Home

GFPE - Computer Science

1. sss 38 3 JMSBICBVALUATION 2 ette diee eati mede e Rs 42 6 CONCLUSIONS AND FUTURE WORK eere 45 6 TC onc US TGT s seele 45 6 2 Fut te Wolke net 46 APPENDICES e oe nt es 47 A Gene Finding Features GFF Definition essere 47 B O 49 C How To Add a New Gene Finding Program Into the GFPE Package 64 REFERENCES tada ai A O e tea 65 CHAPTER 1 INTRODUCTION Computational gene identification plays an important role in genome projects and numerous programs have been developed to address this problem Selecting the best gene finding program or programs for a new organism or category of sequences can be time consuming and error prone as well as problematic for the following reasons 1 The approaches used in gene identification programs are often tuned to one particular organism accuracy for one organism or class of organism does not necessarily translate to accurate predictions for other organisms 2 The performance of the gene finding programs may depend on the parameter settings used to perform the analysis 3 Published evaluations of gene identification programs are often not only limited to a particular organism but may report only a subset of the available metrics This use of different metrics by the authors of different gene finding programs complicates the comparison of results The effort required to reproduce these studies to verify the results or to generate a consiste
2. threshold 80 and type 3 exons for which the COP exceeds the same threshold 2 4 Jian Wang s Work in Evaluation Package The gene finding evaluation programs used in this GFPE graphical user interface tool to predict gene finding accuracy at the coding exon and protein levels are a set of Java classes developed by Jian Wang and using a command line interface The prediction result GFF format converter programs were also written by Jian Wang The data used to draw the bar chart in the tool were averaged using Jian Wang s program The FFG program was developed by Dr Eileen Kraemer Dr Jinhua Guo and Jian Wang at the Department of Computer Science the University of Georgia 14 15 CHAPTER 3 A WALKTHROUGH OF THE USAGE OF THE TOOL Before running the program this software should be installed typically on a Windows based machine In addition a dataset consisting of DNA sequences and their annotation files must be prepared Then the following steps are performed to accomplish the gene finding program evaluation work 3 1 Add Files to the Prediction Files Table The system will automatically open a new experiment when it is started If desired the user can close this new experiment and open a previously saved experiment to continue his work To add a sequence or sequences to the Prediction Files Table the user must select a single table cell or an area in the first column click the add file icon in the tool bar
3. u _ Generate PostScript graphics _ Print GeneMark 2 4 predictions in addition to GeneMark hmm predictions I Translate predicted genes into protein Cancel Defalt OK Fig 3 2d Setup Parameters Window for GeneMark Program 20 5 Select Program and Setup Parameters Window Fig 3 2e Setup Parameters Window for Pombe Program amp Select Program and Setup Parameters Window TER perre Fig 3 2f Setup Parameters Window for FFG Program 21 Gene Finding Program Experiment Kit CER Experiment File Program amp Parameters RunProgram Display Help Oleaje laa a a Ele lolo Bel el Presion fies Test Coding Accuracy Test Exon Accuracy Test Protein Accuracy 8 Annotation file GenScan Program HMMgene Program GeneMark Program Pombe Program FFG Program 2a23_anno off a 4e5 anno gff b8j10 anno gff H123E02_anno gff natvig_anno gf 4 selected tabbed index is 0 selected table row range is 1 1 selected table column range is 2 2 Selected file number is 5 Fig 3 3 A table with five gene finding programs setup 3 3 Run Programs Figure 3 4 presents the data flow through the gene finding evaluation process To run the gene finding programs the user should select an area in the Prediction Files table click the Run Program menu and click on the Run Prediction All five gene
4. Allows the user to operate on data add files to the table save the tables Actually save the data Vectors of each table open a saved file to display the original stored file names on the table read the user selected files and display them on the GUI text window 25 ThreadsRunning Checks on running threads and updates information about the running threads on a popup window every 10 seconds 26 ThreadsStop A stopFlag is set to true to stop the active multithreaded execution in the case that the user does not want to continue running the prediction function The stopFlag is placed near the end of the prediction execution function in the GoColdSpringHarbor GoDenmark GoMIT GoFFG GoGaTech classes before writing the prediction result into a file CHAPTER 5 USER EVALUATION For GUI based software designed for use by scientific researchers the life cycle development relies heavily on user requirements user evaluation and user feedback Accordingly during the development of this tool user evaluation was obtained The users were researchers and graduate students at the University of Georgia Users were asked to use the tool and feedback was obtained GFPE User Friendly Environment version 1 0 Version 1 0 ofthe GUI dealt with the appearance of the menu bar four spreadsheet like tables attached by the tabs of Prediction Files Test Coding Accuracy Test Exon Accuracy and Test Protein Accuracy and a text display wi
5. Go will execute TestExonAccuracy execute args where args is a string that contains the names of the gene annotation file and the prediction result file from the first table If the user selects the fourth table Test Protein Accuracy table Go will execute TestProteinAccuracy execute args where args is a string that contains the names of the gene sequence file annotation file and the prediction result file from the first table GoColdSpringHarbor Runs the Pombe program using multithreading on a remote server at the Zhang Lab Cold Spring Harbor Laboratory GoDenmark Runs the HMMGene program using multithreading on a remote server at the Center for Biological Sequence Analysis Technical University of Denmark GoFFG Runs the FFG program on local drive using multithreading The FFG program is a set of C codes written by Kraemer et al they were compiled into an executable file FFG exe for use in this package GoGaTech Runs the GeneMark program using multithreading on a remote server at the School of Biology Georgia Institute of Technology 14 15 16 17 18 19 20 21 22 40 GoMIT Runs the GenScan program using multithreading on the Burge Laboratory web server at Massachusetts Institute of Technology Gui The start up class of this tool It creates a JFrame and adds a JMenuBar to the top and a JTextArea to the bottom and the rest ofthe JFrame is used for a JTabbedPane on which JInternalFrames
6. Staben 1994 The FFG algorithm begins by identifying possible start and stop sites as well as left 5 donor sites center splice branch sites and right 3 acceptor sites Frame numbers are associated with start and stop sites Any subsequence matching the pattern GTRNGT is identified as a potential left site any subsequence matching the pattern CTRAC is identified as a potential center site and any subsequence matching the pattern Y AG is identified as a potential right site Then the algorithm traverses the list of start sites and builds a list of primitive ORFs Each ORF ends at the first stop site encountered in the same reading frame in the sequence At this point each ORF has one exon Next the algorithm repeatedly traverses the ORF list For each ORF the algorithm examines the last exon in its list and attempts to extend the ORF to include another exon This is possible if a splice site can be found within the exon That is if the exon contains a left 5 donor site and both a center branch site and right 3 acceptor site can be found within an acceptable distance currently set to 300 base pairs If these are located another exon is added to the list for that ORF otherwise the ORF is marked as complete Extension terminates when all ORFs are marked as complete Finally the algorithm deletes ORFs that are less than 300 bp in length an ORF less than 300 bases is not likely to be a gene
7. above procedures and the class GoFFG java and Go java to add it into the GFPE package 10 65 REFERENCES Borodovskky M and McIninch J GeneMark parallel gene recognition for both DNA strands Comput Chem 1993 17 123 133 Burge C and Karlin S Prediction of complete gene structure in human genomic DNA J Mol Biol 1997 268 78 94 Burset M Guigo R Evaluation of gene structure prediction programs Genomics 1996 34 353 367 Chen T and Zhang M Q Pombe a gene finding and exon intron structure prediction system for fission yeast Yeast 1998 14 701 710 Edelmann S E and Staben C A statistical analysis of sequence features within genes from Neurosora crassa Exp Mycol 1994 18 70 81 Java Tutorial http java sun com docs books tutorial Kraemer E Wang J Guo J Hopkins S Arnold J An analysis of gene finding programs for Neurospora crassa Bioinformatics 2001 17 10 901 912 Krogh A Two methods for improving performance of an HMM and their application for gene finding In Proceedings of Fifth International Conference on Intelligent Systems for Molecular Biology AAAI Press Menlo Park CA 1997 pp 179 186 Lukashin A V and Borodovsky M GeneMark hmm new solutions for gene finding Nucleic Acids Research 1998 26 1107 1115 Malacinski G M Freifelder D Essentials of Molecular Biology 3 edition Jones and Bartlett Publishers Inc 2002 325 326 66 11 Sanger Center G
8. java E GoGaTech java E GoMIT java esi Gui java feel Help java HttpUpload java E InternalTableFrame java E ReadFile java amp readme html E RightClickMenu java el RowsAppend java E RowsEdit java E SpreadsheetModel java E TableManager java E ThreadsRunning java E ThreadsStop java E WriteToFile java 2 KB 4 KB 1KB 33 KB 7 KB 38 KB 37 KB 8 KB 22 KB 253 KB 22 KB 2 KB 23KB 8 KB 9 KB 4 KB 9 KB 7 KB 26 KB 2 KB 10 KB 9 KB 5KB 19 KB 4 KB 3KB 14 KB 7KB 25 KB 7 KB 3KB 1KB HTML Document JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File Application JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File HTML Document JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File JAVA File Fig 1 Directories of the software package and java files in the hard drive 2 Layout Of The Main Screen Once the program is started the main screen appears as seen in Figure 2 This initial screen contains the following parts 51 Gene Finding Program Experiment Kit Experiment File Program 8 Parameters Run Program Display Help 30 0 0 ejisjiejejja 14 Prediction Files Te curacy Test P Sequence file Annotation file CD Fig 2 Layout of the main screen A JMenuBar on the top of the screen contains the m
9. sembled from small nuclear ribonucleoprotein particles snRNP that recognize specific sequences on the primary transcript The spliced mature mRNA is transported to the cytoplasm while the lariat excised portion remains in the nucleus and is degraded a Primary transcript 5 splice site je G U i Exon 2 Figure 2 1 The production of mature mRNA by splicing of the primary transcript 2 2 Gene Finding Programs and Their Approaches Computational gene identification plays an important role in genome projects Numerous programs have been developed to address this problem In this thesis a graphical user interface is designed for a program that can evaluate five commonly used gene finding programs GenScan Burge and Karlin 1997 HMMGene Krogh 1997 GeneMark Borodovsky and McIninch 1993 Pombe Chen and Zhang 1998 and Find Fungal Gene FFG developed at the University of Georgia GenScan Burge and Karlin 1997 is a general purpose gene identification program that analyzes genomic DNA sequences from a variety of organisms including human other vertebrates invertebrates and plants For each sequence the program applies a probabilistic model of the gene structure and compositional properties of the genomic DNA for the given organism to determine the most likely gene structure This model includes consensus sequences involved in transcription and translation length distributions and compositional differences GenScan i
10. then select single or multiple files and click Open The files will be added to the sequence file column in the order they are selected Figure 3 1 Similarly the user may choose the second column to add annotation files All four tables will have the same sequence and annotation files in the first two columns regardless of the table through which these files were added To see the content of the added files the user should select them click the display icon in the tool bar and the files will be displayed in the bottom text window 16 Gene Finding Program Experiment Kit l pa x Experiment File Program amp Parameters Run Program Display Help wi e mea e l v JE8 eee Rjal e elle Sequence file Annotation file e Open y 6 al B8 8 Look in sequence files 32223 34e5 359j10 3 H123E02 Dima File Name 223 4e5 b9j10 H123E02 natvi Files of Type All Files he selected tabbed index is 0 he selected table row range is 1 4 he selected table column range is 1 1 Fig 3 1 Add files to the Prediction Files table 3 2 Add Programs and Setup Parameters To specify the gene finding programs to be executed the user may append or insert any of the available gene finding programs to the table The user may also set up or modify the program parameters or delete an unwan
11. AAA pred gff to the end of the field String array predictionResult_suffix to setup the prediction file name suffix add data testCoding_results AAA to the end of the field String array testCodingResult filePath to setup the coding level test output result file path add data testExon_results AAA to the end of the field String array testExonResult_filePath to setup the exon level test output result file path add data testProtein_results AAA to the end of the field String array testProteinResult_filePath to setup the protein level test result file path Refer to the class Go java Code parameter related panel and methods in the class ColumnInsert java ColumnModify java and ColumnAppend java Add AAA Program to the end of the field String array programs to allow user choose the new program code an AAA program parameter panel update the methods MyltemListener saveProgramPramaters and saveProgramDefaultParamaters code the new methods getAAA Parameters and getDefaultAAA Parameters In the class Go java find the method public void runPrediction add a boolean judgement to determine when to run AAA program Refer to the class Go java Modify the file GoAAA java Make it be able to write the prediction result into a file on the local disk and display the file properly in the table For a gene finding program that can be downloaded to the local machine refer to the
12. Figure 5a 5b 5c 5d and 5e The user can select a program from one of five programs in the drop down JComboBox and then setup its parameters Insert Program Insert a user selected program to the left hand side of the selected column in each table Insertion to the left hand side of the first or the second column is not allowed To insert select a column located click Program 55 amp Parameters menu click Insert Program Column Then the user can select a program from the popup window and then setup its parameters Modify Program Change program parameters or using a new program to replace the original selected program in the selected column To modify selected a column to be changed click Program amp Parameters menu click Modify Program Column or click the to be changed column head directly The user re setup the parameters without changing the program in the popup window or select a new program to replace the original program and then setup new program parameters Figure 6 shows a table setup with five selected programs Delete Program Delete a program from each table Pay attention to do this all the data and the column can not be recovered anymore once the deletion is complete To delete click Program amp Parameters menu click Delete Program Column Append Blank Rows Append a number of blank rows to the end of each table To append click Program amp Parameters menu click
13. and a small menu is popped up allowing the user to quickly access the file edit buttons Add Display Cut Copy Paste and Delete The table head ColumnHeadListener class extends MouseAdapter and handles MouseEvents on the user appended or user inserted table column heads JTableHeaders Whenever a left mouse click or right mouse click event happens on those table column headers a ColumnModify object will be created and a Program and parameters Modifying Window is popped up allowing the user to modify the program parameters or change the original program to a new program and set up the related parameters 4 3 Data Operations The primary classes involved in management of data structures are TableManager java and SpreadsheetModel java The class SpreadsheetModel extends DefaultTableModel and implements Serializable It is the backbone of the table data and contains 4 important Vector variables columnNameVector filePath Vector fileNameVector and parameterVector The TableManager class allows the user to operate on data add files to the table save the tables actually save the data Vectors of each table load a saved experiment and display the original stored file names on the table and read the user selected files and display them on the GUI text window 32 Data Storage The columnNamevVector stores column headers the user appended or user inserted program names Initially it has only three columns name is n
14. in the GoColdSpringHarbor GoDenmark GoMIT GoFFG GoGaTech classes just before writing the prediction result into a file The GenScan program on the Burge Laboratory web server at Massachusetts Institute of Technology is run by the GoMIT class using HttpUpload The parameters and file are posted through multipart form data The HMMGene program on a server at the Center for Biological Sequence Analysis Technical University of Denmark is run by the GoDenmark class using HttpUpload Again the parameters and file are posted through multipart form data The GeneMark program on a server at the School of Biology Georgia Institute of Technology is run by the GoGaTech class using URLConnection The parameters and file are posted through name value pairs 36 The Pombe program on a server at the Zhang Lab Cold Spring Harbor Laboratory is run by the GoColdSpringHarbor class using URLConnection The parameters and file are posted through name value pairs The FFG program written in C code and compiled into an executable file FFG exe is run on the local drive by the Java class GoFFG implemented in multiple threads using the Java Runtime exec File file method 4 6 Gene Finding Program Evaluation The primary class involved in running the evaluation function is Go java A set of other major classes in the GFPE directory are provided by Jian Wang and have been compiled into Java class files The Go class is designed to run
15. spreadsheet like table can be created and displayed HttpUpload A helper class to package up files and parameters as a multi part upload bundle to a web server InternalTableFrame Creates JInternalFrames containing two JTables Two TableColumnModels are created for the two JTables One for the row headers and the other for the spreadsheet like table A JViewport is created to hold the row header table It has some accessor methods for the class SpreadsheetModel It also includes three event listeners table cell selection listener column head click listener and right click mouse button listener ReadFile Reads in a file from the designated directory on the hard disk WriteToFile Writes a new file to the designated directory on the hard disk RightClickMenu A small menu allowing the user to quickly access the file edit buttons Add Display Cut Copy Paste and Delete RowsAppend Appends rows to the end of the 4 tables RowsEdit A subclass of Java s JFrame that lets users specify how many rows to be inserted into the current table or delete the selected rows from the table For both insertion and deletion the operation is applied to all four tables 41 23 SpreadsheetModel The backbone of the table data it contains four important Vector variables columnNameVector filePathVector fileNameVector and parameterVector which are used to store column name file path and name file name and parameters 24 TableManager
16. work e Enabling undo and redo function to allow user back and forth between the current and previous status e Enabling print feature allow user to print some results e Reporting failure if unable to setup connection or unable to get the result back when access remote servers to run the gene finding program predictions APPENDIX A Gene Finding Features GFF Definition Sanger Center 2000 Fields are lt seqname gt lt source gt lt feature gt lt start gt lt end gt lt score gt lt strand gt lt frame gt group lt seqname gt The name of the sequence Having an explicit sequence name allows a feature file to be prepared for a data set of multiple sequences Normally the seqname will be the identifier of the sequence in an accompanying fasta format file An alternative is that seqname is the identifier for a sequence in a public database such as an EMBL Genbank DDBJ accession number Which is the case and which file or database to use should be explained in accompanying information lt source gt The source of this feature This field will normally be used to indicate the program making the prediction or if it comes from public database annotation or is experimentally verified etc lt feature gt The feature type name We hope to suggest a standard set of features to facilitate import export comparison etc Of course people are free to define new ones as needed For example Genie splice detectors account for
17. A USER FRIENDLY ENVIRONMENT FOR GENE FINDING PROGRAM EVALUATION GFPE by WEICHENG ZHANG Under the Direction of EILEEN KRAEMER ABSTRACT A graphical user friendly environment for GFPE Gene Finding Program Evaluation written in Java is developed to evaluate the prediction accuracy of gene finding programs GenScan HMMGene GeneMark Pombe and FFG This tool aims to simplify and or automate the process of executing the gene finding programs on sequences of interest and of collecting and analyzing the results The GUI is designed to be similar to a spreadsheet table The user can add cut copy or paste gene sequence files and annotation files to the table add delete or modify the GFPE program and parameters select an area on the table to represent the execution of various gene finding programs on remote servers automatically collect the prediction results and draw bar charts to compare evaluation accuracy at the coding level exon level and protein level It provides a convenient user friendly environment and an efficient file management and execution system thus saving the user substantial time INDEX WORDS Gene Finding Program Evaluation GFPE GenScan GeneMark HMMGene FFG Pombe Graphical User Interface A USER FRIENDLY ENVIRONMENT FOR GENE FINDING PROGRAM EVALUATION GFPE WEICHENG ZHANG BENG Beijing Technology and Business University China 1988 MENG Beijing Technology and Business University China 1993 MS The Uni
18. Append Blank Rows A popup window will ask how many rows to be added input a number then click ok button Insert Blank Rows Insert a number of blank rows before the selected row in each table To insert click Program amp Parameters menu click Insert Blank Rows A popup window will ask how many rows to be inserted input a number then click ok button Delete Rows Delete a number of selected rows from each table Pay attention to do this all the data and the rows can not be recovered anymore once the deletion is complete To delete select the rows to be deleted click Program amp Parameters menu click Delete Rows f Select Program and Setup Parameters Window Select a Program GenScan Program v Organism Vertebrate wv Suboptimal exon cutoff optional 1 00 v Print options Predicted peptides only To have the results mailed to vou enter vour email address optional Cancel Default OK Fig 5a Select Program and Setup Parameters Window for GenScan Program 56 amp Select Program and Setup Parameters Window Select a Program HMM gene Program v Organism Options Human and other vertebrates Predict Signals best prediction C elegans File with annotation optional Cancel meat OK Fig 5b Select Program and Setup Parameters Window for HMMGene Program amp Select Program and Setup Parameters Window TBR Select a Program
19. Experiment Kit m 0 Ix Experiment File Program amp Parameters Run Program Display Help Be s Poelle Elle eee GenScan Program HMMgene Program GeneMark Program Pombe Program FFG Program _4e5_GenScan_pred14 gff 4e5_GeneMark_pred7 gff mE 4e5 FFG predi3gff j10 i i b9j10_GenScan_pred14 gff b9j10_GeneMark_pred gff not evaluated yet V10 FFG pred12 gff 23E02_anno gff H123E02_GenScan_pred H123E02 FFG predii gff vig anno gff The Running Program Display Window HMMgehe Program datatprediction_resultsHMMgene 4e5_HMMgene_predd gff Pombe Program datalprediction_resultsiPombel4e5_Pombe_pred1 gf IHMMgene Program datalprediction_resultsiHMMgenelb9 10_HMMgene_pred9 gff Pombe Program datalprediction_resultsiPombelb9j10_Pombe_pred gff IHMMgene Program datatprediction resultsiHMMgene H123E02 HMMgene pred JGeneMark Program datalprediction resultsiGeneMarkiH123E02 GeneMark pred Pombe Program datatprediction resultsiPombelH1 23E02 Pombe pred gff 4 otal running programs left 7 No file is displayed Fig 3 5 Monitor running program window To evaluate the gene finding results the user can choose to test at three levels the coding level exon level and protein level by going to the appropriate table selecting an area to be tested clicking
20. FF 2000 GFF Gene Finding Features Specifications Document http search cpan org src LDS AcePerl 1 83 docs GFF_Spec html 12 Wang J Kraemer E GFPE Gene finding Program Evaluation Bioinformatics 2003 to appear
21. GeneM ark Program v Species H sapiens v Output Options Email Address required for graphical output or sequences longer than 100000 bp ml _ Generate PostScript graphics Print GeneMark 2 4 predictions in addition to GeneMark hmm predictions I Translate predicted genes into protein Cancel Defalt OK Fig5c Select Program and Setup Parameters Window for GeneMark Program 57 Fig5d Select Program and Setup Parameters Window for Pombe Program amp Select Program and Setup Parameters Window Sle prone Fig 5e Select Program and Setup Parameters Window for FFG Program 58 Gene Finding Program Experiment Kit EE Experiment File Program amp Parameters Run Program Display Help Eem Biber MEMBER Belle Bel el _ Prediction Files Test Coding Accuracy Test Exon Accuracy Test Protein Accuracy 6 Annotation file GenScan Program HMMgene Program GeneMark Program Pombe Program FFG Program 2a23 anno gfr 4e5 anno gff b9j10 anno gff H123E02 anno gff natvig_anno gff 4 selected tabbed index is 0 E selected table row range is 1 1 B selected table column range is 2 2 Selected file number is 5 Fig6 A table setup with five gene finding programs 6 Run Program Menu This menu is designed to assist user run the gene finding programs in the first table Prediction Files table run test coding accu
22. HMMgenel la exon 1 9590 9808 0 926 I besiparse cds 1 Sequence HMMgenel la exon 2 9862 12399 0 880 1 besiparse cds 1 Sequence HMMgenel la lastexon 12584 12771 0 966 0 besiparse cds 1 Sequence HMMgenel la CDS 8831 12771 0 479 bestparse cds 1 Fig 3 6 Display file content through the bottom text window 3 4 Draw Bar Charts To compare the evaluation results the user can view coding exon or protein level accuracy bar charts To do so the user should go to the appropriate evaluation table select an area that contains the files to be compared click on the Display menu and click on Draw Bar Chart A popup window will display the bar chart as seen in Figure 3 7 3 8 and 3 9 The user can then click the tabs in the Coding Level Accuracy Chart or the Exon Level Accuracy Chart to check the different evaluation results amp Coding Level Accuracy Chart HMMgene GenScan GeneMark Pombe Programs Fig 3 7 Evaluation results bar chart coding level accuracy B Exon Level Accuracy Chart HMMgene GenScan GeneMark Pombe Programs Fig 3 8 Evaluation results bar chart exon level accuracy 25 26 2 Protien Level Accuracy Chart Prediction Accuracy 1 00 0 90 0 80 0 70 0 60 0 50 0 40 0 30 0 20 0 10 HMMgene GenScan GeneMark Pombe Programs Fig 3 9 Evaluation results bar chart protein level accuracy 3 5 Summary This chapter describes the operational features and function
23. When several ORFs overlap the longest one is selected and the others are deleted The reverse complement strand is then generated and the process repeated FFG accepts input sequences in FASTA or plain text format and produces output in the GFF format Sanger Center GFF 2000 a sequence annotation format developed with gene finding in mind 10 2 3 Evaluation Methodology In general prediction accuracy can be measured at three levels at the level ofthe coding nucleotide at the level of exonic structure and the level of the predicted protein product At the protein product level the protein encoded by the actual gene is compared with the protein encoded by the predicted gene Measurements of accuracy at the coding level compare predicted coding value with the actual coding value for each nucleotide along the test sequence Figure 2 2 In this widely used approach predictions are divided into four categories True Positive TP nucleotides classified as coding in both actual and predicted True Negative TN nucleotides classified as non coding in both actual and predicted False Positive FP classified as coding in predicted but as non coding in actual False Negative FN classified as non coding in predicted but as coding in actual TN FN TP i FPi TN FN TP i FN TN i Reality Prediction REALITY roding no coding zZ ba sensitivity ss _ o Y TP FN E o TP FP O o Aa uw EU EN TN 28 Sp
24. a data directory that can store gene sequence annotation and prediction files as well as the prediction result evaluation accuracy files and some temporary files and a readme directory that contains the user manual image pages all seen on the left side of Figure 1 To run this tool JDK preferably version 1 4 or higher should be installed on the machine To compile the command javac java should be used in the gui directory and in the GFPE directory separately To run the tool the command java Gui should be used and also the correct class path should be set up in the system environment variables For Unix systems the user may need to set the DISPLAY variable to an appropriate value 50 C3 annotation files E OD prediction results O FFG O GeneMark O GenScan 5 HMMgene DD Pombe 5 sequence files E 5 testCoding results D FFG OD GeneMark GenScan O HMMgene B Pombe C testExon_results DO FFG O GeneMark E GenScan i HMMgene Pombe E C3 testProtein results B FFG O GeneMark O GenScan OD HMMgene 5 Pombe 5 GFPE 5 convert C3 dat D readme a amp about html E AddParametersWarning java E BasicWindowMonitor java fel Column ppend java ColumnDelete java fe ColumnInsert java e ColumnMadify java E DrawBarChart java E DrawBarChartDataProces Tffa exe E FileEdit java s FilesFilter java E Go java E GoColdSpringHarbor java E GoDenmark java E GoFFG
25. a region of DNA and multiple detectors may be available for the same site as shown above lt start gt lt end gt Integers lt start gt must be less than or equal to lt end gt Sequence numbering starts at 1 so these numbers should be between 1 and the length of the relevant sequence inclusive lt score gt A floating point value When there is no score i e for a sensor that just records the possible presence of a signal as splice5 above you must give something by convention 0 lt strand gt One of or should be used when strand is not relevant e g for dinucleotide repeats lt frame gt One of 0 1 2 or 0 indicates that the specified region is in frame i e that its first base corresponds to the first base of a codon 1 indicates that there is one extra base i e that the second base of the region corresponds to the first base of a codon and 2 means that the third base of the region is the first base of a codon If the strand is then the first base of the region is value of lt end gt because the corresponding coding region will run from lt end gt to lt start gt on the reverse strand group An optional string valued field that can be used as a name to group together a set of records Typical uses might be to group the introns and exons in one gene 48 prediction or experimentally verified gene structure or to group multiple regions of match to another sequence such a
26. ality of GFPE s user friendly environment through a walkthrough usage of the tool and is intended to serve as a brief user manual For the detailed user manual refer to Appendix B CHAPTER 4 DESIGN AND IMPLEMENTATION This software package provides a user friendly environment for GFPE The main part is the GUI interface that displays the operation menu tool bar tables and a text window To permit users to perform gene finding program evaluation the GUI allows users to add cut copy paste or delete files on the table save or load an experiment add gene finding programs and set up the related parameters run prediction and evaluation functions and compare the evaluation results using bar charts Figure 4 1 shows the architecture diagram of this software package The major operations of GUI Construction Handling of Mouse Events and User Selection Data Structures and Operations Handling of Sequence Files Gene Finding Programs and Parameters Gene Finding Program Prediction Gene Finding Program Evaluation and Draw Bar Charts were implemented using Java and are described in the following sections of this chapter To enhance the clarity of discussions of code modules we use different type faces to distinguish between classes that are part of the Java distribution and those written as an element of this thesis project The typeface for java sun com Java API packages is bold Java classes developed in
27. ally convert its format and then do the evaluation Using this tool the users will not need to experience the above tedious process and all the work can be done in one place The benefit of this user friendly environment is that it provides a convenient and efficient file management and execution system allows users to add and edit gene sequences on the table setup gene finding programs and parameters simultaneously run multiple programs on the selected gene sequences monitor the running process automatically collect the prediction results from the remote server and write the format converted result files to the local hard disk Also it allows users to save their own work display files to see the results or visually compare the evaluation results by drawing bar charts CHAPTER 2 RELATED WORK 2 1 Gene Terminology Gene The functional and physical unit of heredity passed from parent to offspring through mitosis Genes are pieces of DNA and most genes contain the information for making a specific protein DNA Deoxyribonucleic acid is a double stranded helix of nucleotides that carries the genetic information of a cell It encodes the information for the proteins and is able to self replicate RNA Ribonucleic acid is an information encoded strand of nucleotides similar to DNA but with a slightly different chemical structure There are three main forms of RNA each with a slightly different function mRNA messenger RNA is the mediating tem
28. ameters for inserting a column to the left of the selected column in the 4 tables ColumnModify A subclass of Java s JFrame that contains 5 gene finding programs and related parameters for modifying a column the user can either change the program or re setup the parameters without changing the program DrawBarChart A subclass of Java s JPanel using Graphics class methods drawLine drawstring and fill3Drect to draw the X axis Y axis and bars It can dynamically adjust the size of the window based on the number of bars and make them evenly distributed along the X axis DrawBarChartDataProcess Processes data files based on the selected table and the selected area in the table calculates the weighted average if multiple rows were selected FileEdit Lets the user edit only the sequence file column and annotation file column by selecting an area in the two columns at any one of the 4 tables to 10 11 12 13 39 perform file Cut Copy or Paste in the two columns Delete can be performed in all the columns Go Runs the appropriate prediction or evaluation programs based on the user selected table If the user selects the second table Test Coding Accuracy table Go will execute TestCodingAccuracy execute args where args is a string that contains the names of the gene sequence file annotation file and the prediction result file from the first table If the user selects the third table Test Exon Accuracy table
29. appropriate prediction or evaluation programs based on the user selected table To run evaluation programs one needs to set up the appropriate class path in the system environment variables If the user selects the second table Test Coding Accuracy table Go will execute TestCodingAccuracy execute args where args is a string that contains the names of the gene sequence file annotation file and the prediction result file from the first table If the user selects the third table Test Exon Accuracy table Go will execute TestExonAccuracy execute args where args is a string that contains the names of the gene annotation file and the prediction result file from the first table If the user selects the fourth table Test Protein Accuracy table Go will execute TestProteinAccuracy execute args where args is a string that contains the names of the gene sequence file annotation file and the prediction result file from the first table 37 Whenever a new evaluation file is to be written to the hard drive the Go class will examine whether the new evaluation file s name exists through the checkFileExist RenameltifExist File myFile method If it exists the new evaluation file will be renamed to avoid overwriting 4 7 Draw Bar Charts The primary classes involved in drawing bar charts are DrawBarChartDataProcess java and DrawBarChart java The DrawBarChartDataProcess class processes data files based on the user selected tabl
30. cted Also useful are the notions of Missing Exons ME and Wrong Exons WE ME indicates the proportion of actual exons with no overlap to predicted exons WE indicates the proportion of predicted exons with no overlap to actual exons 13 Kraemer and Wang 2001 developed a method of selecting a threshold for overlap between actual and predicted exons that relies on the notions of overlap sensitivity and overlap specificity and an initial empirical evaluation Overlap sensitivity is the number of nucleotides in the overlapping region between the predicted exon and the actual exons divided by the number of nucleotides in the actual exon Overlap specificity is the number of nucleotides in the overlapping region divided by the number of nucleotides in the predicted exon A Combined Overlap Percentage COP was defined to be OverlapSn OverlapSp 2 COP In reporting the results of their evaluations Kraemer and Wang 2001 define three categories labeled one star two star and three star The one star category includes only the type 1 exons The two star category includes only the type 1 and type 2 exons Both of these categories may be used to evaluate the ability of a program to exactly locate exon and intron boundaries The three star category combines this information with a measure of the ability of a program to correctly predict coding regions and consists of type 1 exons type 2 exons for which the COP exceeds the
31. d to evaluate 12 how well the sequence signals splice sites start codons stop codons are identified Burset and Guigo 1996 We evaluate the accuracy of predictions at the exon level by comparing predicted and actual exons along the test sequence Figure 2 3 wrong correct Missing exon exon exon Fig 2 3 Measures of prediction accuracy at the exon level Burset and Guigo 1996 Although this approach is widely used no unique criterion has been used to consider an exon as correctly predicted The strictest criterion would score an exon prediction as a correct match only if an exact match exists between actual and predicted start and stop locations i e both splicing boundaries are correctly identified Kraemer and Wang 2001 label these as type 1 predictions A looser criterion scores a prediction as correct if a partial match occurs if at least one of the splice sites has been correctly identified Kraemer and Wang 2001 label these as type 2 predictions Finally a predicted exon may be scored as correct if the overlap between actual and predicted exceeds some threshold Kraemer and Wang 2001 label these as type 3 predicted exons The notions of sensitivity and specificity are still applicable in measurements performed at the exon level Sensitivity is the proportion of actual exons in the test sequence that are correctly predicted Specificity is the proportion of predicted exons that are correctly predi
32. dentifies complete intron exon structures ofa gene in genomic DNA is able to predict multiple genes can deal with both partial and complete genes and can predict consistent sets of genes that occur on either or both strands of DNA The GenScan program may be accessed through http genes mit edu GENSCAN html Parameter settings include a choice of organism vertebrate arabidopsis or maize and a suboptimal exon cutoff value 1 0 0 50 0 25 0 10 0 05 0 02 0 01 HMMGene Krogh 1997 is a program for prediction of genes in anonymous DNA designed for prediction of vertebrate and Caenorhabditis elegans genes The program predicts whole genes and can be used on whole cosmids or even longer sequences It can also predict splice sites and start stop codons If some features of a sequence are known such as hits to ESTs proteins or repeat elements these regions can be locked as coding or non coding and then the program will find the best gene structure under these constraints The program is based on a hidden Markov model a probabilistic model of the gene structure HMMGene can also report the n best gene predictions for a sequence This is useful if there are several equally likely gene structures and may even indicate alternative splicing HMMGene takes an input file with one or more DNA sequences in FASTA format It also has a few options for changing the default behavior of the program The output is a prediction of partial or complete gen
33. e index and the user selected area in the table and calculates the weighted average if multiple rows were selected The weighted average is described in detail in the DrawBarChartDataProcess java file A Vector array is used to store the averaged data For example yAxisData_CodingVector is used to hold CC SMC ACP and AC average data in the coding bar chart yAxisData_exonVector is used to hold Sn Sn Sn Sp Spee Sp Sn Sp 2 n Sp 2 Sn Sp 2 ME and WE average data in the exon bar chart The DrawBarChart class uses the Graphics class methods drawLine drawstring and fill3 DRect to draw the X axis Y axis and bars The number of bars to be drawn in the chart depends on the selected columns programs The DrawBarChart class can dynamically adjust the size of the window based on the bar numbers and make them evenly distributed along the X axis 38 4 8 Summary of Classes Implemented This tool has been developed using object oriented design Some of the major classes created are Is 2 Command line GFPE package provided by Jian Wang ColumnAppend subclass of Java s JFrame that contains 5 gene finding programs and related parameters for appending a column to the 4 tables ColumnDelete Delete a selected column A warning window will popup to make sure the user delete or not ColumnInsert A subclass of Java s JFrame that contains 5 gene finding programs and related par
34. e user also can right click mouse button to get fast access to the above menu item function after selecting table cells 18 Gene Finding Program Experiment Kit Ja x Experiment File Program amp Parameters RunProgram Display Help si jx Prediction Files Test DE EJE ajaja Re xJlojlel ju e Sequence file Annotation file 8 Open E3 Look In sequence files 32223 34e5 L3 b9j10 3 H123E02 3 natvi File Name 223 4e5 b9j10 H123E02 natvi Files of Type All Files he selected tabbed index is 0 he selected table row range is 1 4 he selected table column range is 1 1 Fig4 Add files to the Prediction Files Table 5 Program amp Parameters Menu This menu is designed to assist user append or insert gene finding programs or blank rows to the table change program parameters or delete unwanted programs or unwanted rows from the table There are five gene finding programs in this package they are GenScan HMMGene GeneMark Pombe and FFG program Please perform the following functions on the Prediction Files table Once each operation is done all the other three tables will do the same e Append Program Append a user selected program to the end of each table To append click Program amp Parameters menu click Append Program Column a Select Program and Setup Parameters Window will popup as shown in
35. ecificity Sp TP FP TP FN TF TN Fig 2 2 Measures of prediction accuracy at nucleotide level Burset and Guigo 1996 11 Sensitivity is defined as the proportion of coding nucleotides that have been correctly predicted as coding Specificity is the proportion of non coding nucleotides that have been correctly predicted as non coding An issue that arises in evaluating specificity is that the frequency of non coding nucleotides in genomic DNA sequences is much greater than the frequency of coding nucleotides so that TN tends to be much larger than FP with the result of a tendency toward very large non informative values for specificity Thus in much of the literature on gene structure prediction specificity is instead defined to be TP TP FP the proportion of predicted coding nucleotides that are actually coding Other commonly used metrics based on these categories are the Correlation Coefficient CC defined as p TPxTN FNxFP TP EN x TN FP x TP FP x TN FN the Simple Matching Coefficient SMC defined as TP TN TP FN FP TN SMC the Average Conditional Probability ACP defined as ACP 1 TP TP 2 TN TN 4 TP FN TP FP TN FP TN FN and the Appropriate Correction AC defined as AC ACP 0 5 x2 Burset and Guigo 1996 While nucleotide level metrics are often used to evaluate how well the program locates sequence coding regions exonic structure metrics are typically use
36. enu items Experiment File Program amp Parameters Run Program Display and Help The detailed JMenultems in each JMenu are as shown in Figure 3 A JtabbedPane in the middle of the screen contains four tabs Prediction Files Test Coding Accuracy Test Exon Accuracy and Test Protein Accuracy A spreadsheet like table is attached to each tab The table with the Prediction Files tab displays the DNA sequence files annotation files and gene finding program prediction files The table with the Test Coding Accuracy tab displays the DNA sequence files annotation files and gene finding evaluation result files at the coding level The table with the Test Exon Accuracy tab displays the DNA sequence files annotation files and gene finding evaluation result files at the exon level The table with the Test Protein Accuracy tab displays the DNA sequence files annotation files and gene finding evaluation result files at the protein level The first and second columns holds the same DNA sequence files and corresponding annotation files in all four tables A text area on the bottom of the screen is used to display any file files the user selects from the table or program parameters of the user selected columns 52 Fig 3 GFPE Menu and Menu Items 53 3 Experiment Menu New Open a new experiment The new experiment is like the one shown in Figure 1 in which each blank tab
37. er window into one window e Implementing multithreading to improve the speed of running programs on remote servers and to enable monitoring the running process e Drawing bar charts to compare the evaluation results visually GFPE User Friendly Environment Version 2 0 In version 2 0 the additional features described in the previous user evaluation were implemented The focus was on dynamically appending or inserting the columns to the table using multithreading to improve the gene finding program prediction speed and drawing the bar charts After a demonstration of version 2 0 the user suggested the following additions e Enabling undo and redo functions to allow the user to move back and forth between the current and previous status e Enabling a print feature to allow the user to print some results e Reporting failure if unable to setup a connection or unable to get the result back when accessing remote servers to run the gene finding program predictions e Enabling one bar chart to compare the prediction evaluation results of the different sequences predicted by the same program The above suggestions will be considered in the future work 44 CHAPTER 6 CONCLUSION AND FUTURE WORK 6 1 Conclusion In retrospect the development of the present version 2 0 of GFPE user friendly environment was a good hands on learning experience of programming in the Java language using OOD OOP Ease of user interaction was kept in mind t
38. es have the same number of rows and columns same size table cells and same column headings The first table is used to display gene sequence file names annotation file names and gene finding prediction file names The other three tables are used to display gene sequence file names annotation file names and gene finding program evaluation result file names at the coding level exon level and protein level The number of rows in each table initially is set to 40 but the user may append or insert more rows to the table as needed The number of columns in each table is initially set to 2 but again the user can append or insert more columns to each table The maximum number of columns in each table is 100 The class InternalTableFrame creates a JInternalFrame containing two JTables Two TableColumnModels are created for the two JTables one for the row header JTable that displays the row numbers in the red background cells and the other for the spreadsheet like JTable that displays file names in the white background cells A JViewport is created to hold the row header JTable without the JViewport the scrolling for the two JTables would not match 30 The class InternalTableFrame also has some accessor methods for the class SpreadsheetModel It also includes three event listeners table cell selection listener SelectionListener column head click listener ColumnHeaderListener and right click mouse button listener MyMouseListener The JTextA
39. es in the sequences The output specifies the location of all the predicted genes and their coding regions and scores for whole genes as well as exon scores The HMMGene program is available at http www cbs dtu dk services HMMGene Through the web page users may enter sequences select an organism vertebrate or C elegans specify whether or not to predict signals and specify the number of predictions 1 5 to report The GeneMark gene prediction software takes several forms The original GeneMark program Borodovsky and Mclninch 1993 relied on inhomogeneous Markov chain models of both coding and non coding regions based on analysis of known genes and on the Bayes decision making function to predict genes in Escherichia coli DNA sequences and was then retrained for Haemophillus influenzae Mycoplasma genitalium and other organisms GeneMark Genesis developed for analysis of organisms such as Methanococcus jannaschii and Helicobaccter pylori was designed for the situation in which no experimentally studied segments are available for training The GeneMark hmm algorithm Lukashin and Borodovsky 1998 generates a maximum likelihood parse of the DNA sequence into coding and non coding regions and is designed to more precisely locate the exact gene boundaries This program is available through the web page http opal biology gatech edu GeneMark eukhmm cgi Where the species include a choice of H sapiens C elegans D melanogas
40. finding programs can be run in multiple threads When the prediction results are available the system will write the result files to the appropriate prediction_results folder on the hard drive write the file names to the appropriate table cells and update the corresponding table cells in the other three tables using the words not evaluated yet The results taken from remote server are converted into GFF format to allow further gene finding program evaluation 22 Gene Finding Programs x Evaluation Prediction Results Results GenScan a Coding Level AMM Gene Format Exon Draw Bar Converter Level Charts L y Protein Level Input DNA Sequences Multiple Threads GeneMark Annotation Files Time required for the execution of the gene finding programs can vary from a few Fig 3 4 Data flow through GFPE seconds per file to a few minutes per file depending on the gene finding program and the size of the sequence file Thus results are not immediately available and the user may wish to monitor the running process To monitor the running process the user should click on the Run Program menu and click Show Running Programs The system will update the table to reflect the progress of running programs every 10 seconds in the popup window and display All Done when all the running programs are finished Figure 3 5 23 Gene Finding Program
41. from the popup window e Display Files Display the selected file content To display go to appropriate table select an area that contains the files to be checked click Display menu click Display Files then the selected files will be displayed one by one in the bottom text window Draw Bar Chart Draw coding exon or protein level accuracy bar chart based on the user selected evaluation data The data will be averaged by weight if multiple sequence evaluation results are selected in one column User also can compare single sequence prediction evaluation using the same gene finding program To draw the bar chart go to appropriate evaluation table select an area that contains the files to be compared click Display menu click Draw Bar Chart a popup window will display the bar chart see Figure 9 10 and 11 amp Coding Level Accuracy Chart HMMgene GenScan GeneMark Pombe Programs Fig 9 Draw bar cart coding level accuracy B Exon Level Accuracv Chart HMMgene GenScan GeneMark Pombe Programs Fig 10 Draw bar cart exon level accuracy 62 63 2 Protien Level Accuracy Chart Prediction Accuracy 1 00 0 90 0 80 0 70 0 60 0 50 0 40 0 30 0 20 0 10 HMMgene GenScan GeneMark Pombe Programs Fig 11 Draw bar cart protein level accuracy 8 Help Menu This menu is designed to assist user know how to use the GFPE graphical user interface kit and the version of this sof
42. gram 23 anno gff E 5 anno gff j10 anno gff 23E02 anno gff tig_anno gff E 4e5_FFG predi3gh not evaluated yet lj10 FFG predi2gff H123E02_FF6_pred11 g 1 li amp The Running Program Display Window ry f JIHMMgehe Program datatprediction_resultsiHMMgene 4e5_HMMgene_pred9 gff Pombe Program datalprediction_resultsiPombel4e5_Pombe_pred1 gf IHMMgene Program datalprediction_resultsiHMMgenelb9j10_HMMgene_pred9 g Pombe Program datalprediction_resultsiPombelb9j10_Pombe_pred gff _JHMMgene Program datalprediction_resultsiHMMgenelH123E02_HMMgene_pred GeneMark Program datalprediction_resultsiGeneMarkH123E02_GeneMark_pre Pombe Program datalprediction_resultsiPombelH123E02_Pombe_pred gff otal running programs left 7 No file is displayed Fig 8 Monitor running program window 7 Display Menu This menu is designed to assist user check parameter or file information through the bottom text window and compare the prediction accuracy results using the weighted average values shown in bar charts Display Parameters Display the parameters for the selected gene finding program To display go to the first page table select the column click Display menu click Display Parameters the parameters will be shown in the bottom text window Or click the column head check the parameters
43. h table respectively To run go to appropriate table select an area click Run Program menu click Run Evaluation The system will automatically recognize what evaluation program test coding test exon or test protein to execute based on the selected table When the evaluation result is available the system will write the result file to the appropriate folder on the hard drive and update the table cell using the file name Show Running Programs Show the unfinished running programs To show running programs during the run prediction process click Run Program menu click Show Running Programs see Figure 8 The show window will be automatically updated by the latest unfinished running program information every 10 seconds After all the programs finished it shows All done Stop Running Stop the running programs To stop click Run Program menu click Stop Running which will activate the stop points in multiple thread programs block the result files to be written to the file storage folders on the hard drive and to the table cells 61 Gene Finding Program Experiment Kit l jla IX Experiment File Program amp Parameters Run Program Display Help pijem eel arre Delle Bla e f Prediction Files Test Coding Accuracy Test Exon Accuracy Test Protein Accuracy A T Annotation file GenScan Program HMMgene Program GeneMark Program Pombe Program FFG Pro
44. hroughout the development process Using a visualization based software tool for solving problems associated with gene finding research was found to be quite effective A graphical user friendly environment for Gene Finding Program Evaluation GFPE written in Java is developed to evaluate the prediction accuracy of gene finding programs of GenScan HMMGene GeneMark Pombe and FFG program The middle part of the GUI is designed similar to a spreadsheet table The user can add cut copy or paste gene sequence files and annotation files to the table add modify or delete a column in which the selected program and parameters were set up select an area on the table to represent the execution of various gene finding programs on remote servers automatically collect the prediction results and draw bar charts to compare evaluation accuracy at the coding level exon level and protein level All the user s prediction and evaluation work can be saved as a file for future use This tool aims to simplify and or automate the process of executing the runs of the gene finding programs on the sequences of interest and of collecting and analyzing the results saves users substantial time and command line typing work 46 e It also serves as a basic platform for building further enhancements which could be used to solve other Bioinformatics applications 6 2 Future Work As suggested in the version 2 0 user feedback we may do the following future
45. ing programs GenScan HMMGene GeneMark and Pombe are on remote web servers to which our program must upload the gene sequence files and the user selected parameters The upload methods include either setting up a URLConnection to the remote servers and using the name value pairs POST method to transfer the sequence file and the parameters to the remote servers or using the HttpUpload class to set up an HttpURLConnection to the remote server then packing up the file and the paramters into a multi part upload bundle and posting the sequence 35 file and the parameters to the remote servers The transferred data are queued in the servers to be calculated When the prediction result is available InputStreamReader and get nputStream are used to get the result back GenScan HMMGene GeneMark Pombe and FFG programs can be run by the GoMIT GoDenmark GoGaTech GoColdSpringHarbor and GoFFG class respectively These classes are implemented using multithreading The class ThreadsRunning can monitor the running threads by checking if the threads are alive or not and updating the information every 10 seconds on a popup window until all the programs have finished The class ThreadsStop can terminate the running programs A stopFlag is set to true to stop the active multithreaded execution in the case that the user does not want to continue running the prediction function The stopFlag is placed near the end of the prediction execution function
46. le contains only two columns of sequence file and annotation file Load Open a saved experiment The tables will contains all the saved files and related program columns Close Close the experiment Ask Do you want to save the change you made before close Print Not implemented yet Save Save the current experiment Save as Save the current experiment in the other name Exit Exit from the system 4 File Menu This menu is designed to operate on the first and the second column in four tables Delete can work on all the columns Add Add files to a single column at each time If the single column is the first column or the second column the first column or the second column in all the four tables will contain the same sequence files or annotation files even the user add the files only to one of the four tables To add files select appropriate table by clicking the tab using the mouse and then select a single table cell as a starting cell to which the user wants to add files or select an area click File menu click Add a file selection window will popup the user can select one or more files at each time and then click open the files will be added to the column where the selected single cell is or the first column of the selected area in the order they are selected Figure 4 Cut Cut files from the first or the second column or both columns of the table only Because all the files in other column
47. nce I would also like to thank all the people whoever provided their kind help which was available whenever I needed it Last I thank my parents and brothers on the other side of the globe for their constant encouragement and support throughout my life to them I dedicate my further studies TABLE OF CONTENTS Page ACKNOWLEDGEMENTS zn ren een nies V CHAPTER 1 INTRODUCTION essen alles 1 25 RELATED WORK b ii nase 4 2 Gene Terminology AA L 4 2 2 Gene Finding Programs and Their Approaches nen 6 2 3 Evaluation Methodology is ea 10 2 4 Jian Wang s Work in Evaluation Package 13 3 A WALKTHROUGH OF THE USAGE OF THE TOOL 15 3 1 Add Files to the Prediction Files Table 15 3 2 Add Programs and Setup Parameters 16 3 3 R n Programs O aed e LS 21 E rent 24 3 5 UM een erden ei 26 4 DESIGN AND IMPLEMENTATION 000 ccccsssccsssccessccsesscccessccesssccesssccessscees 27 a I GUL Constructions nec ses mu Fas tta s 28 4 2 Handling of Mouse Events and Selection sss 30 O eese ee seen eoa pt Sae ka L 31 4 4 Handling of Sequence Files Gene Finding Programs and Parameters 11 35 ei e bj ers D eO nel 33 vil 4 5 Gene Finding Program Prediction sse 34 4 6 Gene Finding Program Evaluation esee 36 2 7 Draw Bar Charls ass Saale 37 4 8 Summary of Classed Implemented
48. ndow Interaction with the spreadsheet allows users to select an area in the table add files to that area edit the files using cut copy paste or delete run the gene finding programs and evaluate the prediction results by coding accuracy exon accuracy and protein accuracy and display prediction results or evaluation results in the bottom text window Each table has only seven fixed columns whose column header showing Sequence Files Annotation Files GenScan Program HMMGene Program GeneMark Program Pombe Program and FFG Program After an initial demonstration user feedback was sought The interface look was fine and the participants felt that the tool would save users much effort and time compared to independently running those gene finding programs and collecting the prediction results More concrete ideas about the additional features included 43 e Enabling dynamic addition insertion or deletion of programs on the table allowing users to compare the programs that he she wants or to compare the same program with different set of parameters on one table instead of the initial fixed seven columns e Enabling a mouse right click popup menu for user to quickly access the File edit menu such as Add Cut Copy Paste and Delete e Enabling a mouse click popup window on the column header to change the program or program parameters and merging the separate paramet
49. nt set of metrics is typically prohibitive Numerous gene finding programs have been developed over the last decade Despite these limitations existing methods of gene prediction and models of gene structure are still frequently applied to the newly sequenced organisms for which no model or method has yet been tuned Some gene finding programs work well for the gene prediction even they were not specifically developed for the newly sequenced organisms To find the best gene finding program from the numerous existing methods a lots of trivial and tedious repetitive work need to be done Thus it is important to have a rapid and reliable means to assess the accuracy of different gene identification methods and parameter settings when beginning a new genome project or evaluating a new gene identification program Recently members of the Kraemer Lab at the University of Georgia evaluated several commonly used gene prediction programs GenScan HMMGene GenMark Pombe and a self developed program FFG to compare the ability of these programs to accurately predict gene structure for a particular organism Neurospora crassa Kraemer 2001 Executing these programs on the test sequences collating the results of the various programs and calculating statistics were found to be both time consuming and error prone and motivated the need for development of a standard tool to perform such studies In 2002 a Gene finding Program Evaluation GFPE program wri
50. on the Run Program menu and then clicking on Run Evaluation The system will automatically start the evaluation As the evaluation results are available the system will write the result files to the appropriate folder on the hard drive and update table cells with the corresponding file names To display the evaluation results the user can select single or multiple files in an area in the table click the Display menu and click Display Files The evaluation results will be displayed in the bottom text window Figure 3 6 24 Gene Finding Program Experiment Kit Experiment File Program amp Parameters Run Program Display Help elle a aaa FREIE Melle Be e Annotation file HMMgene Program 2a23_anno gff t 2a23_HMMgene_out gff 4e5 anno gff i 4e5_HMMgene_out gff 4e5 GeneMark outgff 4e5_pombe_out gff b9j10 anno gff j10 genscan out b9j10 HMMgene outgff b9j10 GeneMark outgff b9j10_pombe_out gff H123E02_anno gff H123E02_genscan_out gff H123E02 HMMgene outoff H123E02_GeneMark_out gff H123E02_pombe_out g nag annogf _ Inatvig genscan outgf navig HMMgene outgf nag GeneMark outgff navig pombe outgff date Wed May 30 21 02 01 2001 HMMgenel la C elegans model c sim chmm bsmod SEQ Sequence 16820 4 4230 C 4270 G 4252 T 4068 Sequence HMMgenel la firstexon 8831 9383 0 530 1 besiparse cds 1 Sequence
51. plate between DNA and proteins The information from a particular gene is transferred from a strand of DNA by the construction of a complementary strand of RNA through a process known as transcription Next three nucleotide segments of RNA called tRNA transfer RNA which are attached to specific amino acids match up with the template strand of mRNA to order the amino acids correctly These amino acids are then bonded together to form a protein This process called translation occurs in the ribosome which is composed of proteins and the third kind of RNA rRNA ribosomal RNA Protein made of long chains of smaller building blocks called amino acids Amino acids determine the size shape and length of protein molecules They also give protein molecules the ability to coil and uncoil They are the chemical building blocks from which the cells organs and tissues such as like muscle are made Proteins also serve double duty as hormones enzymes and antibodies which help fight off invading germs Exon Any segment of a discontinuous gene the segments of which are separated by introns Intron Introns are sequences of junk DNA found in the middle of gene sequences These sequences are excised before the mRNA is translated into a protein The function of introns is not known Splicing Malacinski and Freifelder 2002 See Figure 2 1 a and b Pre mRNA splic ing occurs in large ribonucleoprotein complexes known as spliceosomes which are as
52. racy in the second table Test Coding Accuracy table run test exon accuracy in the third table Test Exon Accuracy table run test protein accuracy in the fourth table Test Protein Accuracy table The data flow through the GFPE graphical user interface is shown in Figure 7 The Run Prediction should go first since the evaluation needs the prediction data 59 Gene Finding Programs Evaluation Results Coding Level GEPE Exon Draw Bar Level Charts Protein Level Prediction Results Format Input DNA Multiple Converter Sequences Annotation Files Fig 7 Data flow through the GFPE Graphical user interface 60 Run Prediction Run the user selected programs using multiple threads in the first table To run select an area in the table click Run Program menu click Run Prediction When the prediction result is available the system will write the result file to the appropriate prediction results folder on the hard drive see Figure 1 and write the file name to the appropriate table cell in the table update the corresponding table cells in the other three tables using the words of not evaluated yet User can monitor the run prediction process by clicking Show Running Programs Run Evaluation Run the evaluation programs of test coding accuracy test exon accuracy and test protein accuracy by selecting the files to be evaluated in the second the third and the fourt
53. rea The JTextArea is a multi line area that is used to display file content or program parameters in this tool JScrollPane is used to make the JTextArea scrollable The font and font size are set to Serif and Font BOLD 12 respectively 4 2 Handling of Mouse Events and User Selection The primary class involved in the handling of mouse events and user selections is InternalTableFrame java The class InternalTableFrame includes three private event listener classes table cell SelectionListener table cell MyMouseListener and table head ColumnHeadListener The table cell SelectionListener class implements ListSelectionListener and handles table cell selection Whenever a ListSelectionEvent happens this class will assign the new position values of the selected column starting index selected column end index selected row starting index and selected row end index to the variables collndexStart collndexEnd rowIndexStart and rowIndexEnd respectively These values can then be used to perform appropriate operations such as to add or delete files display a file cut copy or paste files run programs and draw bar charts The table cell MyMouseListener class extends MouseAdapter and handles the MouseEvents that happen on any table cells Only the right click button function is 31 implemented in this class Whenever the right mouse button is clicked on a table cell or a selected area in the table a RightClickMenu object will be created
54. s an EST or a protein See below for examples All strings i e values of the lt seqname gt feature or lt group gt fields should be under 256 characters long and should not include whitespace The whole line should be under 32k long A character limit is not very desirable but helps write parsers in some languages The slightly silly 32k limit is to allow plenty of space for comments extra data Fields must be separated by TAB characters t Here are some example records SEQI EMBL atg 103 105 0 9 0 SEQI EMBL exon 103 132 6 42 0 SEQI EMBL splices 172 173 0 SEQ1 netgene spliceS 172 173 0 94 SEQI genie sp5 20 163 182 2 3 SEQI genie sp5 10 168 177 2 1 SEQ2 grail ATG 17 19 2 1 0 49 APPENDIX B User Manual 1 Compiling And Running The Software The software developed in this thesis work may be downloaded from http iubio bio indiana edu 7780 archive 00000645 To begin using the software a user would download the zip file GFPE zip and then unzip the software to a selected directory on the hard drive In our examples this directory is named gui After unzipping the software files and directories will be shown as displayed in Figure 3 1 In the root directory of gui this software package contains the graphical user interface Java files and an executable FFG program seen on the right side of Figure 1 and a GFPE directory that contains the Java files for the gene finding evaluation programs
55. s are created by different gene finding programs based on the sequence file and annotation file in the first and the second column at the same row If any of the first column or second column were cut the files in other columns will be gone To cut select a single file or an area click File menu click Cut Cut files are stored in a file named tempDataStore for future paste use every new Cut will update this file Copy Copy files from the first or the second column or both column of the table only To make a copy select a single file or an area within the first two columns click File menu click Copy Copied files are stored in a file named tempDataStore for future paste use every new Copy will update this file Paste Paste the cut or copied files which are stored in tempDataStore to the first or the second or both column of the table only To make a paste select an equal size area within the first two columns click File menu click Paste For the Prediction Files Table the table cells that are at the same row as the pasted files and whose column number greater than 2 will become blank For other evaluation accuracy tables the corresponding table cells will be written by the words of no pred result yet 54 e Delete Delete the selected files from any of the four tables To make a delete select the files to be deleted click File menu click Delete Th
56. saved by the class WriteToFile The class ColumnAppend is a subclass of Java s JFrame that contains five gene finding programs and related parameter JPanel windows for appending a program to the four tables The class ColumnDelete can delete a selected program column The class ColumnInsert can insert a program to the left of the selected column in the four tables 34 The class ColumnModify allows the user either to change a selected program or modify the parameters without changing the program 4 5 Gene Finding Program Prediction The primary classes involved in running the prediction functions are Go java GoColdSpringHarbor java GoDenmark java GoGaTech java GoMIT java GoFFG java HttpUpload java ThreadsRuning java and ThreadsStop java The Go class is designed to run the appropriate prediction or evaluation program based on the user selected table If the user selects the first table Prediction Files table the Go class will run the appropriate gene finding programs by checking the table column headers of the selected area if the column header is GenScan Program GoMIT will be run using the selected gene sequences Whenever a new prediction file is to be written to the hard drive the Go class will examine whether the new prediction file s name exists through the checkFileExist RenameltifExist File myFile method If it exists the new prediction file will be renamed to avoid overwriting There are five gene find
57. t to delete the selected column the user should click OK to delete it All four tables will have the same programs in the same order Figure 3 3 shows the table with five gene finding programs 18 amp Select Program and Setup Parameters Window Select a Program GenScan Program Y BL2222 222222 22222222222 2222222222 2 22 22202 EEREZEEAXXkiiktkiik HMMgene Program N GeneMark Program Organism Vertebrate Pombe Program Print options Predicted peptides only NA To have the results mailed to you enter your email address optional null Cancel peak OK Fig 32a Select GenScan program from the drop down JcomboBox amp Select Program and Setup Parameters Window Seles zungen SS Organism Vertebrate M Suboptimal exon cutoff optional 1 00 v Print options Predicted peptides only To have the results mailed to vou enter vour email address optional Cancel Der OK Fig 3 2b Setup Parameters Window for GenScan Program 19 f Select Program and Setup Parameters Window Select a Program HMM gene Program Options CI Predict Signals best prediction v Human and other vertebrates C elegans File with annotation optional f Select Program and Setup Parameters Window Select a Program GeneMark Program v Species H sapiens Mi Output Options Email Address required for graphical output or sequences longer than 100000 bp
58. ted program The same program may be added multiple times with different parameters To add a program the user should click on the Program amp Parameters menu and click Append Program A Select Program and Setup Parameters Window will popup as shown in Figure 3 2a There are five gene finding programs in this package they are GenScan HMMGene GeneMark Pombe and FFG program The user can select one of them from the drop down JComboBox and setup its parameters as seen in Figures 3 2b 17 3 2c 3 2d 3 2e and 3 2f If the user clicks Default the parameters will be set to the default values for that program Ifthe user clicks OK the selected program will be appended to the last column of the table To insert a program the user should select a column click Insert Program in the Program amp Parameters menu then select a program and setup its parameters and click OK The selected program will be inserted to the left of the selected column To modify the program parameters the user can directly click the column heading in the Prediction Files table and re setup the parameters from the popup window or do so using the Modify Program choice in the Program 8 Parameters menu To delete a program the user should select that program column and click Delete Program in the Program amp Parameters menu A popup warning window will ask the user Are you sure you wan
59. ter A thaliana C reinhardtii G gallus O sativa Z mays T aestivum H vulgare M musculus the output options include the choices of Generate PostScript graphics Print GeneMark 2 4 predictions in addition to GeneMark hmm predictions Translate predicted genes into protein The Pombe program was developed to find genes and predict exon intron structure in Schizosaccharomyces pombe Chen and Zhang 1998 In developing the program the authors first extracted a training data set from GenBank checked the annotations for accuracy and removed redundancy Execution of the program involves a number of linear discriminant analyses For example one analysis differentiates between sites introns exons and pseudo sites pseudo introns pseudo exons Initiation sites donor sites and acceptor sites are identified Exon and intron predictions are the result of the combination of three linear discriminant functions Other factors considered include oligonucleotide preferences positional triplet preferences and the location of ORFs The results of these intermediate analyses are then combined through dynamic programming to predict gene structure Pombe is freely available for academic use and is available through the web site at http argon cshl org genefinder Pombe pombe htm FFG is a pattern directed program for gene finding in N crassa based on statistical analysis of sequence features with genes from N crassa performed by Edelman and
60. the selected prediction or evaluation files When saving the tables the four Vectors of columnNameVector fileNameVector 33 filePathVector and parameterVector in the SpreadsheetModel are written into a user named file in Objects by ObjectOutputStream When loading an experiment the stored table data are restored by ObjectInputStream four new SpreadsheetModels are created using the restored data and four tables are constructed using a method of createlnternalTabs model in Gui java with the newly obtained SpreadsheetModels The ColumnAppend ColumnDelete ColumnInsert ColumnModify FileEdit Go GoColdSpringHarbor GoDenmark GoGaTech GoMIT RowsAppend and RowsEdit classes will update some or all Vectors of columnNameVector fileNameVector filePathVector and parameterVector if these Objects are called and executed 4 4 Handling of Sequence Files Gene Finding Programs and Parameters The primary classes involved in handling of sequence files gene finding programs and parameters set up are FileEdit java ReadFile java WriteToFile java ColumnAppend java ColumnDelete java ColumnInsert java and ColumnModify java The class FileEdit allows the user to add sequence files and annotation files to the table by selecting an area in the first two columns of any of the four tables and also to cut copy or paste files in the first two columns or to delete files in all the columns The file data can be accessed by the class ReadFile and can be
61. this project are shown in bold italic and variables and methods are in plain italic 28 GUI Construction Handling of Mouse Events and Selection Data Operations Build Data Handling of Sequences Files Gene Finding Programs and Parameters Gene Finding Program Prediction Data Gene Finding Program Evaluation Data Blank box operation Greyrbox object Draw Bar Charts Figure 4 1 The Architecture Diagram of the software package 4 1 GUI Construction The primary classes involved in the GUI interface are Gui java SpreadsheetModel java and InternalTableFrame java The file Gui java is the main class of this tool It creates a JFrame and adds a JMenuBar to the top and a JTextArea to the bottom and a JTabbedPane on which JInternalFrames spreadsheet like tables can be created and displayed The JMenuBar The JMenuBar holds a JMenu of Experiment File Table Editor Run Program Display and Help choices The JMenu contains JMenultems and may 29 also contain JSeparators The JMenu and JMenultems in this tool are shown in Figure 3 3 The JToolBar The JToolBar contains five groups of the fast access JButtons with image icons The JTabbedPane The JTabbedPane holds four InternalTableFrames that extend JInternalFrame They are the backbone of the tables Each InternalTableFrame contains a spreadsheet like table based on its own SpreadsheetModel All the tabl
62. tten in Java was developed that aims to produce a command line tool to support the task of evaluating gene finding programs Wang et al 2003 The evaluation criteria employed in this package are based on those described in Burset et al 1996 This GFPE program saves the evaluators of gene finding programs substantial effort in calculating the prediction accuracy In 2003 a user friendly environment for GFPE was developed to simplify and or automate the above process of executing the gene finding programs on the sequences of interest and of collecting and analyzing the results The GUI is designed to be similar to a spreadsheet table The user can add cut copy or paste the sequence files and annotation files to the table add modify or delete a column in which the selected program and parameters were setup select an area on the table to represent the execution of various gene finding programs on remote servers automatically collect the prediction results from the remote servers and convert them into standard Gene Finding Features GFF Sanger Center GFF 2000 for further accuracy evaluation This tool also can draw bar charts to compare the prediction accuracy at the coding level exon level and protein level To run these gene finding programs in the absence of the tool described in this thesis users would need go to different websites paste gene sequence or attach a sequence file wait for the result to come back save the result manu
63. tware package e User Manual Describe the user manual To use click Help menu click User Manual e About the Kit Describe version developer and the contact information about the GFPE graphical user interface kit 64 APPENDIX C How To Add a New Gene Finding Program Into the GFPE Package Follow the procedures below to add a new gene finding program into the GFPE package step by step suppose the new program name is AAA and it s on a remote server 1 2 3 4 5 Run AAA program on the remote server Write a basic java class file GoAAA java make upload DNA sequence file upload the related parameters to the remote server get the prediction result back and change it into the standard GFF format success Refer to the class GoMIT java and GoDenmark java Setup result file related storage directories output file names and paths On the local hard drive make four AAA folders in the directories of data prediction_results data testCoding_results data testExon_results and data testProtien_results to hold the AAA program prediction result coding level test result exon level test result and protein level test result respectively In the class Go java add AAA Program to the end of the field String array programs add data prediction_results AAA to the end of the field String array predictionResult_filePath to setup the prediction output result file path add
64. ull Sequence file and Annotation file The files in each table are stored in appropriate directories on the hard drive as shown in Figure 3 1 Their paths and names are stored in two different Vectors of a SpreadsheetModel java to serve different purposes One Vector is the filePath Vector that is used to hold the row data Vector of file paths in which the file paths are stored in the order of the column index and the other vector is the fileName Vector that is used to hold the row data vector of file names in which the file names are stored in the order of the column index The files in filePath Vector are used when its associated table is selected and the gene finding programs or evaluation programs are called The files in fileNameVector are used only for displaying the file names in that associated table The parameterVector stores the parameters of the user selected program The parameters are stored in the String format based on the column index they are null at the index position 0 1 and 2 since those places correspond to the row header Sequence file and Annotation file which have no programs The maximum number of parameters that the parameter Vector can store in each table is set to 100 Data operation All the data operations are through the SpreadsheetModel The TableManager class is a major class controlling data operations such as adding files to the table saving tables opening a saved table file or displaying
65. versity of Georgia USA 2001 A Thesis Submitted to the Graduate Faculty of The University of Georgia in Partial Fulfillment of the Requirements for the Degree MASTER OF SCIENCE ATHENS GEORGIA 2003 2003 Weicheng Zhang All Rights Reserved A USER FRIENDLY ENVIRONMENT FOR GENE FINDING PROGRAM EVALUATION GFPE by WEICHENG ZHANG Approved Major Professor Eileen T Kraemer Committee John A Miller Thiab Taha Electronic Version Approved Maureen Grasso Dean of the Graduate School The University of Georgia August 2003 To my family for their love and support iv ACKNOWLEDGEMENTS Special thanks to my major professor Dr Eileen T Kraemer for her open minded guidance direction and patience during the entire course of my Masters study I also thank Drs John Miller and Thiab Taha for serving on my advisory committee and providing me with guidance I am deeply grateful to Dr William Kisaalita my major professor in my Ph D program at Biological amp Agricultural Engineering Department for his kind support of my study of Computer Science Special thanks also is given to Mr Jian Wang for his GFPE command line package which is run by this graphical user interface I would also like to thank my friends and colleagues in the Computer Science Department and Biological amp Agricultural Engineering Department Bo Qian Nan Li and Hongyu Wang for their companionship and for sharing their programming experie

GFPE - Computer Science

Contents

Download Pdf Manuals

Related Search

Related Contents