationDataSelectionPanel DMPAnnotationDataSelectionPanel AnnotationDataSelectionPanel o A user interface component for selecting a annotation and data file GEPAnnotationDataSelectionPanel o A user interface component for selecting a GEP annotation and data file DMPAnnotationDataSelectionPanel o A user interface component for selecting a DMP annotation and data file correlation circos api correlation circos api _CircosConfigCorrelation CircosConfig CircosConfigHypergeo CircoaDataF lo writer Karyotypas CircosColor l Gene CircosColors Karyotype Link e CircosConfig o An abstract class for generating Circos configuration files and for running the perl Circos script e CircosConfigCorrelation o A specific implementation of CircosConfig made for the Correlation approach e CircosConfigHypergeo o A specific implementation of CircosConfig made for the Pathways approach e CircosDataWriter o An abstract class which allows classes to write a data string to a given file e Links o A class representing a links data file for Circos e Karyotypes o A class representing a karyotypes data file for Circos e Genes o Aclass representing
19. Correlation The Pathways approach has a set of genes which are linked to two group and multiple pathways The set of genes can be enumerated at the group s category span and links can then be drawn between the groups and the multiple pathways modeled as category spans Part 6 Evaluation This chapter will discuss the results of the project This includes how the quality of the produced programs have been measured as well as the clients feedback Part of the evaluation process is testing Both automatic and manual testing has been used throughout the project to ensure that all the code does what it is supposed to do This includes unit testing where individual classes and methods are tested as well as manual functional testing where the entire program is run to see if it performs as required by the design The second part is ensuring that the program meets all the demands of the client Throuhgout the project the client has frequently tested the final build and provided featback such as new feature requests or usability issues Listed below are both the testing methods used in each part of the project as well as the clients comments regarding the final build 6 1 Correlation Complex methods have been tested using automated unit testing and testing of the user interface has been done with a Test Plan see Appendix Correlation Test Plan 6 2 Pathways 6 3 Alignment Most classes and methods in the alignment tool have been tested
20. G TUDelft HeatMapper Expansion Thesis BachelorProject IN 3405 Delft University of Technology Faculty of Electrical Engineering Mathematics and Computer Science Bioinformatics Sietse Au 1382756 Freek Kuethe 1522434 Firdina Lesmana 1037668 Foreword This bachelor thesis is written in completion of our Bachelor degree in Computer Science at the Delft University of Technology This is a collaborative project of the Department of Hematology of Erasmus MC Rotterdam and the department of Bioinformatics of the Faculty of Electrical Engineering Mathematics and Computer Science at the Delft University of Technology The aim of this bachelor project is to aid and provide researchers at Erasmus MC as well as other researchers in similar fields in getting more accurate and rapid interpretation of large scale genomic data We would like thank all of the involved supervisors during this project both from Delft University of Technology as well as from Erasmus MC especially Marcel Reinders of the Delft University of Technology and Erdogan Taskesen of the Erasmus MC Thank you for the useful input suggestions guidance and willingness to supervise our project Delft February 2012 Sietse Au Freek Kuethe and Firdina Lesmana Table of Contents COSSA A A a A A E wiesmaee Gen EE 5 Bart ATi OCT sa aina noaa e a o a E te 7 BATE ANE orot AULD AE EEEE AN A T E A E E S E aiaeelete 8 2 1 Pearson s correlation coc 8 2 2 Computation of correla
21. L type s It might be able to show a possible connection between a pathway and a specific AML type that is related to a mutated specific gene From several studies It is known that there have been several mutated genes discovered that might be the cause of AML such as CEBPa K RAS etc Users in this case researchers will be able to compare genes from patients or samples that belong to one of these groups with another group or with samples from a control group Comparing one AML type with a control group is mostly desired 3 2 Interactions and Design There will be some interactions between the users and this pathway part namely users can access this new pathway functionality by pressing the pathway button on the HeatMapper he will have to choose a csv file containing clinical data to use open that will be imported and read The file containing the annotations will automatically imported once the next button is pressed This csv file has to have comma as separator and the numbers have to use point to denote decimal Users will have to choose totally two files as input for this program part Another form of interaction is that it is also possible for the users to choose which genes of AML types they want to compare and visualize They will have to pick two genes from a list of several genes read from the annotation file and they will have to choose two of them or one from this list and control group to compare For example they can choose to co
22. METRIX_U133A 2 CEBPA AFFYMETRIX_U133PLUS2 BOTH_SNPCHIPS CEBPA_DM CEBPA_SIL CEBPA_SM CKIT_EXON17_D816 CKIT_EXON8_ID COMPLEX CONTROL_GROUP Remove figure 2 Choices of several AML groups User has to choose 2 groups to compare Functionalities of the buttons on this screen e Click on one group on the left column and click on Add button to move it to the right column or to select it e Click on one group on the right column and click on Remove button to return this group to the left column When two groups are added the Next button will become selectable Pressing this button will go on further to the next screen e Exit button will close the application Note clinical data DMP and GEP will also be automatically loaded to determine which AML groups are available to compare as shown in figure 2 In this figure one group is already added to the right column Multiple Test FWER On this screen there is a field to threshold User can fill in another value for this threshold A standard threshold is set to 0 05 This number must use point to denote decimal otherwise this application will not accept it Below this field there is a checkbox to decide whether to apply multiple test or not j Multiple Test Please fill in alpha 0 05 Apply multiple testing figure 3 Multiple test screen Functionalities of the buttons on this screen e Clicking on
23. ORY NAMES 3 8852 S232555 ro 2555 p i sn OAD roe 1001001 AASN yer gyoad B5t4 tavsan zooeeTo01901 se weer gon east fe ON 5 Se N Figure 14 Screenshot of a part of a circular diagram generated using CircosConfig This subset allows NN the display of gene names at specific locations on a circular diagram the display of links between 2 arbitrary locations on a circular diagram the display of category span names on a circular diagram the user to select any color each category span on a circular diagram the user to select any color for links bundling of links All of these functionalities are displayed in Figure 14 except for the bundling of links CircosConfigCorrelation implementation The specific implementation for the Correlation approach in Part 2 for Circos was quite straightforward As correlations are calculated each correlation value is linked to two genes Given a threshold a correlation value becomes a link when the correlation value gt threshold the names of the genes can be extracted by looking the data up with the GeneSymbols see Appendix and at the same time the chromosomes can be looked up by extracting the data from gene symbols as well These links genes and chromosomes are then put into the CircosConfig CircosConfigHypergeo implementation The implementation for the Pathways approach in Part 3 for Circos is even more simple than the CircosConfig
24. Using the GEP and DMP gene symbols sets the rows containing the gene symbols in those sets are gathered Internally there is a conversion from gene symbol to its respective gene unique identifier Correlation values are then calculated given the input and saved in an Object Example Input for calculation genes yp 1genel genes yp gene3 samples 1234 2345 4566 Conversion gene symbol to gene unique identifier geneUids p 1 gene geneUids py p 2 gene2 Rows containing gene1 for GEP and gene3 for DMP non relevant attributes are excluded for convenience rOW oep gene uid 1 1234 gt 1 2345 2 4566 3 rOW pyp gene uid 2 1234 gt 4 2345 7 4566 gt 6 Using the formula in 2 1 Pearson s correlation coefficient if we take X as 1234 gt 1 2345 gt 2 4566 gt 3 andYas 1234 gt 4 2345 gt 7 4566 gt 6 then the correlation value 0 6546 a positive correlation The correlation matrix would look like this 1 2 0 65 2 6 3 3 Creating a visualization of the input data based on the input See Part 5 Circos for the implementation of Circos Part 3 Pathways The expansion of the HeatMapper with visualization functionality pathways to visualize genes and possible pathways linked to those genes will be further elaborated in this chapter First problem analysis and its solutions will be explained Thereafter a UML diagram a function diagram and several screenshots of the us
25. a genes data file for Circos e CircosColors o A data structure for storing all the colors used in CircosConfig e CircosColor o A single color definition for Circos Gene o A data structure for saving the gene name chromosomal location e Karyotype o A data structure for saving the category span e Link o A data structure for saving a Circos link Appendix Class Diagram Alignment This appendix contains the UML class diagram of the Alignment part of the project The Alignment part of the project is explained in chapter 4 AlignmentPanel alignment Alignment attFileChooser DataFileChooser annFileChooser DataFileChooser loadFile in name string in location string 1 ocket AllignmentSocket gui AlignmentPanel read in attname string in file in annotation void HoadAttribute in proberows in samplenamesilist in atiname string addAttribute in attribute Attribute void removeAttribute in name string void setThresholds in upperT string in lowerT string attributeEnable in attname string selectSamples In selected list of samples deselectSamples in elected list of samples 1 selectAllSamples deselectAllSamples merge in elected list of samples 1 AllignmentSocket import in filedirectory string in name string AlignmentParser Parse in file in annotation in name lt unspecified gt addRow ProbeRow String ifi A
26. according to their location on a chromosomel Bioinformatics Sequence and Genome Analysis This way multiple types of data or different samples can be placed in different tracks to be compared This allows researchers to find certain regions of interest on the genome The goal for the Alignment part of the project is to visualize expression methylation CNV and SNP data per sample using tracks There are already programs able to do this called genome browsers These programs are great for comparing different types of data as long as it is links to specific locations on chromosomes They often already contain basic information about the different genes on the human genome A detailed description and comparison between different commonly used genome browsers can be found in the orientation report of this project While these genome browsers can certainly be used to visualize the data they often require very specific input formats and are not very interactive This part of the project will be to allow the user to not just visualize the data but also interact with it The requirements for this project will be to create a tool where the user after uploading the data can first select what types of data and samples he wants to show The user should also be able to set thresholds This will allow the user to focus on extreme values Lastly the user should be able to merge samples into groups These group samples will show the average values in their trac
27. always the sample id Aditional columns provide properties such as gender and age You can browse to such a file using the choose clinical data file button By giving a specific column to search in such as gender as well as a desired value such as male you can search for a group of samples that meet that search criteria E clinical Data Searcher i lol x choose clinical data file Column name search value search C deselect merge select figure 7 The clinical data searcher tool If you press search without giving a column name you will recieve a list of all column names and if you press search without giving a search value but only a column name you will recieve a list of possible values present in that column After you have recieved a list of samples you can directly select deselect or merge these samples from the Clinical data search window Loading into IGV Once you are done selecting samples and data types you can press finish This will load all selected tracks into IGV Tracks will be sorted on sample first and data type second To load the files into IGV the program will create output files called bedGraph files These files are stored into C AlignmentOutput After loading is done the result should look something like figure 8 Fie view Tracks Help 2101_9MP pata patch dl lil as adele dl chal cai an e ae a E aci l 2244_0MP fannst p
28. art of the project can be divided into three distinct functions reading the input allowing the user to manipulate the data and visualizing the output These functions are performed using a GUI called AlignmentWindow and a manager class called Alignment Reading the input The reading of the input works much alike the other parts of the project For each attribute RNA expression methylation CNP SNPs a file must be uploaded with the values for each probe and sample pair as well as an annotation file to map the probes to the correct location on the correct chromosome The Alignment class contains an instance of a class called AlignmentParser whichs reads and stores the data Due to the similarities an abstract class called FileParser which is used throughout the entire project to parse CSV files is inherited by the class AlignmentParser The AlignmentParser in turn also uses another sub class of FileParser called AnnotationParser To store the data the program uses a Probe objects each containing the chromosome number start and end value for a specific probe set These objects are basically little containers for the data For performance these containers need to remain as small as possible which is why they do not contain a probeset id or sample value Two other container classes annotationProbe and ValueProbe in turn each contain a Probe objects but add a name for the Probeset the Probeset id used in the input files and the value for a
29. cle JDBC Overview 2011 http www oracle com technetwork java overview 141217 html Bioinformatics Sequence and Genome Analysis Mount D W Bioinformatics Sequence and Genome Analysis 2004 IGV user guide Judy McLaughlin IGV user guide 2011 Appendix Problem Description Acute myeloid leukemia AML is a form of leukemia Leukemia is a type of cancer of the blood the adjective myeloid suggests that the cancer is related to the bone marrow or spinal cord There are different subtypes of AML each marked by certain combination of genetic mutations The Erasmus University Medical Center Rotterdam is conducting several researches in determining many subtypes of AML At this moment some researchers at the Erasmus University Medical Center Rotterdam use an existing visualization tool called HeatMapper a visualization tool that allows accurate and rapid interpretation of the data obtained by large scale gene expression profiling Our assignment will be to expand the HeatMapper which is developed by several researchers of the Department of Hematology and the Department of Bioinformatics at the Erasmus University MC Rotterdam The expansion of the HeatMapper should aid researchers at the Erasmus University Medical Center Rotterdam as well as other researchers in similar field The expansion should consist of three new functions e Calculating the correlation between gene expression profile data and DNA methylation profile data and consequentl
30. compare groups of samples A drop down box should allow the user to easily select an attribute and get some basic information about this attribute such as its name if the attribute is enabled and disables and what the current thresholds if any are The user should then be able to change this information via some text fields and buttons An effective way to allow the user to enable or disable certain samples is to keep a list of sample names The user should be able to add or remove names from this list The user should be able to add or remove multiple samples at a time perhaps by typing multiple sample names into a text field The merge function will work in a similar way The user can type multiple sample names into a text field Using a button the user can then create a new sample containing the average values of all the input samples and automatically add it to the list to be displayed Clinical data searcher To help the user find samples he might want to select or merge an extra tool will be created This tool can select a clinical data file This is also a csv file containing properties of each sample test subject such as age and gender After browsing to such a file the user can give a property to search on as well as the requested value To give an example consider a clinical data file with three columns The first column gives the sample ids The second and third column give the age and gender respectively The first row g
31. correlation would indicate that large values of X are associated with small values of Y and large values of Y are associated with small values of X 2 2 Computation of correlation coefficients The approach provides three ways to pick the genes to be correlated 1 Select 1 24 chromosomes and correlate the GEP data with the DMP data of the genes in the chromosomes 2 Build two lists of gene symbols one list of gene symbols in the GEP data and one list of gene symbols in the DMP data These genes are then correlated with one another It is possible that gene symbols in the GEP data do not exist in the DMP data vice versa Therefor the computation of the output is split up in calculating the correlation of three sets A B C visualized by the colors red green and blue respectively gt ii intersection x intersection intersection x DMP intersection intersection x GEP intersection OW ou Figure 1 Schematic overview of a matrix in which data has to be calculated DMP represents the columns of gene unique identifiers GEP represents the rows of gene unique identifiers DMP and GEP are split up in three sets A Intersection of GEP and GEP whereas only half of the matrix needs to be calculated as the values are mirrored values of the values in the red area in A B Intersection x DMP intersection this is a full heat map as each value in the matrix is unique C Intersection x GEP intersection this is also a full heat ma
32. de 26 Elnicaldata ar dd a O 27 VISA o 27 A D Implementar IE EEE SE OTEA 28 Reading th PUE A AAA a e 28 Data Mapu ot lion celo A actes 29 Nistalizino theo tp t da AA A di 29 PAC a 30 Sl WOY CICOS A A A A NT 30 5 2 Implementation asic isda eee eA at Ol ae wales oan da sve ral alan EE eRe ed Sst DRC EE E E 30 5 2 1 Str Of Ci COS ali 31 5 22 Circos APT SpeCiliCati OMe A ida 32 CircosConfigCorrelation implementation ccccecscecsseeesecesceeeseeeeeceseceseeceeeeneeeeseenseeeeneneees 32 CircosConfigHypergeo implementatiON ooconnocnnoccnococoncconnnconccnonononcconnocnnncnnnnconn cn conan nccnnnnncnns 32 IP ATU Gis Evaluation ennan O teed E test N 33 A A Ae A a a S aE a E EA aai 33 A n e EEEE a NEE EEA E E EE E RE EREEREER EROTIES EAAS 33 OS AOMEN a EET E a 33 Part 7 Conclusion and recomienda E es 33 Correlations mena in ia O A a A RA eaa Re a T a 34 Pathways ena edar A dd 34 A E E AEEA AAE AE E A E E A 34 Appendix Probl m Description ecssiscrnernnsinnnii ei a a E a e ii 35 Appendix Correlation class HIS diia 36 CA O nnii en i a e aaa a e EA SA sana e Ea EA a ER eee 36 conelation Mbresa O NE 39 correlation components seliniCal data sacar acces A a 39 correlation components annotationdataselectiONpANE cooocccnoccconccoonoconcconnnconnconon cn conan cn connncconnncno 40 Correlation CIT COS PESA NA ica 42 Appendix Class Diagram Aliment morosidad aran etica 44 Appendix Pathway Class AM A la 45 Appendix Correla
33. e same as in the other parts of this project Each attribute is imported as a CSV file with a corresponding annotation file The data file contains a large table Each row represents a probe set The first column give an ID and the following columns represent each sample The annotation file is also a large table that provides the chromosome number gene symbol start and stop for each probe set The tool will need to collect the data for each probe set out of both files by matching the names on the On the GUI the user should be able to browse to these files Once the location of both files have been chosen he can give the attribute a name and start importing it He can repeat this for multiple attributes He should also be able to delete attributes he no longer needs Interaction Once one or more attributes have been loaded the user should be able to do some simple tasks The user should be able to see what attributes are loaded He should also be able to see how many samples have been loaded and how many of those are selected He can then enable and disable certain samples and attributes This will give the user control over what tracks should be visualized He should also be able to set thresholds per attribute so that he can look for extreme values Finally he should be able to make sample groups where multiple samples are merged into one sample by averaging out all the tracks per attribute This could be useful when a researcher wants to
34. ed to the Probes write method which will check its own value to determine whether it needs to write If so it will write its own data as a line in an output BEDgraph file Using a socket these BEDgraph files are loaded into IGV via a port call This is done using a class called AllignmentSocket which handles the connection to IGV That means that if IGV is running and set to allow port commands this program can directly load the data into IGV to be visualized This option allows the user to make further adjustments to the visualization in IGV and compare the data to other tracks loaded into IGV Part 5 Circos Circos is a visualization tool used to generate circular visualizations in Part 2 and Part 3 5 1 Why Circos Circos is an established visualization tool for generating circular visualizations of data which contain a lot of relationships between objects 5 2 Implementation To integrate Circos into Part 2 and Part 3 a Java wrapper around the original Circos perl package had to be implemented In Figure 13 it is visible that there are 3 classes dedicated to export files Genes Karyotypes and Links Each of these writes an output data file for Circos while CircosColors writes its output to a predefined circos configuration file circos conf which is written to the user s disk by CircosConfig In the next section the structure of Circos is explained what it requires and what the Circos API implementation can provide After that
35. er The software pattern used will be the model view controller This is a way of separating data object logic program logic and user interface logic A package will be created for model view and controller In the case of the functionality to be implemented the model will contain data objects which hold data parsed from input files the view will contain the user interface implementation and finally the controller will contain the code for manipulating the data objects Packages correlation components annotationdataselectionpanel view correlation main parser H i A i AA 1 p i Y correlation main view T 1 i i E AAA i Y A z correlation components clinicaldata view y Y correlation circos api correlation main controller correlation components clinicaldata controller T 1 i i i A a o al o i i i y y correlation main model correlation lib correlation components clinicaldata model correlation components clinicaldata parser Figure 4 Overview of the code structure In Figure 4 the model view controller pattern is visible the views do not have direct control over the data they delegate high level instructions to the controllers which then manipulate the data in the models 2 6 2 Front end The front end is built with the built in drag and drop user interface editor of the NetBeans IDE By using the NetBeans IDE more time can be spent on the implementation of back end of the fu
36. er interface will be displayed Finally details about the methodology and the implementation will be furthermore elaborated 3 1 Problem Analysis It is essential for researchers to have a better visualization in pathway analysis because of the increased complexity of data sources Therefore is this expansion to the HeatMapper MADEx This expansion will be able to display pathways and their significantly related genes in a circular diagram This circular diagram will also be visualized in the sameway as Correlation part from chapter 1 using Circos Cell behavior is regulated by a complex network of intracellular and extracellular signaling pathways Researches and experiments have discovered the identification and characterization of the components that make up such signaling pathways Researchers have discovered many important biological pathways through laboratory studies and experiments of cultured cells bacteria mice and other organisms Many of these identified pathways have similar working in humans How these pathways are controlled how they communicate with each other and most importantly what happens when they do not function properly this is what pathway analysis implies MolDevices The main target of this project part is to get other form circular presentation of visualization This gives researchers another view of visualization which genes belong to a specific pathway and which pathways are significantly related to specific AM
37. f pathways that are significantly related to gene symbols in the two groups chosen before Gene symbols that not belong to a group in a GSEA file will be connected to OTHER Some screenshots of user interactions in pathway part are shown below Please choose a csv DMP or GEP file then a csv clinical data file DMP File GEP File eones Cesar Ji Figure 8 First screen after pathway has started User has to choose one csv file DMP or GEP data and one clinical data S Please choose 2 AML groups Selected groups AFFYMETRIX_U133A AFFYMETRIX_U133PLUS2 BOTH_SNPCHIPS Figure 9 Screenshot when user has to choose two groups within AML User can add one group at a time from left list to the list of selected groups The Next button will be enabled when there are two groups are on the right list 0 05 Apply multiple testing Figure 10 Screenshot when user has to fill an alpha to threshold found p values User can check or uncheck the checkbox for multiple test s Samples N Clinical data import _ Import data CSV Between Take 2 groups 2 AML groups 1 AML group and control group N T test 2 groups Paro PN 1 Multiple test P_ P N FWER m Threshold lt alpha Set diff exp genes Import gene sets Enriched x gene sets Input for Circos Circos Figure 11 Function diagram of pathway functionality This figure displays the whole process of pathway expansion
38. he GEP set and the black gene symbols are from the DMP set 2 4 Requirements 2 4 1 Reading Affymetrix annotation files Affymetrix annotation files are comma separated values CSV files generated by a company named Affymetrix The annotation files contain information about each probe from a DNA micro array Reading annotation files is essential in order to information about probe sets such as gene symbol name chromosomal location can be extracted which can be used for filtering selecting and grouping genes 2 4 2 Reading human sample files containing GEP or DMP data values Annotation files give information about each probe but in order to calculate correlation coefficients sample values are needed Each probe has n GEP samples and m DMP samples each sample has a unique identifier The identifier is essential because the intersection of the sample identifiers of both the GEP and the DMP probe set needs to be used as each sample represents a person it s not useful to see if person X s GEP value has any linear relation with another person Y s DMP value 2 4 3 Letting the user set a correlation threshold The user needs to set an absolute correlation threshold both negatively and positively correlated variables will then be displayed that will correlate the genes between GEP and DMP The more stringent value will result in less correlation between GEP and DMP 2 4 4 Letting the user select genes of interest With the same analogy a
39. hen a csv clinical data file f Annotation File Clear All Fields figure 1 Home screen of pathways part In this example a DMP csv file is already selected Functionalities of the buttons on this screen e First please choose an input file of a gene expression or a methylation data If itis a methylation data click on DMP file button for a gene expression data click on GEP file button e After choosing a DMP or a GEP file the Annotation file button will become selectable Please choose a clinical data file as input by clicking on the Annotation file button e Clicking Cancel button will return to HeatMapper initial screen e Clear all fields button will reset chosen file s Next button will become selectable when an appropriate input file and a clinical data file are entered pressing this button will go on further to the next screen Note It might take a little while to load the csv files completely AML groups selection After previous steps a screen with two columns left and right will appear The left column contains a number of AML groups which exactly two of these groups have to be selected Groups that have been selected will show up on the right column Due to the limitation of two groups being selected this program allows maximum two groups on the right column j Please choose 2 AML groups ML groups Selected groups AFFY
40. hway process would go on further with this 100 probesets These probesets use special probeset annotations that are different from the ones in the Gene Set Enrichment Analysis GSEA files These annotations have to be matched using either DMP annotation or GEP annotation file loaded earlier Results from this matching are gene symbols This matching process finds place in DMPAnnotation java or GEPAnnotation java Found gene symbols from steps above will further be matched with one of the five selectable GSEA files Only one file that has been chosen by the user will be loaded A matching process finds place after this loading process and it will return a list of pathways or categories that these gene symbols belong to These gene symbols and their pathways will then be used in CircosConfig java to automatically generate input txt files for Circos Two HashMaps have to be created in order to create an object The first one contains mapping between the two chosen groups and their gene symbols while the second one contains mapping between gene symbols to their related pathways Please refer to Circos implementation part 5 Part 4 Alignment 4 1 Problem Analysis It may be useful to the researchers to view certain parts of their data as graphs and compare it to other types of data This can be done using a DNA sequence alignment A sequence alignment is when the different data points such as probe sets for expression data are placed in a track
41. ignment with the highest matching percentage will be used Example chr6 30856165 30867931 95 63 p21 33 chr4 30856169 30867967 80 32 p21 33 In this case the first data of the first alignment will be used chromosome 6 with start position 30856165 and stop position 30867931 As a sanity check the column chromosomal location will be checked against the data with the highest matching percentage if the chromosomes do not match the entry will not be used Attribute 1 has its own respective column in the GEP Affymetrix annotation file As for attribute 5 the value is a string containing several gene symbols separated by a triple forward slash In order to aggregate aliased gene symbols gene symbols are given an internal unique identifier whereas multiple aliases of the same gene symbol can be identified with one unique gene identifier The GEP data file is similar to the DMP data file where the columns represent the value of a specific sample in a specific probe set A clinical data file is a file where the rows are samples and values and the columns the attribute name of said values The extraction is therefore trivial each row is represented by a data structure which has a attribute to value mapping 2 6 3 2 Manipulating the input data The data has been extracted to compute correlations The input for the correlations is a set of GEP gene symbols a set of DMP gene symbols and a set of sample identifiers
42. ile for the annotation or data file selector User selects file for the clinical data file selector ue v v c D Check if there are any files selected in the Check if there are any files selected in the Check if there are any files selected in GEP annotation and data file selectors DMP annotation and data file selectors the Clinical data file selector oO O Both annotation and data files have been selected A annotation and data files have been selected Asynchronically parse the annotation and data files while Asynchronically parse the data file while showing a Showing a progress bar for the progress of the parsing progress bar for the progress of the parsing y y c Click selected genes Click intersection of all genes tab Click genes within selected 2 chromosomes tab oO Ri T T User fills the DMP list and GEP list with gene symbols to be correlated User selects chromosomes to be correlated fe y Y y o Mark selected genes that exist in the parsed data in RS A Mark selected 5 a ee A R gt Click Correlate DUMON eeestis T the respective lists chromosomes 3 o E Y Calculate the Pearson s correlation coefficient for the selected gene sets for the selected samples from the clinical data a Iterate through the genes and create links for each Sc pair of genes that have a coefficient gt correlation o 2 threshold on ei So o gt co Generate data and configuration files E e A A A P s i o gt E HA Y Bacadionthana
43. in figure 5 Ensure that port commands are enabled for port 60151 Also consider how much memory IGV will need for your intended goal After that run the alignment tool File View Tracks Help x hgi amp y la y f General Tracks Mutations Charts Alignments Probes Advanced Proxy Y Enable port 18 Enable port to send commands and http requests to IGV J Edit server properties Reset to Defaults Fn A AEE httn fiav broadinstitute orafaenomesinenct figure 5 Setting IGV to allow port commands Loading in attributes The first two things you will see on the tools graphical user interface seen on figure 6 are the info window which is not displaying much info for now and the Add a new data type section Data types refer to different types of data files that can be loaded into the program such as expression or methylation data Any type of genomic data can be loaded into the program as long as a data file containing the values for all the probe set and sample pairs as well as an annotation file containing the chromosome start and end position for each probe set are provided These files must be provided in the same csv format as used in the other parts of this project To program will store the data per sample Each sample will have multiple tracks for each data type First you should give the data type a name This name will be used in IGV After that the select data File and select annota
44. ives the name of each column and all following rows the values for each sample The user can select gender as the search property and male as the requested value This will return the sample ids of all male samples These ids can then be used to select deselect or merge the samples as explained previously Visualization Once the user is done modifying the tracks he should be able to visualize the tracks This should be a relatively quick process so that the user can go back make new modifications and visualize again At first an implementation using awt and swing was considered but it is more practical to use an existing genome browser for this This will allow the user to also compare other data to the data visualized by the tool In the orientation report a comparison of different genome browsers is given Integrative genomics viewer IGV was chosen due to its simple java based api and the ability to run it locally instead of through a website IGV user guide The tool will create output files which can be loaded into IGV using a socket call If the user does not have IGV open at the time that these files are generated he can still import them manually into IGV later In both cases the results will look similar to the picture below ee ii q lo All f r l TA ine Figure 2 A screenshot of IGV showing different data tracks 4 3 Implementation The design of the Alignment p
45. ks HeatMapResultContainer o A data structure which saves a heat map in a one dimensional array CorrelationResultContainer o A data structure which saves correlations in a two dimensional array PRowContainer o A container for PRow objects GEP o A container for GEPRows it represents the data extracted from the GEP annotation and data files DMP o A container for DMPRows it represents the data extracted from the DMP annotation and data files PRow o A data structure which holds the variables needed to identify a row in a an annotation and data file GEPRow o A data structure which represents a row in the GEP annotation and data file DMPRow o A data structure which represents a row in the DMP annotation and data file GeneSymbols o A static class which is thread safe it contains data about every gene parsed into the application ChromosomeListModel o Model for an element in the user interface GenesListModel o Model for an element in the user interface InvalidElementException o An exception thrown when an element in a parsed row in either the annotation or the data file is deemed invalid correlation main controller SelectedGenesCorrelationController F StatusFeedBack gt CorrelationController ChromosomeCorrelationController CorrelationControllerPropertyChangeListener AllGenesintersectionCorrelationController PFileParserPropertyChangeListener lt T gt Prog
46. ks This can be useful when comparing groups such as control groups versus patients This tool should then be able to visualize the data quickly and allow the user to readjust the options 4 2 Design The alignment part of the project will be a simple tool where the researchers can import the data This data is then stored in tracks each assigned to a sample Each sample has multiple tracks for each type of data These different types of data are referred to as attributes The tool will have a simple one window GUI where the user can perform all the required steps before pressing a button and visualizing the data This can be seen in the function diagram shown bellow Let us first focus on the input Input attribute tables CSV files Input window Scanner at least one table 3d array with al attribute and probe Select n Select m Group samples Samples Attributes and mean Or all samples attributes array Lower and upper thresholds Optional Set Thresholds for each attribute Write BEDgraph files 3d array with sample attribute and probe y Create Tracks for Sample and Attribute names gt each sample and Chrm Start end value attribute pair y BEDgraph file Create feature y Import into IGV via Socket call Figure 1 function diagram showing the different actions a user can take Input The input looks th
47. ll samples by using the select all and deselect all buttons To select or deselect specific samples you must write their names in the field seperated by semicolons Lets give an example where you only want to display the samples 2281 2282 and 2283 The easiest way to do this would be to first deselect all samples and then enter 2281 2282 2283 in the text field Press select to select these samples The info window will always give up to date info on how many samples are loaded and selected You can also select and deselect via the clinical searcher which will be explained later in this manual Merging samples Merging samples works much the same as selecting samples Merging takes the tracks from several selected samples and creates a new sample with the average of all those tracks First you must give a name for the new sample You then write the names of the samples you want to merge seperated by semicolons and press the merge button If any of the selected samples have tracks of the same data type the new merged sample will recieve a track of that data type where the values for all probe sets in that track are averaged Merging samples can also be done via the clinical searcher The clinical data searcher The clinical data searcher as shown in figure 7 is a tool to allow users to find samples based on a clinical data file A clinical data file is a comma seperated value file where every row represents a sample The first column is
48. mpare CEBPa with the control group or CEBPa with K RAS After this interaction process they can also fill in an alpha to threshold p values found from the t tests before Users also have the possibility to choose if he wants to apply multiple test or not Under this alpha field there is a checkbox which he can check or uncheck Thereafter user has to choose one of five files c1 c5 that contains gene set collections These files are already downloaded from the Gene Set Enrichment Analysis site GSEA GSEA Chosen file will then be imported and added to the database automatically These five GSEA files are e c1 positional gene sets e c2 curated gene sets e 3 motif gene sets e c4 computational gene sets e c5 gene ontology GO gene sets Figure 11 shows a diagram of some actions the users need to perform in order to get the desired end visualization import clinical data choose 2 AML groups fill in alpha to threshold p values apply multiple test FWER choose 1 gene set collections Figure 6 User interaction diagram without multiple test FWER Finally in Figure 6 below an image that illustrates the expected end image that approximately will be displayed using circos OTHER y 7 hay gua CHR2Q14 ao CHR12Q21 o SS S Chy Re Loy AD g o oe Roy Qe ex G CHR19Q13 capita Figure 7 Expected end result displayed in Circos Outside the ring are categories o
49. nctionality Correlation NetBeans IDE 7 1 Be Edt yew Navigate Source Refactor Run Debug Profe Team Tools Window Hep ERE 2098190 e 278Y0D8 0 E StartPage x C CorrelatonPrototype java x CinicaDataSamplePicker java lt gt gt 5 Poete x El Sare oen noy AO AA As E Swing Containers i Panes E Tabbed Pane JL Split Pane Use the Connection Mode button n the trobar to establish a connection between components lt eae Tool bar panar 2 C intemal Frame 5 Layered Pane Swing Controls Correlation threshold 95 9 Bundle drcos inks ene a E Toggen h a Check Box Radio Button Z Button Group e GEP affymetrix annotation Cick here to choose fie status label U Combo Box Hust L Text Feld i gt Text Area EE scroll Bar QD Sider i3 GEP data samples Click here to choose file Bar El Formatted Field E Prd Fid 5 TE Spinner Separator T TextPane e DMP annotation Cicchere to Goose e eee E Edtorpane ij ree Teie Y E Swing Menus B DMP data samples Click here to choose fie Bs Menu Bar E Menu SI leis x an_ s O 2 8 Beans Binding beansbinding 1 2 1 jar Properties Binding Events Code 5 8 commons math 2 20 No status El EE 10k 1 7 Default Cinicaldata Click here to choose fie f defaultCloseOperation EXIT_ON_CLOSE E y Click the combo box to select an attribute Select items for that list m E
50. not properly develop into different types of blood cells such as neutrophils dendritic cells and macrophages It is characterized by the rapid growth of abnormal white blood cells in the bone marrow It is caused by mutations and chromosomal aberrations that obstruct the development of stem cells to properly functioning blood cells After years of research several subtypes of AML have been revealed Technological advances in DNA sequencing has allowed computerized processing of DNA data Raw data is hard to interpret for human researchers therefor there is a lot of interest in having computers display bio molecular data in visually attractive and informative way for humans This thesis will discuss the expansion of an existing tool for visualizing biological data HeatMapper Three extra approaches to visualize data will be built into application Each approach is elaborated in respectively the Correlation Pathways and Alignment chapter The three approaches are e Correlation There exist applications to create correlation matrices but it s mainly manual work done in general mathematical analysis packages Visualization of correlation matrices is mostly done by heat mapping but the disadvantage of heat mapping is that the context is lost it s not clear for the user what each matrix element represents in human genome The Correlation approach attempts to solve the contextual problem by allowing the user choose multiple groups of samples and cho
51. of interest and generate a correlation image S Design Preview CorrelationPrototype Correlation threshold h 95 JV Bundle circos links GEP affymetrix annotation cick here to choose file imbid GEP data samples Click here to choose file DMP annotation cick here to choose file status bbe DMP data samples cick here to choose file No status Clinical data cick here to choose file Click the combo box to select an attribute Select items for that list a Selected genes Intersection of all genes Genes within selected chromosomes Delimiter s Correlate Enter genes of GEP data set GEP data set genes DMP data set genes Enter genes of DMP data set PP Select all genes in GEP data set PP Select all genes in DMP data set File input By clicking on the text fields it s possible to choose the respective input files by selecting them in a file dialog Look in di neworiginaldata y 2 e a El dinical_data_GEP_and_DMP csv El dinical_data_GEP_and_DMP xlsx Recent Items al DMP_annotaties_hg19 csv EL DMP_data_344 csv A GEP_annotaties_hg19 csv GEP_data csv Desktop D My Documents Computer e Network Filtering on Clinical Data 10 x Correlation threshold h 95 JV Bundle circos links GEP affymetrix annotation s_viz neworiginaldata GEP_annotaties_ha19 csv Parsing aS GEP data samples Desktop data_viz neworiginalda
52. oll mpila at cadera makilka aas day My mael A IRA 2244 08P A lean ak a at 2255_DMP pt pol o 00 pao ape ol Le gph gg e o pl a to nent pas lios audi ici lc bss to ceca dl tic 2751_DMP Kostas paisa dd aask palyda olle wr i pt y adn shoal La alain ai figure 8 Example visualization in IGV
53. or example the intensity of methylation near a certain gene and the intensity of the gene expression of a gene Research shows that methylation may play a role in the regulation of gene expression Grewal Sl Rice JC Calculating the correlation between methylation and gene expression data may help to gain a clearer insight into the relationship of methylation in the regulation of gene expression First the idea of correlation will be defined and in what way it is applied to the data sets Then the problem analysis requirements design and implementation follow 2 1 Pearson s correlation coefficient The Pearson correlation coefficient is an indicator for a positive and negative linear correlation The correlation coefficient r is calculated by taking two variables say X Y Y 1 1 9 7 i 1 r n 1 s 8 5 where x y are the averages of X and Y respectively and where s sare the standard deviations of X and Y respectively where X x x Y Ypo Observe the following if x and y are predominantly above average OR predominantly below average then x x y y would be positive 1f x 1s predominantly below average and y is predominantly above average vice versa then x x y J would be negative What can be seen here is that a positive correlation indicates a general tendency that large values of X are associated with large values of Y and that small values of X are associated with small values of Y A negative
54. ose genes of choice within one data set to generate specific visualizations of which the context can be altered by the user e Pathways A genetic pathway is a set of interactions between multiple groups of genes who depend on each other to provide a function to the cell The Pathways approach aims to displays which pathways are significantly related to specific types of AML Users are allowed to choose two groups within AML which they can compare and visualize Just like the correlation approach the data is presented in a circular diagram Alignment Data is visualized by aligning multiple data sources Unlike the other two approaches the data is viewed not in a circle but in horizontal tracks each track contains data points of a certain data source such as a specific patient Each data point corresponds to a specific location on the genome and displays its value ona corresponding location on the track By aligning different tracks below each other one could easily compare different data sources while looking at specific places of the human genome The elaboration of each approach will contain an analysis of the visualization design of the functionality design of the user interface and implementation details and decisions Finally the thesis will end with an evaluation of each way of visualization followed by the conclusion and recommendations chapter Part 2 Correlation Different types of information can be measured from genes F
55. p as each value in the matrix is unique 3 Create a heat map of all like the red area marked with the letter Ain Figure 2 In a heat map both axes in this case the GEP axis and the DMP axis have corresponding elements genes Only half of the matrix has to be computed as the 2 half white is an exact mirror image of the 1 half red 2 3 Problem analysis The correlation approach allows the user to select a correlation threshold for a large data set and visualize this in a circular diagram It allows users to generate a circular diagrams given a threshold for the underlying correlation values Correlation between two variables in this case a DMP variable and a GEP variable can be computed by using the Pearson s correlation coefficient Correlation is usually visualized in correlation matrices however by displaying the correlation in a matrix the context of each correlation value is fades away the user will not be able to grasp the entire picture by just looking at the calculated correlation values A way to solve this is by visualizing each matrix entry as a relation between a DMP variable and a GEP variable as a link in a circular diagram see Figure 2 By visualizing the matrix entry in a circular diagram it instantly becomes clear to the user how strongly genomic ranges relate to one another given a threshold Figure 2 high threshold 0 8 of the entire genome filtered on a clinical data set The red gene symbols are from t
56. pen Clinical Data Searcher Finish select samples deselect samples select all deselect all figure 6 The main graphical user interface of the alignment tool Data type options In the data type options section loaded data types can be selected via the dropdown box You may then e set thresholds for the data type e enable or disable the data type or e delete the data type Thresholds determine what probe sets are shown or not shown when the data is visualized in IGV Probe sets with values that fall outside the threshold range will be ignored To set the thresholds edit the upper and lower thresholds valued and click set Thresholds What if you only want to see outliers If the upper thresholds is equal or lower then the lower Thresholds the program will only display outliers and ignore all probe sets in between the two values Note that the default of zero for both thresholds means that all probe sets are shown by default The next button in this section can be used to enable or disable the entire data type Disabled data types are entirely ignored during visualization If you permanently want to remove a data type you can unload it with the delete button Selecting samples The next section on the main panel is allow you to select what samples should be shown in IGV By default all samples are enabled when they are added You can select or deselect a
57. rayList lt String gt removeSlashAndSplit ArrayList lt String gt getGeneSymbolByCol ArrayList lt String gt matchGene String gene String col String Figure 17 UML Class Diagram pathway part Appendix Correlation Test Plan This test plan contains several test cases which are manually performed by a tester on program execution Test case 1 User clicks on choose file field Open File dialog pops up Test case 2 User presses OK after choosing a file in Open File name is written in the File dialog choose file field Test case 3 User has selected both an annotation and data File starts loading and fills in file for GEP or DMP the chromosomes in the chromosomes tab and selects the genes if any in the selected genes tab Test case 4 User has selected a clinical data file File starts loading and after finishing fills both the checkbox and list Test case 5 User has pressed the correlate button on the Correlation of data has started chromosomes tab or intersection of all genes and after it has finished it will tab or selected genes tab show an image if there are correlations with a certain threshold or an empty image if there aren t any Test case 6 User has selected some properties in the clinical Selections stay in place unless data section user changes Appendix User manual User Manual _HeatMapper Expansion _ Table of Contents Part Apt A A A A AA ds
58. re gains and or losses of DNA sequences occur Single Nucleotide Polymorphisms are DNA sequence variations that occur when a single nucleotide A T C or G in the genome sequence is altered an application developed by the Erasmus MC for displaying heat maps matrix filled with correlation values Pearson s correlation coefficient a value between 1 and 1 representing negative or positive correlation a span of genomic positions genomic position on the human genome entirety of a human s hereditary information a company that manufactures DNA microarrays location on a chromosome code is thread safe when its data remains consistent even if multiple threads manipulate the data an arranged representation of multiple DNA or RNA sequences where areas of common properties are aligned In the case of this project probe data is aligned based on the location of the probe in the human genome a statistical hypothesis test that is used to evaluate whether two sets of measures are essentially different a discrete probabilty distribution that arises when a random draw is made among two distinct types without replacement Gene Ontology KEGG Kyoto Encyclopedia of Genes and Genomes genome browser an application to visualize data concerning genomes and their expression It is a graphical interface to display information read from a genomic data Part 1 Introduction Acute Myeloid Leukemia AML is a form of cancer where immature blood cells do
59. ressPrinter CorrelationController CorrelationType e StatusFeedBack o An abstract class which lets the CorrelationController dispatch status messages to the user interface incomplete e CorrelationController o An abstract class containing all the methods to correlate selected genes and samples e SelectedGenesCorrelationController o The controller for the selected genes tab e ChromosomeCorrelationController o The controller for the chromosomes tab e AllGenesIntersectionCorrelationController o The controller for the all genes intersection tab e CorrelationControllerPropertyChangeListener o A listener for CorrelationController which keeps track of the progress of the calculations e PFileParserPropertyChangeListener o A listener for the PFileParser classes it also keeps track of the progress of the parsing of files e ProgressPrinter o A class dedicated to printing the progress in the console for debugging purposes e CorrelationController CorrelationType o An enum which holds the type for each of the 3 children classes of CorrelationController correlation main parser DMPDualPFileParser DualPFileParser lt T gt GEPDualPFileParser e DualPFileParser lt T gt o Aclass for parsing 2 files at once namely annotation and data files e DMPDualPFileParser o A class for parsing DMP annotation and data files e GEPDualPFilePaser o A class for parsing GEP annotation and data files co
60. rrelation lib correlation lib CorrelationSet PearsonsCorrelationBoxedFloat e CorrelationSet o A static class which contains frequently used methods for calculating set operations on correlation results PearsonsCorrelationBoxedFloat o A class which is a wrapper around the Apache Commons Math library component it has extended the existing class to support arguments for boxed floats correlation components clinicaldata correlation components clinicaldata parser ClinicalDataFileParser ClinicalDataFileParser o A parser for clinical data files correlation components clinicaldata view ClinicalDataSamplePicker e ClinicalDataSamplePicker o The user interface component for selecting samples based on properties in clinical data correlation components clinicaldata model ClinicalData ClinicalDataComboBoxModel ClinicalDataListModel Sample e ClinicalData o A data structure for holding clinical data It also serves as a model for selecting and deselecting certain clinical data attributes e ClinicalDataComboBoxModel o A model for a user interface ComboBox component e ClinicalDataListModel o A model for a user interface List component e Sample o A data structure holding the attribute to value mappings of clinical data correlation components clinicaldata controller ClinicalDataFileParserPropertyChangeListener ClinicalDataFileParserProperty
61. rsionsencecrested Run Circos perl script through external program call in Java Circos image gt Display Circos image Figure 3 Flow chart Calculating correlation part represents the three ways to pick genes in 2 2 2 5 1 Loading and parsing of data To be able to do calculations there has to be a way to take files as an input A file location needs to be identified Once the file location has been identified the file has to be parsed Once the file has been parsed the following part will allow the user to select genes of interest 2 5 2 Selection of genes and calculating correlations Selection of genes is important for convenience two extra ways of selecting genes have been added intersection of all genes and genes within selected chromosomes The first will only calculate correlations coefficients of genes which are contained in both the DMP and GEP files The latter will allow users to correlate genes of entire chromosomes with one another 2 5 3 Displaying the resulting visualization Circos requires a configuration file and data files in a certain format These files will need to be generated accordingly to the selection of genes Once this is done the perl script can be activated given the configuration and data files The perl script will then generate a circular diagram 2 6 Implementation This part will elaborate the implementation decisions made 2 6 1 Model view controll
62. s setting a correlation threshold it is up to the use to decide which genes are interesting 2 4 5 Letting the user select samples of interest Samples represent patients and each patient has different properties The user would like to select samples with certain properties 2 4 6 Displaying data in a circular diagram A circular representation of data gives more information about relations between genes than tabular data An external perl script called Circos Circos will be used as it is an established tool for generating genomic visualizations In the next section the general design will be elaborated with the help of a flow chart 2 5 Design The process of getting a visualization is split up in three distinct parts see Figure 3 e Loading and parsing of input data e Selection and calculating correlations e Displaying the resulting visualization This is also the order of the flow of the application GEP DMP Correlation visualization Optionally fill in correlation threshold this is a positive number the default value is 0 95 Check that the correlation threshold is between within 0 1 where the correlation threshold is a real number g oO me To i e l a User selects file for the annotation or data file selector User selects f
63. stead of memory tables because it turns out after several tests that this program will stop working it deals with java memory heap space problem On the other hand disc based tables will work slower but it can handle bigger data The main programming APIs are SQL and JDBC stands for The Java Database Connectivity the industry standard for database independent connectivity between the Java programming language and a wide range of databases SQL databases and other data sources such as spreadsheets or flat files The JDBC API also provides a call level API for SQL based database access It is possible to use Java programming language to utilize Write Once Run Anywhere capabilities for applications that require access to enterprise data JDBC The JDBC API mainly has three features e Establish connection with a database or access any tabular data source e Send SQL statements e Process the resultsets 3 4 2 Processing data After importing files to a database next step would be comparing which sample numbers belong to two groups that have been chosen by the user This step can be found in class GroupAndAnnoCrossCheck java This class has doCrossCheck method that returns sample numbers that can be found in the annotation file as well as in that groups Applying t test to these sample numbers from two groups is the next step How many times this t test are done would be the same as the total number probesets in the annotation file
64. step by step It starts with choosing a csv file data file as input and displaying a circular diagram in circos as output Steps in between input and output will be elaborated in 3 3 below 3 3 Methodology This methodology sub chapter gives explanations about calculations and methods that need to be applied to the data in order to get end result of this pathway functionality generating an input file for circos 3 3 1 T Test After inserting and importing input file to a database file next step would be to test two groups using t test T test will give the observed significance level or p value that associated with these two groups this two tailed t test compares the means of these groups even if they have different numbers of samples Several p values p pn are obtained from these p tests which after that are thresholded by an alpha which filled in by the user TTest If the user choses to apply multiple test obtained p values from t tests will be corrected using recently developed multiple test methods Suppose that there are 20000 p values which means that there are 20000 separate hypothesis tests If for example a standard p value cut off alpha of 0 05 is being used then 20000 0 05 1000 genes would be expected to be significant by chance 3 3 2 Multiple Testing Hois true Hois false Do not reject Ho Correct decision Incorrect decision 1 a 1 B Reject Ho Incorrect decision Correct decision a B
65. ta GEP _data csv PITT UIT TLL LLL LLL DMP annotation i_viz neworiginaldata DMP_annotaties_hg19 cs Parsing fe DMP data samples jop data_viz neworiginaldata DMP_data_344 csv AAA AAA Parsing complete Clinical data ktop data_viz neworiginaldata dinical_data_GEP_and_DMP csv ITT TTT AAA AAN Click the combo box to select an attribute Select items for that list 119 23 Deselect all neg Reset ia Selectal Selected genes Intersection of all genes Genes within selected chromosomes Delimiter s Correlate Enter genes of GEP data set GEP data set genes DMP data set genes Enter genes of DMP data set PP Select all genes in GEP data set JP Select all genes in DMP data set It is possible to choose an attribute and then select the items that you desire Only the samples matching the selected criteria will be used in the calculations The Deselect all button deselects everything in the list of the currently selected attribute The Select all button selects everything in the list of the currently selected attribute The Reset button deselects everything if nothing is selected all samples will be used Correlation options There are 3 ways to select genes of interest each with a slightly different user interface Selected genes There is a delimiter field two text areas and two lists The delimiters are applied to the input in the two text areas The input is parsed real time and parsed
66. the specific implementations of CircosConfig CircosConfigCorrelation and CircosConfigHypergeo will be explained correlation circos api CircosDataFileWriter ea Rea Genes Karyotypes Links CircosConfig CircosConfigCorrelation CircosConfigHypergeo CircosColor CircosColors Gene Karyotype Figure 13 Circos API code structure 5 2 1 Structure of Circos To be able to generate an image with Circos the following files need to be generated e a Circos configuration file o this file contains settings for the size and height of the image spacings etc file location of the genes karyotype and links file e agenes file o this file contains the genes and its location on a certain chromosome e akaryotype file o this file contains the size name and color of each chromosome e a links file o this file contains source and destination locations where links needs to be drawn On top of that the following files need to be copied to the system e circos 0 56 o Circos script e custom fonts o Circos uses certain custom fonts which need to be included e bundlelinks tool o this is a tool to reduce the number of links in the links file given a minimum link span 5 2 2 Circos API specifications The implementation of the Circos API does not contain the entire set of functionality of the Circos perl package but it is a subset which suffices for this project chrY yo AFERE CATEG
67. tion buttons will allow you to browse to the files you want to import You can also type in the path manually but you will need to use double backslashes After that you can press load The loading process can take a while depending on the size of the input files You can load multiple files at once manipulate already loaded data types and use the clinical searches while you wait Please note that the sample selection and merge functions may not work properly until all data types are loaded Once the data types are loaded the info window will display the total amount of samples loaded and selected as well as some basic info about each loaded data type An unfortunate property of the tool is that the program can be quite memory intensive If you find that you have performance issues when loading in large files it may be required to increase the heap size allocated to java The most reliable way to do this is to create a new system variable called _JAVA_OPTIONS with a body of Xmx500m or Xmx1g This should set the java maximum heap space to 500 megabyte or 1 gigabyte respectively Please ensure that you have enough RAM for this COTO toa his is the info window Add a new data type hame file location select data File annotation file select annotation load Data type options im upperT lowerT set Thresholds delete Select samples merge samples name merge O
68. tion TestPlan ai A oi 45 Appendix User manal sisses enia en ensina esee is 47 Glossary neutriphil dendritic cell gene DNA nucleotide base DNA sequencing DNA microarray DNA spot probe set DNA probe pathway methylation DMP gene expression GEP CNV SNPs SNIPs HeatMapper correlation matrix correlation value Pearson s correlation coefficient genomic range genomic position human genome affymetrix chromosomal location thread safe DNA sequence alignment t test hypergeometric distribution GO Most common type of white blood cells An immune cell a molecular unit of heredity of a living organism is made up of DNA Deoxyribonucleic Acid is the inheritable characteristics of an organism basic building block of DNA determining the order of the nucleotide bases in a DNA molecule a collection of DNA spots attached to a chip a spot on a microarray in which a certain DNA sequence is defined a set of probes of a DNA microarray a labeled segment of DNA used to find a specific sequence of nucleotides in a DNA molecule a set of genes which have a relation with each other the addition of a methyl group CH3 to an existing molecule Methylation in promoter regions of a gene can lead to silencing of that specific gene DNA Methylation Profile the process which uses information in a gene to synthesize a gene product Gene Expression Profile Copy Number Variations are alterations in DNA sequences whe
69. tion COSTAS tea 8 a O O 9 2 4 A E eth dl en E E od al hea hs ae tah E E a 10 2 4 1 Reading Affymetrix annotation TUES c cis cclasteteccsctnessaneuasoesscdeascdashivearssdbsousaeadeouceedaacdetes 10 2 4 2 Reading human sample files containing GEP or DMP data valu8S oooconccnnnnccnoncccnnccnns 10 2 4 3 Letting the user seta correlation threshold a ia di 11 2 4 4 Letting the user select genes OIM RS 11 2 4 5 Letting the user select samples of IM tas 11 2 4 6 Displaying data in a circular DIAM airada ti ceci ita 11 2I DDT O PATA 11 2 5 1 Loading and parsing AAA O 12 2 5 2 Selection of genes and calculating correlatiONS oonoocnnnccnnconionnnonaconnconnonononononnnnonnnnnnn ns 12 2 5 3 Displaying the resulting YiSUAlUIZ MO a 12 PA exec r a a E a ace Maa sidetaosen mene en Aus ttet 13 2 6 1 Model view EN hae ER 13 2S 2 O 13 216 3 BACK di e 14 2031 Parsing th np aa 14 2 6 3 2 Manipulating the input da NA Nealon aaiintiavhd oiteual ans 15 2 6 3 3 Creating a visualization of the input data based on the IMpUt ooooonocccnnnncccnonocccnnnncnnnnns 16 Part 3 Pathways arenei darte li Catiatets 17 Del Probl m AA A E E A E ENE 17 3 2 ntet ctiots and DS A as 17 33 Methodology pee AS A A eaa E e E Ra ia 21 ee A E E OPA o E E N 21 A A A E E OA 22 3 4 pimienta On TAS AE 22 3 4 1 Redding and Importing Ple ido 22 3 42 POCERO 23 Part4 AUN O 25 ARIAS E A IS 25 SA AE EA EE E E S ih tedunGalnaaesdh 25 OP a 26 Interaction ti a
70. using automated unit testing using Junit an Eclipse plugin However the functionality of the gui and the output generation and visualization in IGV have been tested manually The Alignment tool was able to meet most requirements but not all For example the client requested the feature to give different data types different colors in IGV He also asked that the thresholds feature would highlight certain probe sets instead of hide all other probe sets These requests appeared difficult to implement in IGV However all the requirements requested originally where met One problem with the program is its memory usage The loading of large input files can sometimes take up to 10 minutes and the java heap size can get quite large This requires the user to adjust his maximum heap size if he wants to load large files Visualization can also be slow if the user wants to display a large number of samples in IGV Part 7 Conclusion and recommendations This chapter is a summary of the overall results of the project and gives recommendations to improve the created tools Correlation The aim of this part of the expansion was Calculating the correlation between gene expression profile data and DNA methylation profile data and consequently displaying the correlations in a circular diagram given a certain threshold The part of the expansion does exactly this and more During the project new requirements such as filtering by samples came up which
71. utes that needs to be extracted from the first four files are unique identifier of each probe set chromosome in which the probe set is located the starting position in the chromosome the stopping position in the chromosome the gene uid which is measured OY nb a lee mapping of sample name to value So the data structure containing these 6 attributes will be called a row Here follows a definition of a row row uid chr start stop g s where uid chr start stop g and s correspond to 1 2 3 4 5 6 respectively The extraction of the data for the DMP files are quite straightforward the values in the columns of the DMP annotation file correspond to attributes 1 5 and the values in the columns of the DMP data file correspond to attribute 6 The GEP Affymetrix annotation file is less straightforward attributes 2 4 need to be extracted from the column called Alignments The value in this column is formatted as follows chromosome start position stop position strand alignment matching percentage chromosomal location Example chr6 30856165 30867931 95 63 p21 33 This would be an probe set in chromosome 6 with start position 30856165 stop position 30867931 a positive strand a matching percentage of 95 63 and at chromosomal location p21 33 It is also possible that there are multiple alignments these are separated by a triple forward slash I P In that case the data of the al
72. y displaying the correlations in a circular diagram given a certain threshold e Displaying pathways of a gene expression profile or a DNA methylation profile in a circular diagram that significantly related to specific genes e Create a tool to easily align expression methylation SNP and CNV data in a genome browser Appendix Correlation class diagrams Packages correlation components annotationdataselectionpanel view correlation main parser H i A i A AAA 2 r I y correlation main view T T l ae i Y AA 4 z ai z f correlation components clinicaldata view T i i Y Y correlation circos api correlation main controller correlation components clinicaldata controller 1 i l A a AAA s C j i i Y y i correlation main model correlation lib correlation components clinicaldata model correlation components clinicaldata parser Figure 15 The class overview In the pages hereon each class will have a short description correlation main correlation main view e CorrelationView o This is the user interface of the application correlation main model Fai HeatMapResultContainer Coso oO E CorrelationResultContainer GEP DMP PRow GEPRow ChromosomeListModel DMPRow GeneSymbols GenesListModel InvalidElementException e LinksExporter o An interface for correlation value containers to export lin

BScEindverslag - Repository TU Delft

Contents

Download Pdf Manuals

Related Search

Related Contents