Home

NetGestalt User Manual

1. 02 Markers for disease free survival B 03_Stage IV vs Stage i 0 Signature genes for Stage IV vs Stage _ 02 Signature proteins for Stage IV vs Sta 04_MSI vs MSS hs on 01 Signature genes for MSI vs MSS b 02 Signature proteins for MSI vs MSS Figure 35 Clinical relevance based on CRC tumor tissue Sub category 01 Clinical relevance snapshot only contains one track Clinical relevance snapshot which shows the summary of clinical relevance including markers for disease free survival markers for overall survival signature genes for MSI vs MSS and signature genes for stage IV vs stage I for all the genes based on multiple datasets 34 Sub category 02 Survival contains two lower level sub categories 01 Markers for overall survival and 02 Markers for disease free survival Each of them contains eleven tracks including 1 eight SCT files recording the correlation signed logp and log2 Hazard Ratio between gene expression and survival overall survival or disease free survival based on Cox regression model and four datasets two tracks for each of the four datasets 2 three summary tracks one continuous track and two binary tracks summarizing the results based on the four datasets by order statistics Wang et al 2013 Sub category 03 Stage IV vs Stage I contains two lower level sub categories 01 Signature genes for Stage IV vs Stage I and 02 Signature proteins for Stage IV vs Stage I 01 Signature
2. Figure 32 Track co visualization in NetGestalt 30 III Portal descriptions There are multiple portals available for NetGestalt Each portal contains both the protein protein interaction network views for a given species along with a chromosome view Currently there are separate portals for the following species Homo sapiens Arabidopsis thaliana Drosophila melanogaster Mus musculus Rattus norvegicus Caenorhabditis elegans Saccharomyces cerevisiae and Danio rerio Each portal comes preloaded with tracks For the default human portal the context 1s generic with preloaded track limited to some example tracks as well as functional tracks from sources like KEGG Reactome GO and DrugBank The rest of the species portals come preloaded with just GO functional tracks Two portals the Human Colorectal Cancer CRC portal and the Human Clinical Proteomics Tumor Analysis Consortium CPTAC portal both come preloaded with tracks containing the experimental results from specific studies The data sources and data processing methods for these portals are described below 1 Generating Views Each portal contains at least one view derived from a protein protein interaction PPI network and one viewed based on chromosomal locations See section 11 2 a for more on chromosome view vs network views Currently the human portals CRC CPTAC and default human portals contain two network views hprd corresponding to the
3. H AXxe w 8 o 0 8 LI Upload Track File TA A Enter Gene Symbols 2 500 5 000 7 500 Browse All Tracks x ollapse all Sabates Bellver et al 2 E Ey NetGestalt Tracks SHTETESISE ea Enna TSeRatio E E 01 Main Data e Sabates Bellver Normal Adenomas tStatistic e Sabates Bellver Normal Adenomas logPsigned Sy Colorectal cancer progression Galamb et al Sabates Bellver et al E Cy TCGA Glioblastoma multiforme U Copy Number J DNA Methylation Gene Expression O Somatic Mutations E E 02 Functional Tracks Y Cell Map E c Gene Ontology Biological Process Cellular Component Molecular Function e Sabates Bellver Normal Adenomas Siggene Sabates Bellver Normal Adenomas 19468 Gene Expression Omnibus GEO zu This track shows a gene expression dataset comparing human colorectal adenoma to normal mucosa with 32 samples in each group The dataset was downloaded from GSE8671 It was generated using the Affymetrix Human Genome U133 Plus 2 0 Array platform by Sabates Bellver et al PMID 18171984 The dataset was preprocessed using Contributed by U HumanCyc the robust multi array average RMA package Probe set identifiers L L KEGG Deacon IDs were mapped to gene symbols based on the mapping provided by p the GEO database Median expression levels from multiple probe sets LJ Net corresponding to the same gene were calculated to represent t
4. a shared profile a Ovarian_Cancer ec 02 Functional Tracks L Cell Map al correlations 1 3 copy number alteration 1 3 molecular subtype Ee mutation zc Gene Ontology Li Biological Process L Cellular Component a Molecular Function L a vs TP53 mutation a BI phosphoproteome ms global La TCGA hypermethylations GJ TCGA mRNA RNA seq a TCGA_mRNA_microarray TCGA_somatic_copy_number_alterdti e TCGA somatic mutations 1 3 Colon Cancer E Cg Ovarian Cancer 1 3 02 Functional Tracks 2 HumanCyc NCI pod Reactome 03 Drug related Tracks 4 3 03 Drug related Tracks Figure 38 Overview of CPTAC and TCGA track types available Tracks derived from data in CPTAC and TCGA for Colorectal breast and ovarian cancers blue box Proteomic tracks from CTPAC are further divided into shared and unshared peptide data sets red boxes Data sets are further divided between profile tracks clinically annotated matrices of the experimental data shown in black box and correlation tracks Correlation tracks are further divided into categories clinical annotations copy number alterations molecular subtypes mutations shown in purple box a Protomic phosphoproteomic and glycoproteomic alterations from CPTAC cohorts Tracks derived from the CPTAC omics data for the colorectal breast and ovarian tumor cohorts are included for all three cancer types red boxes in Figure 37 as follows
5. Main jeli Zoom to a Gene Enrichment Results EL U IL e Search feature MANGIA Y N LAD eite t fh ie aby l n h i MIR Mat shakes vemm do eddy A Vor oret ti BT Nb YA t i 1b TM nud H i 1 i E i E pha le In T Locanon t unt Ot ENES UM LUN MI RET EI i chide 3d 1 v T EMDT u lth ta AE i AA TTTI Tissue TH ieee be Fon pi l nde 48 En Feature Location Sample GSM215083 Value descending colon VM at Please select analysis level EA day eden Mu To p LUE Me f m A I Li Mop hit pa I Le l n n f gene level vj t BE AR r AIN nme n Please select a feature Abreu T E AANA Aid se pe T N P a4 t TW ven KL Wh i l Location E Sn EMIL WR ey o oh y i HI Select a test Perform FDR adjust Yes No Output track s Figure 21 Setting up a statistical test using a categorical annotation feature tumor Location in this example 22 lo 1 000 2 000 3 000 4 000 5 000 6 000 7 000 8 000 9 laa mi nn 1 8 u s cm DEH gt LI ai M X eas aan nnd gt A A 0 2 500 5 000 7 500 r u i i i i u TA o Sabates Bellver Normal Adenomas Ve qat dU HE ERO Fue een I ng Mv 1 t mh Ed y DW x oe TIE ME inwa ee safe EGIT MR 1m pe vaj ortos du pera dm rere H me ptss I hard dreri im u k t mes prem xs b eti 1 a pr
6. drop down box see orange box in Figure 7 and then finally click the Submit button to upload the file to NetGestalt For network analyses based on user uploaded networks the user can use any type of IDs matched with the IDs in the user uploaded data tracks NetGestalt supports five types of track files corresponding to different data types Please see below for file preparation guidelines Examples for the five types of track files can be downloaded from the Upload Track dialog see brown box in Figure 7 Track Vier Abos Cum view PA viam o Aef D 300 11000 1500 2000 2 500 3000 3500 4000 14 500 5 000 5 500 16 000 14 500 1 200 7 500 16000 16 500 9 000 19 500 10 000 10 504 11 00d 11 5007 A O O Sanates Recover Nora Acerca O E Gon nun merr Cd by Ces La perras Cards AO Ep Vias trado mecs peer ri perc det comparing bomes E a 9o we ma mes lt lt oh 37 gr mor ot groso The distent wee derribado hon CSEBATI L ee paran wang de AU mem Domen Cim LU P8910 oc Anas piste by Sda Aor of of PMID 182719941 The alma propr sumo wang Gu AAA Esa e tay settings PALA par ago Peete wt ac ro 106 mets mopped tr gree mba aad ca fhe mery poet add b do CALO de Medica piprase bevels freee eee ligne pete ute rr da t o eee gem rra oh a lt ts spram Me games c4 primm od average deesse keti of a campers c Pe dat Moser Moor amd coing y adds were wt to oad e
7. i i ae TH Location T dd jT Ni xi ji UT d ge MUR i ALP ta Size NI Je pj Please select analysis level Wut Kat UNT M ade zn lu 4 Viewer Wa U u mi jr M nd P vie d yr i Y pa bi qu u pene ah gone level brek Va etude Li ajh dai o n Nie ote Hee I v i i messer RM h T I ve ses mi Please selecta feature RUD UE PL edd add sh Nach Mirae Value adenoma KAT ai I 1 i TERR i Tissue MU MILI hine rum I n i TN KIT I i LE Select a test adenoma vs normal normal vs adenoma t test Wilcoxon rank sum test Perform FDR adjust 8 Yes No Output track s iog p value test statistic Figure 22 Setting up a statistical test using a binary annotation feature adenoma vs normal in this example viii Survival annotation data and CCTs CBTs When users select a categorical annotation feature e g Overall survival etc See section II 2 c 111 Track sample information TSI file users must select whether or not to perform an FDR adjustment see red box in Figure 21 Next users must select which results to output as new SCTs which currently includes the hazard ratio of the survival analysis and or the signed p values logl0 transformed with same sign as hazard ratio for the Wald test waldtest Likelihood ratio test logtest or Score test sctest see blue box Finally the user should click Go green box The cox
8. eC copy number alteration 1 7 molecular subtype J mutation 41 03 spectral count shared CI spectral count unshared Ga TCGA_hypermethylations TCGA mRNA microarray Ga TCGA_somatic_copy_number_alterati HL TCGA somatic mutations E C3 Ovarian Cancer L3 02 Functional Tracks E C 03 Drug related Tracks Figure 40 Overview of CPTAC track colorectral data sets available outlined in red Precursor AUC and spectral count data from liquid chromatography tandem mass spectrometry LC MS MS based shotgun proteomic data on 90 TCGA tumor samples Ovarian and Breast Proteomic profile tracks The ovarian and breast cancer profile tracks are CCTs containing the processed mass spectrometric proteomics data for liquid chromatography tandem mass spectrometry LC MS MS based shotgun proteomic and phosphoproteomic data on 105 TCGA primary solid breast cancer tumor samples generated by the Broad Institute and 122 TCGA primary solid ovarian cancer tumor samples generated by Pacific Northwest National Lab Additionally LC MS MS shotgun proteomic and glycoproteomic data on 84 TCGA primary solid ovarian cancer tumor samples generated by the Johns Hopkins University was also made into profile tracks see red boxes in Figure 39 Data 1s available at the shared and unshared peptide level and reported at the protein level The following steps were used to prepare the data tracks 1 The Protein Reports containing the TRAQ data files were d
9. 15 i Network OXI ANISTON es se ah e p DEED EE 16 un Gene DEIOFIUZALCOM db te 17 bz Gene Set Enrichment 1 pn Ds o cuu tdi AUS ee 17 MEE WII CI MM rm 17 d Statistical analysis oos boo eate e n 18 i Gene level vs Module level tests eeessesseseeseeeseeeeeeeeen nnne nnne nennen nennen 18 ii Continuous annotation data and CCT CBT statistical testing sseeeeesssss 19 iii Categorical annotation data and CCT statistical testin Qi nana eee 20 iv Categorical annotation data and CBT statistical testing ssesesseeeeesssssss 21 v Binary annotation data and CCT statistical testiQg ooooonccnnnccnnnonoccnnnnnnnnnnnnanornnnnnnnnnnnnnos 21 vi Binary annotation data and CBT statistical testing eese 21 vii Survival annotation data and CCTS CBTS eese nnne nnne 23 Es Data transforma 24 15 Value bascd termo carioca 25 p Presence Dased terna e pain 25 lr Aod nk rapera E M ee eee 26 Il ZOOM 0 CONG ca 27 Je Track COMP AMIS ON ps o e A Sih cts eee kak 28 k Mi CO VIS WAZA OM rr PE 29 l Switch between different VLeVY Sana aaa ann nep eee neneve eee 29 MI Portal d SGFIPLIONS iiic nE E Pa 31 MS Generatio VIC DT 31 2 The Colorectal Cancer CRC portal track description eese 31 a Genomic and proteomic alt
10. All clinical feat ures TCGAbiotab release 090413 xlsx b Breast cancer https cptc xfer uis georgetown edu publicData Phase II Data TCGA Breast Cancer TCGA B reast Cancer Metadata BRCA All clinical features TCGAbiotab rl 020314 xlsx c Ovarian cancer https cptc xfer uis georgetown edu publicData Phase II Data TCGA Ovarian Cancer TCGA Ovarian Cancer Metadata OV All clinical features TCGAbiotab CPTAC S020 xlsx 4 Additional annotation sources from the following publications a Colorectal cancer data 1 CPTAC clinical data from Zhang et al 2014 including the proteomic subtypes identified in the study http www nature com nature journal v5 13 n7518 extref nature13438 sl xisx ii Clinical data from TCGA Colorectal study 2012 including the methylation clustering MLHI silencing and MSI status subtypes and hypermutation status 43 http www nature com nature journal v487 n7407 extref nature 1 1252 s3 zip b Breast cancer data 1 Clinical data from 2012 TCGA Breast study including various categories of clustering http www nature com nature ournal v490 n74 1 S extref naturel 1412 s2 z1 c Ovarian cancer data 1 Clinical data from 2012 TCGA Ovarian study http www nature com nature journal v474 n7353 extref nature10166 s2 z1 5 TSI files were additionally filtered out as follows a Ifthe data for a given annotation feature is uniform b Ifthe data is binary or categorical and lack
11. TCGA mRNA RNA seq HC TCGA mRNA microarray 1 0 TCGA somatic copy number altera EZ TCGA somatic mutations 1 3 Colon Cancer E lt Ovarian Cancer e JHU proteome ms global E shared Figure 43 Example of statistical analysis track subfolders in the CPTAC portal tree browser red boxes Statistical association tests were conducted for all CPTAC and TCGA derived profile tracks in the CPTAC portal 45 Expand all Collapse all vs gender EC 01 Main Data a Breast_Cancer E Colon Cancer Ec VU proteome ms global aS precursor_AUC_shared E o precursor AUC unshared a E molecular Qi vs Hypermutated LJ vs_MLH1_silencing J vs_MSI_status on vs Proteomic subtyp ta 7 vs_Transcriptomic_su E XL mutation aC spectral_count_shared i E C3 spectral count unshared EH TCGA hypermethylations EH TCGA mRNA microarray Figure 44 Example of tracks available from the statistical association tests conducted on the precursor AUC unshared peptides CCT profile track purple box Each of the subtrack annotation features for the profile track were tested against each gene in the track Features with at least one significant gene q value 0 05 have SCTs generated for the test statistic p values q values and an SBT of significant genes Each subtrack annotation features falls into one of four categories clinical copy number alteration molecular subtype and mutation In the above examp
12. and a chromosome view bottom After selecting a network view a one dimensional layout of network nodes will be shown right below the menu Figure 3 the ruler at the top of the upper panel Below the ruler bars in different length and thickness represent modules at different hierarchical levels of the network The thick bars correspond to modules in the best partition of the global network Modules at this level can be split into smaller ones represented by thin bars Alternating bar colors green and orange are used to help users distinguish neighboring modules If a chromosome view 1s selected Figure 3 lower panel the thick bars correspond to chromosomes in numbered order from left to right The smaller bars correspond to chromosomal bands The module information can be hidden clicking the grey double arrow button at the right bottom of this region see blue boxes in Figure 3 b Upload a view Users can also upload their own views into NetGestalt by clicking the Upload button below the Select button in the menu see red box in Figure 4 After clicking an Upload View dialog window will pop up In this window a user can click the Choose File button to select a local file and then click the Submit button to upload the view to NetGestalt NetGestalt accepts two types of view files nsm network files or msm gt data matrix files input network and track information combined
13. error message The maximum upload file size is SOMB d Enter gene symbols as a track To enter gene symbols and add them as a new SBT in NetGestalt the users can click the Enter gene symbols button in the Track menu see red box in Figure 8 and then enter or paste gene symbols in the Enter gene symbols dialog The input will be automatically separated into valid gene symbols i e gene symbols included in the current view and invalid ones The total number of valid gene symbols will be shown at the bottom of the window Clicking the GO button entering a track title in the opened dialog and then clicking the Add button will add the valid gene symbols as a new SBT Figure 8 View Y About Current view network view hprd Browse System Tracks i 11 500 2000 2 500 3 000 13 500 4000 4500 5 000 5500 6000 6500 7 000 7 500 8000 8500 90 a Upload Track File A e Enter Gene Symbols 2 500 3 750 5 000 6 250 7 500 8 750 Clear All Tracks Enter gene symbols x Enter or paste gene symbols APC TP53 AOC3 FN1 MUC58 STAB1 ERBB2 ABC5 Valid gene symbols APC TP53 AOC3 FN1 ERBB2 Invalid gene symbols MUC58 STAB1 ABC5 Figure 10 Add gene symbols as a track in NetGestalt 11 4 Visualize Tracks NetGestalt uses different methods to visualize different types of tracks a CCTs NetGestalt visualizes a CCT track with a heat map with colors ranging from bl
14. into a single file which can both be generated by the R package NetSAM http www bioconductor org packages release bioc html NetSA M html The NetSAM package as well as the NetSAM manual can be downloaded in the Upload View dialog window see brown box in Figure 4 After uploading the network the current view will be automatically switched to the uploaded network view which can also be found in the drop down menu under View Select Track Y About Current view network_view hprd LLL E 0 3 000 4 000 5 000 6 000 7 000 8 000 9 0 Choose a file to upload 30 fr PA Choose File No file chosen NetGestalt supports two types of view files e NSM networks in an edge list format can be converted to he nsm format using the NetSAM function in the R package sect please refer to the NetSAM manual for instructions D rise of the function Here is an example nsm file exampleNetwork nsm e MSM data matrices can be converted to the msm format using the MatSAM function in the R package NetSAM Please refer to the NetSAM manual for instructions on the use of the function Here is an example msm file exampleMatrix msm Submit Cancel Figure 6 Upload a view in NetGestalt c Delete a view A view in use cannot be deleted To delete an uploaded network view the user has to switch to another network view first and then click the delete button below the Select button in the menu
15. proteomic and phosphoproteomic breast cancer data generated by the Broad Institute proteomic colorectal cancer data generated by Vanderbilt University proteomic and glycoproteomic ovarian cancer data from Johns Hopkins University and finally proteomic and phosphoproteomic ovarian cancer data from Pacific Northwest National Laboratory 37 Expand all Collapse all Ey 01 Main Data Er ra Breast_Cancer BH 3 BI proteome ms global E Y BI phosphoproteome ms global m TCGA_hypermethylations amp 7 TCGA mRNA RNA seq 1 7 TCGA mRNA microarray a TCGA_somatic_copy_number_altera E CJ TCGA somatic mutations Ee Colon Cancer amp C VU proteome ms global H LJ TCGA hypermethylations 1 TCGA mRNA microarray t TCGA somatic copy number altera E C3 TCGA somatic mutations J Ovarian Cancer H JHU proteome ms global 20 JHU glycoproteome ms global PNNL_proteome_ms_global 2 03 PNNL phosphoproteome ms gipbal EJ TCGA mRNA RN seq gE C3 TCGA mRNA microarray Bi oi 39 Overview of CPTAC track types available Proteomic tracks are outlined in red For the all three cancer types the CPTAC data was obtained from the CPTAC Phase II Data Portal https cptc xfer uis georgetown edu publicData Phase II Data Proteomic data from CPTAC is separated into shared and unshared peptide result sets red boxes in Figure 36 Data sets are further divided into profile and correlation analysis results data sets Co
16. samples generated by the Broad Institute 122 ovarian from Pacific Northwest National Labs and 84 samples from Johns Hopkins University TCGA profile tracks in CPTAC portal The TCGA profile tracks available in the TCGA data portal are also available in the CPTAC portal but are limited to ONLY those samples available in the CPTAC data sets If samples in the CTPAC data sets are not available in the TCGA data sets those samples will be added to the track with NA values for all genes shown as a grey vertical line see Figure 40 In the current version of the CTPAC portal the TCGA data is take from the 08 21 2015 version of the TCGA result sets 40 The following data sets from TCGA were added to the CTPAC portal 1 Somatic mutation data 1 11 Mutation data was obtained from the MutSig analysis results data sets from TCGA e g http gdac broadinstitute org runs analyses 2015 04 02 data COADREAD 20150402 gdac broadinstitute org COADREAD TP MutSigNozzleReport2CV Level 4 2015040200 0 0 tar gz Silent or non somatic mutations are filtered out 111 CBTs are generated for each version of MutSig available version 1 5 2 0 CV and 2CV where mutation present 0 mutation not present 2 mRNA transcriptome microarray data 1 Lowess normalized log2 transformed transcriptome data was obtained from TCGA e g http gdac broadinstitute org runs stddata 2015 04 02 data COADREAD 2 0150402 gdac broadinstitute or
17. under View and select the view they would like to delete Only user uploaded views can be deleted 3 Add tracks NetGestalt provides multiple options for adding tracks to the track viewing area Hovering the mouse over the Track menu see red box in Figure 5 different options for retrieving and adding tracks of interest will be shown in a drop down menu a Browse system tracks A user can click the Browse System Tracks button to browse all the tracks included in the NetGestalt database In the Browse All Tracks dialog all the tracks are organized into a tree structure The user can click the Expand all button see blue box in Figure 5 to expand the tree and select a category of tracks by clicking a leaf node of the tree such as the leaf node Sabates Bellver et al Then a list of tracks contained in the category will be shown in the top right part of the dialog see purple box in Figure 5 When clicking one track in the list such as the Sabates Bellver Normal Adenomas track see black box in Figure 5 detailed information associated with the track will be shown in the bottom right table Finally clicking the button see brown box in Figure 5 will add the track to the track viewing area Track View Y About Current view network_view hprd Browse System Tracks 2 000 3 000 4 000 5 000 6 000 7 000 8 000 EN Search System Tracks mm iii A a
18. 7 1 TCGA A6 2677 01 68 cecum 0 635 0 TCGA A6 2678 01 43 transverse colon 0 437 0 TCGA A6 2679 01 73 ascending colon 0 145 0 TCGA A6 2680 01 72 hepatic flexure 0 465 1 TCGA A6 2681 01 73 cecum 0 882 1 iv Single continuous track SCT file Definition a SCT file sct 1s a tab delimited text file that contains statistical analysis results derived from a data set e g mean median sum variance t statistic p value File format description an SCT file contains at least two columns the first column with gene ids e g gene symbols and the second column with statistic scores data for the genes Up to three statistic scores three data columns can be included in an SCT file Columns must be separated by tab The first row of the file lists the track names We suggest that a track name should contain some descriptions about the corresponding statistic score For fold changes it is recommended to perform a log2 transformation For p values it is recommended to perform a logl0 transformation and add the direction of change to the transformed p values e g signed p values Missing values are represented by NA Duplicated row names or column names are not allowed No special characters for row or column names Example GeneSymbol TrackNamel TrackName2 Genel 2 2 4 8 Gene2 1 3 1 5 Gene3 5 3 7 6 Gene4 1 1 0 7 v Single binary track SBT file Definition a SBT fi
19. AC portal The data sources for the TSI files data sources were as follows 42 l Significantly mutated genes identified in the sig genes txt result set available in the TCGA COADREAD study MutSig2CV TCGA analysis pipeline e g http gdac broadinstitute org runs analyses 2015 04 02 data COADREAD 20150402 gda c broadinstitute org COADREAD TP MutSigNozzleReport2CV Level 4 2015040200 0 0 tar gz 2 Significantly amplified or deleted focal regions identified in the TCGA COADREAD study Focal region level results the significantly amplified or deleted genes are identified from the all lesions conf 99 txt analysis result sets available in from the TCGA Copy Number Gistic 2 0 analysis pipeline e g http gdac broadinstitute org runs analyses 2015 04 02 data COADREAD 20150402 gda c broadinstitute org COADREAD TP CopyNumberLowPass Gistic2 Level 4 2015040200 0 0 tar gz The non thresholded values for each significantly amplified or deleted focal region are reported in the tsi file 3 Selected TCGA clinical data provided by the CTPAC portal a Colorectal cancer https cptc xfer uis georgetown edu publicData Phase II Data TCGA Colorectal Cancer CPTA C TCGA Colorectal Cancer Protocols and Clinical Data COAD All clinical feat ures TCGAbiotab releasel 090413 xlsx and https cptc xfer uis georgetown edu publicData Phase II Data TCGA Colorectal Cancer CPTA C TCGA Colorectal Cancer Protocols and Clinical Data READ
20. CHOSE TO ji rw t I H if i I 3 46e 12 0 773 MINI TUI y i ER o Made P YA d de MT H nd b View p rim vd nt d pi mk prune ye 1 1 Prnt AM du js n 3 p E H V TP T A i vit Mis m 15 i E i l 10 Entries Per Page age lt Di 9 irem E E Hnc PM B be n m Pm gom um e sed vom i i abb am nd nt te ni ide q a T ti e 9 Mp od mpra Pw 9o VOV uM S CR p k i Wr i Original Track LORII EIA i l VEN u wit nomo TE hi T Sabates Bellver Normal Adenomas at l PAN ine eol d o e SAER kihi Uhi b Wind NI Feature Name Size Test Name Method cor pearson Pval adjusted Cutoff yes 0 05 Figure 19 Module level statistical test results p values and test statistics are initially shown in the Significant Modules panel on the left red box When users click on the individual module names in the Significant Modules panel a new SCT will be automatically generated blue box showing the gene level statistical results non fdr adjusted for only those genes in the selected module green box Alternatively if users click Add all significant modules purple box a new composite track is generated orange box where each row represents a hierarchical level in the selected view and significant modules are either colored red or blue depending on the sign of their test statistic ii Continuous annotation data and CCT statistical testing When users select a continuous
21. HPRD human protein protein interaction PPI network http www hprd org and iRef corresponding to the Ref human PPI network http wodaklab org 1RefWeb For the non human portals the protein protein interaction data of different organisms from BioGrid was processed with the NetSAM package to create a PPI view See section IL 2 b for more on NetSAM For the chromosome views the chromosome information for eight organisms was downloaded from BioMart We generated the chromosome view based on the gene positions in each chromosome and the band information The first level in the view contained all genes in all chromosomes and the second level was the genes in each chromosome If the organism contained the band information the bands of each chromosome will be shown from the third level The gene positions in the view were determined by the gene positions in the chromosome 2 The Colorectal Cancer CRC portal track description As shown in Figure 31 the current version of NetGestalt CRC portal contains 1 genomic epigenomic transcriptomic proteomic and clinical data for The Cancer Genome Atlas TCGA CRC cohort 2 mRNA expression data from several Gene Expression Omnibus GEO CRC cohorts that come with survival information 3 data from CRC cell lines including drug response genomic and transcriptomic data from the Cancer Cell Line Encyclopedia CCLE project and shRNA screen data from the Achilles project and 4 functional
22. NetGestalt User Manual December 16 2015 Table of Contents I Introduction 1 2 IT ee el a eae aoe nie Sa a he Sah na el ne ee ie tede netas os i da lb Dod elas o bo nee hie Gas e o De AOS Selecta portal DE SEL AVIE asses rr get b Upload A VIEW ie du dim deii A 5 YO PP E UL UU O E 6 Add ATE ACS EE A Browse svstem IFAJCRS in e Ru LE 6 De SOAR CI EEAC KS at ke dun d I 7 vM Upload track Sn EEE 7 i Composite continuous track CCT flle ooooononnnnonnncnonononocnnnnnnnnnonnnoccnnnnncnnnnonanrcnnnncnnnnnnnos ii Composite binary track CBT file aaa eee nene rene eneve veren eee vezen eee eee iii Track sample information TSD file aaa aaa aaa anen eee enen eee eee eren eneve eee eee eee eee iv Single continuous track SCT HUQ aaa aaa en pere e en neeee nene eee rene eee eee eee v Single binary rack SBT ME sni oe e o i nt Do CO d Enter gene symbols as a track asserere retener poen rr nen Re epu eds operas p rse 11 Visualizo BICIS CENE E ME cM P GN Sisi e da 12 CEE c MCN M qe NM e RM 12 COS DS odes RR 12 DSS AAA O e E Lu ca Moe cte a tcuct 12 Zoom dn 0 Ut TEAC KS nn a a a a a a Click bars representing predefined modules eee eee 14 DP Vingiig pec 14 c Double Cl6E uoo e nt de T 14 Oc DM ours te daas E udine ed ner IEEE LI LI LE IE EU E EE 14 Analyze tracks A dis Network anal o eee O 15 D Module enrichment ooo rd poaae ono pe aa dm
23. S L 09 Correlation between SCNA and protein 4 3 02 Clinical relevance Figure 34 Genomic and proteomic alteration based on CRC tumor tissue Sub category 01 Omics snapshot only contains one track Omics snapshot which shows the summary of alterations somatic mutation somatic copy number alterations epigenetic alterations differential expression at mRNA level and protein level for all the genes based on data from the TCGA CRC tumor cohort Sub category 02 Somatic mutation contains four tracks including one CBT file recording the binary mutation matrix two SBT files recording the significantly mutated genes and genes mutated in at least 5 of all the CRC samples and one SCT file recording mutation counts in log scale i e log2 mutation count 1 Sub category 03 Somatic copy number alteration SCNA contains four tracks including two CCT files recording the gene level SCNA matrix and focal SCNA matrix two SBT files recording genes in the focal amplification regions and focal deletion regions Sub category 04 Epigenetic silencing contains two tracks including one CCT file recording the methylation matrix and one SBT file recording the candidate epigenetically silenced genes Sub category 05 Differential mRNA expression contains six tracks including one CCT file recording the gene expression matrix three SCT files recording the t statistic values signed logP p values were calculated based on t test and log2 fold change
24. a small region containing the selected symbol and use a vertical red line to highlight the gene in all tracks in the track viewing area Figure 28 27 el A 3000 14000 5000 O A 18000 9000 10000 11000 E i NFEZL1 TEF PATZ NFEA 3 mi MAFB bet Ai 3 E i Figure 30 Zoom to a gene in NetGestalt j Track comparison NetGestalt uses an interactive Venn diagram to help users compare different SBTs The user can first select the binary tracks from the Track Comparison section located in the left panel of the page see brown box in Figure 29 An SBT automatically appears in this section when it is added to the track viewing area and will be removed from the section after it is deleted At most three tracks can be compared at the same time As soon as tracks are selected a clickable Venn diagram will be shown below the track names Figure 29 To help users 28 distinguish different binary tracks easily NetGestalt uses the same color for the selected track name in the Track Comparison section circle in the Venn diagram and upper and lower borders of the binary track visualized in the viewing area Clicking each part of the Venn diagram will add a new binary track to the track viewing area For example the user can click the overlapping part of the Venn diagram and add a new binary track User can also click the blank region of the Venn diagram to add the union of the genes Gene
25. ased filtering For SBTs users can use the Presence based filtering feature to focus on genes present in the current visible range By clicking the Presence based Filtering button Figure 24 all genes present in the current visible range from the SBT will be identified and data for these genes from all tracks in the track viewing area will be displayed in a new webpage Figure 25 This feature is only activated when the number of genes in the current visible range from the SBT is between 2 to 100 29 1000 Jace ET 1900 5900 T Jess 17000 8000 5 500 cece 11009 TI a gt e o e gt o ki ea c tt h sa a c e e o 2 DAA UE YT ua D Network Analysis 9 Qese Set Ennchemen RR a o RS Apo wi tih va ET nt T ape qi Fo Sabaot Beaver Pn D Track information P aa Um a u DNI P m Le PRO vi Delete r v e v4 M Li OR Preeren i mo ROWSE TE iE x da tt ho x Er Sat Os Y 4 yt 1 ARTE ge n f Figure 26 Presence based filtering in NetGestalt Shine zampe heating Hida track fe F T T F BRET Sl n SRFX Al RHTIBJ RADA A EET Sabates Delhoe Normal Ad n ma s logPalgned Figure 27 Output for presence based filtering in NetGestalt h Node link Graph NetGestalt also provides the traditional 2D network visualizati
26. ata Transform Hs Subtrack Annotation Lakos gh ete T BA AAP led Md P nio ib VUE MES oe e dd AIN NO P IE DAR EA o E pash hie m Sabates Bellver_Normal_Adenomas_ logPsigned MX a igs if genti AMI JU UG gl ka Ly sl Mil Ald hi A dii Mba ne ty lur vo Jy Md VW UL Sabates Belver Normal Adehomas Sigaene M I TCGA GBM Mutation Combined v t I Figure 11 Visualize four types of tracks in NetGestalt The top track is an example of a Composite Continuous Track CCT showing continuous data per gene x axis per sample y axis green box Below is a Single Composite Track SCT showing one continuous value y axis per gene axis purple box Next a Single Binary Track SBT shows binary data per gene via the presence or absence of a vertical black black box Finally a Composite Binary Track CBT shows binary data per gene x axis per sample y axis where a true value is presented by a red dot and false is gray yellow box After adding the tracks into the track viewing area NetGestalt automatically shows the track name on the top left of the track see red box in Figure 9 User can click the double arrow button located above the top left of the first track to hide these track names see blue box in Figure 9 Dragging a track name can change the vertical position of the track not supported by IE Hovering the mouse over a track name all track manipulation and a
27. cer Cell Line Encyclopedia study Category 05 Somatic copy number alteration contains one track CCLE CRC CNA AffySNP6 which shows the copy number alteration matrix of CRC cell lines from Cancer Cell Line Encyclopedia study 02 Functional Tracks Cell Map Os Figure 37 Functional tracks d Functional tracks 3 All the functional tracks are included in the Category 02 Functional Tracks red box in Figure 35 These tracks are from six databases Cancer Cell Map 10 pathways Gene Ontology 808 Biological Processes 231 Cellular Components and 383 Molecular Functions HumanCyc 267 pathways KEGG 200 pathways NCI 223 pathways and Reactome 1108 pathways 36 3 Clinical Proteomic Tumor Analysis Consortium CPTAC Portal As shown in Figure 36 the current version of NetGestalt CPTAC portal contains 1 proteomic genomic epigenomic transcriptomic and clinical data for the colorectal breast and ovarian cohorts of both The Cancer Genome Atlas ICGA and CPTAC studies 2 statistical correlation analysis results and 3 functional tracks including Cell Map pathways GO Biological Processes GO Cellular Components and GO Molecular Functions HumanCyc pathways KEGG pathways NCI pathways and Reactome pathways green box in Figure 36 Expand all Collapse all Expand all Collapse all 01 Main Data zw 01 Main Data 2 3 Breast_Cancer H Breast_Cancer Ec Bl proteome ms global O Colon_Cancer
28. d be uploaded together with the CCT CBT file File format description a TSI file ts1 1s a data matrix in which each row represents a sample and each column represents one feature of the samples Sample features can be divided into four data types binary data BIN e g mutation status categorical data CAT e g tumor stage continuous data CON e g age and survival data SUR e g overall survival Binary data do not have to be 0 1 but must contain exactly two categories e g yes no or tumor normal The first column lists the sample names Sample names must match exactly those in the corresponding CCT or CBT The first row lists the feature names and the second row indicates the data type for each feature must be one of the following BIN CAT CON or SUR Each cell in the matrix is a value for corresponding sample and feature For survival data time and event are separated by Missing values are represented by NA Duplicated row names or column names are not allowed No special characters for row or column names Example Barcode Age Anatomic neoplasm Colonpolyps Overall survival data type CON CAT BIN SUR TCGA A6 2670 01 45 sigmoid colon 0 259 0 TCGA A6 2671 01 85 sigmoid colon 0 437 0 TCGA A6 2672 01 82 transverse colon 0 1321 1 TCGA A6 2674 01 71 sigmoid colon 0 NA NA TCGA A6 2675 01 78 sigmoid colon 0 434 0 TCGA A6 2676 01 75 cecum 0 143
29. e the tracks a Network analysis 1 Module enrichment For SBTs or SCTs NetGestalt can help users identify which modules represented by bars with different length are significantly correlated with the tracks For an SBT NetGestalt uses the Fisher s exact test to identify enriched modules identified from the active network Enrichment p values are corrected for multiple comparisons by calculating the False Discovery Rates FDRs Hovering the mouse over a track name several buttons for track analysis will be shown in a drop down menu Figure 12 Clicking the Network Analysis button red box in Figure 12 several options for network analysis will be shown Choosing the Module enrichment and clicking Go blue box in Figure 12 the enriched network modules will be listed in a table in the Enrichment Results section located on the left panel of the page red box in Figure 13 The user can click column title to sort the results by the corresponding column The user can navigate through the pages by clicking the buttons right below the table or select the number of entries per page by clicking the drop down menu at the bottom of the section see purple box in Figure 13 Clicking one entry in the table will add overlapping genes between the enriched module and the original binary track as a new binary track see green box in Figure 13 named by the user The user can also add all enriched modules in a new composite track b
30. eins comparing MSI with MSS CRC samples based on Wilcoxon test and protein data from Zhang et al 2014 4 3 01 Primary Tumor Tissue m 02 Predictors of drug sensitivity 03 Essential genes 04 Gene Expression L 05 Somatic copy number alteration Figure 36 Clinical relevance based on CRC tumor tissue 35 c Tracks based on CRC cell lines All the data tracks derived from CRC cell lines are included in the Category 02 Cell Lines red box in Figure 34 These tracks are divided into five categories 01 Drug response snapshot 02 Predictors of drug sensitivity 03 Essential genes 04 Gene Expression and 05 Somatic copy number alteration Category 01 Drug response snapshot only contains one track Drug response snapshot which shows the Spearman s correlation coefficient between response activity area of the 24 compounds and mRNA expression of all the genes Category 02 Predictors of drug sensitivity contains 24 tracks recording the Spearman s correlation coefficient between response activity area of 24 compounds and mRNA expression of all the genes separately Category 03 Essential genes contains one track Achilles RNAi CRC Cellline specific Essential Genes which shows the CRC cell line specific essential genes from Project Achilles Cheung et al 2011 Category 04 Gene Expression contains one track CCLE CRC ExpGene AffyUl33 2 which shows the gene expression matrix of CRC cell lines from Can
31. enpwotrely end ety vola hroms Deg fus RN e Pt veg vaha Cay pa MAK Syri harris Mp evo met s n a pr pa rr wee o gi med t Org A pui Da par Figure 8 Search tracks in NetGestalt View Y About Current view network_view hprd Browse System Tracks 2000 amp 3000 4000 5 000 16 000 7 000 8 000 90 Search System Tracks Upload Track File Enter Gene Symbols Clear All Tracks Upload Track x E Choose a track file Choose File No file chosen ID type hgnc_symbol v NetGestalt supports four types of track files Please refer to NetGestalt manual for instructions on how to prepare track files Here are some examples for each type of track files SBT Sabates Bellver Normal Adenomas Siggene sbt SCT Sabates Bellver Normal Adenomas statistic sct CBT TCGA GBM Mutation Combined cbt CCT Sabates Bellver Normal Adenomas cct TSI Sabates Bellver Normal Adenomas tsi TSI is an optional annotation file for corresponding CCT Submit Cancel Figure 9 Upload tracks in NetGestalt i Composite continuous track CCT file Definition a CCT file cct is a tab delimited text file that contains data for a composite track with multiple related sub tracks with continuous data e g microarray gene expression data for samples from the same data set The file name will be used as the track name File format de
32. er alteration data was obtained the from the TCGA Gistic 2 0 SCNA pipeline e g http gdac broadinstitute org runs analyses 2015 04 02 data COADREAD 20150402 gdac broadinstitute org COADREAD TP CopyNumberLowPass Gistic2 Level 4 2015040200 0 0 tar gz Non thresholded gene level Gistic scores were used for CCT creation 41 OO O 15 2000 12500 3 00 0 1250 2 500 c epee coo we ees 000000 ee 0 TCGA BRCA _Methylation_HumanMethylation450 e 4 q Samples in CPTAC breast cancer tracks not available in TCGA methylation data added to track as all NA values TCGA_BRCA ExpGene AgilentG4502A 07 3 BRCA normal Figure 42 Example of two TCGA derived profile tracks in the CPTAC portal The top track of methylation data contains a large number of blank sample entries greyed out vertical lines corresponding to the CPTAC breast cancer samples that are not available in TTCGA s methylation data The TCGA mRNA transcriptomic data bottom track has data for all CTPAC samples Because sample information is limited to the samples used in the CPTAC study any samples in the TCGA data that are not also in the corresponding CPTAC data are excluded from the tracks in this portal b CPTAC portal tsi track data sources TSI files containing clinical annotation data and other significant findings significantly mutation genes and somatic copy number alterations from publications are generated for each track in the CPT
33. erations in the TCGA CRC tumor cohort 32 b Clinical relevance based on CRC tumor tissue oooooooooncnncccnccnncnnnnnnnnnononancnonnnnnnnnnnnnnnnnnnnnnnnnnnnanos 34 c Tracks based on CRC cell liDes ooooooocccnnnncncnnnnoccnnnnnnnonononoccnnnnnnnonnnorcnnnnononnnnnornnnnnnnnnnnnnos 36 d E unctonaltrackSuca ito 36 3 Clinical Proteomic Tumor Analysis Consortium CPTAC Portal sess 37 a Protomic phosphoproteomic and glycoproteomic alterations from CPTAC cohorts 37 Colorectal proteomic profile tracks unatinn ada ld 38 Ovarian and Breast Proteomic profile tracks nana aaa aaa aaa ne eee nenen nene eee 39 b CPTAC portal tsi track data sources oio io dat oi Et date ndim dia te vika d r 42 I Introduction NetGestalt is a novel data integration framework that allows simultaneous presentation of large scale experimental and annotation data from many sources in the context of biological networks or genomes to facilitate data interpretation and hypothesis generation The NetGestalt framework provides various features for data query upload visualization and integration This manual introduces all features that can be accessed through the user interface Section II as well as several portals developed based on the NetGestalt framework Section III II User Interface 1 Select a portal There are multiple portals available for NetGestalt Each p
34. fewer genes and all direct neighbors of these genes in the network can be retrieved and shown in a new SBT 16 together with the seed genes in the original SBT The enriched neighbors option works for any SBTs Specifically for each non seed gene in the network all direct neighbors of the gene are retrieved and evaluated for the enrichment of the seed genes using the Fisher s exact test All non seed genes significantly enriched with seed neighbors according to a user defined FDR are identified and shown in a new SBT together with the seed genes in the original SBT iii Gene prioritization To prioritize genes in an SBT NetGestalt provides a Gene prioritization feature under Network Analysis brown box in Figure 12 Specifically for each seed gene in the selected SBT all direct neighbors of the gene are retrieved and evaluated for the enrichment of other seed genes using the Fisher s exact test Seed genes significantly enriched with other seed neighbors according to a user defined FDR are identified and shown in a new SBT b Gene Set Enrichment NetGestalt compiles information from Gene Ontology GO and other five pathway databases including cell map human cyc kegg nci pid and reactome to help users identify GO terms or pathways that are significantly correlated with an SCT or SBT Figure 14 shows an enrichment result red box for a binary track based on GO biological process BP database Enrichment analyse
35. g B et al 2014 Proteogenomic characterization of human colon and rectal cancer Nature In press 47
36. g COADREAD Merge transcriptome agilen tg4502a 07 3 unc edu Level 3 unc lowess normalization gene level data Level 3 2015040200 0 0 tar gz 3 RNAseq data 1 11 Normalized RSEM gene level RNAseq data was obtained from the HiSeq rnaseq v2 normalized gene level data sets e g http gdac broadinstitute org runs stddata_ 2015 04 02 data COADREAD 2 0150402 gdac broadinstitute org COADREAD Merge rnaseqv2__illuminahi seq rnaseqv2 unc edu Level 3 RSEM genes normalized data Level_ 3 2015040200 0 0 tar gz The normalized count values provided by TCGA are log2 value 1 transformed and generated as a CCT 4 Hypermethylation data 1 11 Methylation data was obtained from the Human Methylation 450 TCGA data set e g http gdac broadinstitute org runs stddata 2015 04 02 data COADREAD 2 0150402 gdac broadinstitute org COADREAD Merge methylation human methylation450 jhu usc edu Level 3 within bioassay data set functio n data Level 3 2015040200 0 0 tar gz For genes with multiple probes the probe set found to be most anti correlated with the HiSeq rnaseq v2 normalized gene level data sets data is used when generating the CCTs when paired rnaseq data is available Only solid tumor patients are used for correlations If not available take the mean of the probe sets For correlation ties use first tied probe listed in file 5 Somatic Copy Number Alterations SCNA data 1 11 Somatic copy numb
37. genes for Stage IV vs Stage I contains fifteen tracks including 1 twelve SCT files recording the differential expression t statistic signed logp and log2 Fold Change of genes comparing Stage IV with Stage I CRC samples based on t test and four datasets three tracks for each of the four datasets 2 three summary tracks one continuous track and two binary tracks summarizing the results based on the four datasets by order statistics 02 Signature proteins for Stage IV vs Stage I contains only three tracks recording the differential expression W Statistic signed logp and log2 Fold Change of proteins comparing Stage IV with Stage I CRC samples based on Wilcoxon test and protein data from Zhang et al 2014 Sub category 04 MSI vs MSS contains two lower level sub categories 01 Signature genes for MSI vs MSS and 02 Signature proteins for MSI vs MSS 01 Signature genes for MSI vs MSS contains fifteen tracks including 1 twelve SCT files recording the differential expression t statistic signed logp and log2 Fold Change of genes comparing MSI with MSS CRC samples based on t test and four datasets three tracks for each of the four datasets 2 three summary tracks one continuous track and two binary tracks summarizing the results based on the four datasets by order statistics 02 Signature proteins for MSI vs MSS contains only three tracks recording the differential expression W statistic signed logp and log2 Fold Change of prot
38. he aov or kruskal test R functions are used to calculate the results o 1 000 2 000 3 000 4 000 5 000 6 000 7 000 8 000 9 Es Feature Size x Sample GSM215097 Value 1 5 A 0 2 500 5 000 7 500 o Sabates Bellver Normal Adenomas M I I i e b t ti y Pron al i mn weis am dra aer mb iL n su T i 8 sh iT om r h k 4 Location SAPRA hi 1 Ba i LI _ Please select analysis level i i I f Tissue i I T KJA i N j V REIG Id gene level id B M i b ob eins yv t nbn i ios d LA Please select a feature 4 i l DIT b th eee Tie a d lin iii pn diji eo rar rn or so et rr 7 usa whe sed trp nd dd mase a DU ae winds RETI E TIT E mI E Size IRR l if H Selecta test vel AY ui correlation test Select a method pearson spearman Perform FDR adjust Yes No Output track s Y log p value test statistic Figure 20 Setting up a statistical test using a continuous annotation feature tumor Size in this Go 20 example v Categorical annotation data and CBT statistical testing When users select a categorical annotation feature e g tumor location etc See section 2 c 111 Track sample information TSI file in a CBT the statistical test used depends on whether or not the test 1s to be conducted as the gene level in which case the CBT data will be binary or at the module level in which case the module level CBT data will be a continuou
39. he gene i 7 Reactome expression level For heat map visualization we performed gene wise d normalization on the expression matrices by subtraction of the average EI 03 Drug related Tracks expression level of all samples in the dataset Moreover floor and ceiling DrugBank values were set to 1 and 1 respectively and any value lower higher than the floor ceiling value was set to the floor ceiling value Genetype HGNC Symbol 7 ES T Figure 7 Browse tracks in NetGestalt b Search tracks A user can click the Search System Tracks button below the Browse System Tracks button to search for tracks of interest in the database After clicking the Search System Tracks button a Search system tracks dialog will open The user can input a key word in the box see red box in Figure 6 NetGestalt will list the names of all matching tracks below the box Hovering the mouse over the 1 button of a track will display the detailed information about the track Finally the track can be added to the track viewing area by clicking the button c Upload tracks Users can upload their own data into NetGestalt by clicking the Upload Track File button below the Search System Tracks button see red box in Figure 7 In the Upload Track dialog the user can click the Choose File button to select a local file see blue box in Figure 7 then select the type of gene identifier from the ID type
40. igure 10 A track of gene symbols will appear at the top of the track viewing area when 1t is completely zoomed in see red box in Figure 10 To zoom out the user can click the top green bar at root level see blue box in Figure 10 E E A 1909090 110 000 EE 1 un Sabates Bellver Normal Adenomas vi E ZO gt M a EM S LI s E ji mm Sabates Bellver Normal Adenomas Siggene Figure 12 Click bars representing predefined modules to zoom in the tracks If a user is interested in a region that is not represented by a predefined module NetGestalt provides two additional zoom in methods for visualizing any regions of the one dimensionally ordered network See below b Alt drag A user can drag the mouse across a region of interest while holding the Alt key down see Figure 11 c Double click A user can double click a region of interest to zoom in the tracks A user can hold the Shift button and double click the tracks to zoom out d Pan When the tracks are zoomed in a user can drag anywhere in the track panel to pan 14 o Jim 200 I399 Jess ses 6000 7000 8000 es fis 11000 K Figure 13 Zoom in the tracks by holding the Alt key down and dragging the mouse 6 Analyze tracks The current version of NetGestalt provides eleven features to help users analyz
41. le sbt 1s a tab delimited text file that contains lists of genes in separate rows e g significant genes from differential expression analysis File format description an SBT file contains track name track description and gene ids e g gene symbols in the track Each row represents a track and columns are separated by tab Up to three tracks three rows can be included in an SBT file To enable meaningful enrichment analysis the user can include in the first row an All track that contains the reference gene symbols for the tracks in the SBT file e g all genes on the microarray platform from which the differentially expressed genes were identified If this information is not provided enrichment analysis will be based on all genes in the network If the All track is provided all genes in the other tracks should be included in the All track No special characters for Track names Cells in red should not be changed Example All Description GeneSymbol 11 GeneSymbol 21 GeneSymbol 22 TrackName 1 Description GeneSymbol 11 TrackName2 Description2 GeneSymbol 21 GeneSymbol 22 NetGestalt uses the following rules to determine the file type of a given file 1 Use the file extension to determine the file type Ignore the txt file extension For example both test sbt and test sbt txt are treated as a SBT file 2 If that fails NetGestalt cannot determine the file type and displays an
42. le a single clinical feature gender was found to be significantly associated with at least one gene red box The resulting tracks are shown in the panel on the right blue box References Cancer Genome Atlas Network 2011 Integrated genomic analyses of ovarian carcinoma Nature 474 609 615 Cancer Genome Atlas Network 2012 Comprehensive molecular characterization of human colon and rectal cancer Nature 487 330 337 Cancer Genome Atlas Network 2012 Comprehensive molecular portraits of human breast tumours Nature 490 61 70 Cheung H W et al 2011 Systematic investigation of genetic vulnerabilities across cancer cell lines reveals lineage specific dependencies in ovarian cancer Proceedings of the 46 National Academy of Sciences of the United States of America 108 12372 12377 Shi Z Wang J and Zhang B 2013 NetGestalt integrating multidimensional omics data over biological networks Nat Methods 10 597 598 Subramanian A et al 2005 Gene set enrichment analysis a knowledge based approach for interpreting genome wide expression profiles Proceedings of the National Academy of Sciences of the United States of America 102 15545 15550 Wang J et al 2013 Integrative genomics analysis identifies candidate drivers at 3q26 29 amplicon in squamous cell carcinoma of the lung Clinical cancer research an official journal of the American Association for Cancer Research 19 5580 5590 Zhan
43. les will be sorted according to the rightmost feature 17 O 1800 11000 11500 17 000 2500 3000 2 500 4009 4300 5 000 15 500 6 000 16 500 17000 17 900 18 000 14 500 000 12 500 10 00 19 509 11 009 11 590 E a sa m 8 gt IM gt gt KANE r EJ e tg di i ne o y Nis IIR ya te ajrim i ugg t p 3 y H 33 E IV o Gaunt tuat ro s por rat k wb cp Jf ER ji d tx Feet sf PELE OE aes mss eid d Lr Hi E Manor i abi ili 5f Pe eer w trys NI os gt o TE a wees Us srk a o cats n r SEP i Marta EI ETER EI Ladies PES bo aiae if A emn i l K q taki 8 qat fizd 14 ei iiid thire bdis i Roi on it M PER 1 PIA Bi Figure 17 Subtrack annotation in NetGestalt d Statistical analysis The users can conduct statistical association tests on CCTs or CBTs e g clinical information like age gender etc when subtrack annotation data is available see section 6 c above for more on subtrack annotation data The Statistical Analysis button will be activated on a track s drop down menu if annotation data is available The statistical test used depends on the data type of the annotation data selected and the type of track data whether it is a continuous or binary track i Gene level vs Module level tests Statistical tests can be conducted at both the gene and module level on the subtrack annotations association with CCTs or CBTs F
44. lorectal proteomic profile tracks The colorectal cancer profile tracks are CCTs containing the processed mass spectrometric proteomics data for liquid chromatography tandem mass spectrometry LC MS MS based shotgun proteomic data on 90 TCGA primary solid tumor samples generated by Vanderbilt University Data is available at the shared and unshared peptide level and reported at the protein level see red box in Figure 38 The following steps were used to prepare the data tracks 1 The Protein Reports containing the spectral count and precursor AUC data files were downloaded from the CTPAC data portal https cpte xfer uis georgetown edu publicData Phase II Data TCGA Colorectal Cancer 2 The data is normalized as follows for both spectral count and precursor AUC data at the shared and unshared peptide level a Ifduplicate samples exists the sample with the highest signal 1s kept b Following deduping the data is normalized using the global normalization method 38 ii i Sum total signal for each sample Take max observed sum and generate normalization factors for each as max sum this sample sum ii Multiply all values in sample by that sample s normalization factor c Log transform the normalized values as log2 value 1 Expand all Collapse all EC 01 Main Data H E Breast_Cancer Colon Cancer E SY VU_proteome_ms_global a precursor AUC shared precursor AUC unshared profile correlations EJ clinical
45. mprove data visualization Pain sr AA A NENNEN AA AA A AA AA AA A A a Teor oa Gare sora laj motet nren notedirag meder Select a color scale acheme AT aM um Era rum Y vr M Cmm CORO cai M Figure 24 Data transformation in NetGestalt 24 f Value based filtering The Value based Filtering features allows users to filter for interesting genes e g differentially expressed genes from an SCT After clicking the Value based Filtering button in the menu Figure 23 a filtering dialog will be displayed The maximum and minimum values of the track are shown at the top of the dialog The user can input the parameters to filter the track After providing the name of the new track and clicking the Add Track button an SBT will be added to the track viewing area red box in Figure 23 Genes in the new SBT barplot are colored according to the original values in the SCT with blue for negative values and red for positive values Sabotes Delwer Normal Adenomas ogPsigned a ovp p Network Analysis Ales Apnd ope didis AA i E cana bi me 9 Gene Se Enrichment 9 Node Ank Graph Visible Range O Data TY x Track y Fiker Q Export Track Maximum 60 49 Track Morris 59 08 Dee Please select a range less Pan greater than poner ma ans less Pan e less Tan 30 or greater han 30 i Py Track u Figure 25 Value based filtering in NetGestalt g Presence b
46. nalysis features will be shown in a drop down menu see orange box in Figure 9 When clicking the i button see brown box in Figure 9 a table containing detailed information about the track will be shown Clicking the e button will export the track data see gray box in Figure 9 Clicking the x button will remove the corresponding track from the viewing area see gold box in Figure 9 We will introduce other buttons in the section Analyze tracks The user can also click the Clear all tracks button in the Track menu to remove all the tracks from the track viewing area see green box in Figure 5 5 Zoom in out tracks NetGestalt provides multiple methods to visualize tracks at different scales described below 13 a Click bars representing predefined modules NetGestalt uses horizontal bars to represent network modules sub networks at different hierarchical levels see purple box in Figure 10 The lengths of the bars correspond to the size of the modules For a specific hierarchical level genes within a module are highly connected whereas genes from different modules are loosely connected Most of the predefined network modules have been demonstrated to be functionally spatially or dynamically homogeneous Shi et al Nat Methods 2013 These modules can help users to easily associate subnetworks with experimental data User can click the corresponding bar to zoom into a module for further analysis see F
47. on using the Cytoscape Web plug in For CCTs and SCTs when the number of genes in the current visible range is less than 500 the Node link Graph Visible Range button in the drop down menu will be activated red box in the top plot of Figure 26 By clicking the Node link Graph Visible Range button network structure for all nodes in the current visible range will be displayed left bottom plot in Figure 26 For an SCT NetGestalt can color the genes in the network according to their original values in the SCT right bottom plot in Figure 26 26 For SBTs if genes in the track are colored the corresponding genes in the network graph will also be colored with the same colors left bottom plot in Figure 27 In addition by clicking the Node link Graph Present Nodes button red box in Figure 27 a network that only contains genes within the visible range and present in the SBT will be displayed right bottom plot in Figure 27 5 KY 12000 13 000 KE UTC NN UC NN 8 000 19 00 10 000 US NN S Mn eS Figure 28 Visualize network structure for CCT and SCTs in NetGestalt i Zoom to a gene The users can use the Zoom to a gene feature to zoom directly to a gene When typing the gene symbol in the Zoom to a gene box see purple box in Figure 28 NetGestalt will list all matching symbols below the box on the fly After selecting a symbol NetGestalt will zoom in to
48. or gene level tests the selected tests will be performed for all genes in the selected track with new SCTs generated for each type of test result which can include p values and test statistics see Figure 16 For module level analysis principle component analysis 1s used to combine all of the genes in the module into a single value and then the statistical methods are applied as normal see Figure 17 O e 3 000 4 000 5 000 6 000 7 000 8 000 9 A 9 2500 5 000 7 500 Gene_Sabates Bellver_ Normal_Adenomas_Size a oo I i ta so i se v y ll Figure 18 Gene level statistical test results are shown as new SCTs In this example the Pearson correlation test statistics red box and the log p values blue box were generated 18 o TE D T1000 12000 3 000 4 000 5000 6 000 7 000 8 000 9t Zoom to a Gene hide Enrichment Results hide o dr s m S A pt 7 500 No enrichment analysis performed Significant Modules Significant Network Modules Total 90 Module Nam p adjust stats 3 83e 9 0 662 959e 14 0 799 255e 10 0711 208e 7 0 607 130e 11 0 755 2 x zi Sabates Bellver Normal Adenomas B Lens i 3 74e 9 0 671 KIT tul ki ES dee 4 sss I Hd ble 0 00000 0 586 tj at My way imn 7 13e 13 0777 Lo bd orn O TEDE i pi b ah LI nno men T j I X i 4 25e 7 0 598 1 l TI m Wt I rh hil Wr 1
49. ortal contains both the protein protein interaction network views for a given species along with a chromosome view as well as a large set of functional information from sources such as KEGG Gene Ontology and Drugbank Currently there are separate portals for the following species Homo sapiens Arabidopsis thaliana Drosophila melanogaster Mus musculus Rattus norvegicus Caenorhabditis elegans Saccharomyces cerevisiae and Danio rerio In addition there are two human portals populated with publicly available data The Human Colorectal Cancer Portal which includes omics data from various colorectal tumor cohorts including The Cancer Genome Atlas TCGA cohort as well as colorectal cancer cell lines and the Human CPTAC Portal which includes data from both the TCGA study and the Clinical Proteomics Tumor Analysis Consortium CPTAC study of human breast colon and ovarian cancer Users must select a portal on the Netgestalt home page Figure 1 Please select a portal to enter a Human Colorectal Cancer Portal Human CPTAC Portal Copy Arabidopsis ved Fruit Fly Figure 1 Screen image of portal selection dropdown box 2 Seta view To perform analysis in NetGestalt a user should first select a view When hovering the mouse over the View menu see red box in Figure 2 different choices for setting the view will be shown in a drop down menu a Select a view Track About Current view netwo
50. ownloaded from the CTPAC data portal https cptc 39 xfer uis georgetown edu publicData Phase II Data TCGA Ovarian Cancer and https cptc xfer uis georgetown edu publicData Phase II Data TCGA Breast Cancer 2 The ITRAQ data provided by CPTAC is processed as follows a Ifduplicate samples exists the sample with the highest signal 1s kept b The iTRAQ data provided by CPTAC is already normalized so no additional normalization steps are required See CPTAC normalization steps described here https cptacdcc georgetown edu cptac documents CDAP ProteinReports des cription 20140708 pdf Expand all Collapse all E 01 Main Data E a Breast_Cancer ca Bl_proteome_ms_global _ Bl_phosphoproteome_ms_global UJ TCGA hypermethylations amp 7 TCGA mRNA RNA seq HC TCGA mRNA microarray TCGA somatic copy number alteratii E C TCGA somatic mutations CI Colon_Cancer ES Ovarian_Cancer H JHU proteome ms global amp C3 JHU glycoproteome ms global H PNNL proteome ms global H I PNNL phosphoproteome ms global H I TCGA mRNA RNA seq L TCGA mRNA microarray E HI TCGA somatic copy number alterati c 4 0 TCGA somatic mutations E ay 02 Functional_Tracks a 4 23 03 Drug related Tracks Figure 41 Overview of CPTAC track breast and ovarian data sets available outlined in red TRAQ data from liquid chromatography tandem mass spectrometry LC MS MS based shotgun proteomic data on 105 breast cancer
51. ph R function provided in the survival library is used to calculate the results 23 pO 000 2 O00 3 000 4 000 5 000 6 000 7 000 8 000 oe gt A g S ENT Tey F3 SRSA TIM a a LH rd V PS TE LARNER DA RE AR e Set LIU Ne RAD S prn Ns WM TCGA BRCA SCNA SNP6 GISTIC2 thresholded E let erre deje d aod tenet e Re IA A Be let eh UN al i H Ci 1 at s 4 y al On es d i y 7 MI VIA d n 1 rre miu Y uri A e d a s x Ferien Ern riore iram maf gis aed a Wale oe t Please select analysis level TENE a aO e x A A ARR AB ever MIR A rra yy LT mt p a e d n es prem y gene level Please selecta feature esh overallsurvival Select a test xy survival proportional hazards Feature overallsurvival_Event Sample TCGA_A2_A0SW_01 Perform FDR adjust Value 1 0 Yes O No Output track s hazard ratio signed logtest logipvalue vi signed sctest logipvalue signed waldtest logipvalue Figure 23 Setting up a statistical test using a survival annotation feature overall survival in this example e Data transformation To better visualize the CCTs NetGestalt provides a Data Transform feature which allows users to perform gene wise standardization by subtracting the gene wise mean or median and set floor and ceiling values for the data Figure 22 Similarly using the Data Transform feature users can also set floor and ceiling values for an SCT to i
52. rk_view hprd fo p 9 Delete hsapiens chromosomeView M oc SB lt _ lt O nl TS K y Upload Ref is PA 0 2 500 5 000 7 500 Figure 3 Set a view in NetGestalt When hovering the mouse over Select all views provided by the system will be shown in a menu The currently active view is shown in grey while others are shown in black The user can select a view by clicking the name After setting the view the user can find the name of the selected view and corresponding category at the top right of the window see brown box in Figure 3 In the current version NetGestalt contains two categories of views network views and chromosome views Each portal contains a single chromosome view and at least one network view For the human portals the hprd and REF correspond to the HPRD human protein protein interaction PPI network http www hprd org and Ref human PPI network http wodaklab org iRefWeb respectively v i w e Track View Current view network view iRef e TT 1000 T2000 3 000 4 000 5000 6 000 7 000 8 000 9 000 10 000 11000 Ti u 0 2 500 5 000 7 500 10 000 Track Y View Y Current view network_view hsapiens_chromosomeView A 110000 115 000 20 000 25 000 30 000 PRERE Ph me a im me a Dee iT EE 0 5 000 10 000 15 000 20 000 25 000 30 000 Figure 4 Examples of a network view top derived from the REF protein protein interaction network
53. s and two SBT files recording the up regulated and down regulated differentially expressed genes Sub category 06 Differential protein expression contains six tracks including one CCT file 33 recording the protein expression matrix three SCT files recording the W statistic values signed logP p values were calculated based on Wilcoxon test and log2 fold changes and two SBT files recording the up regulated and down regulated differentially expressed proteins For 07 Correlation between mRNA and protein 08 Correlation between SCNA and mRNA and 09 Correlation between SCNA and protein each of them contains two tracks both of which are SCT files These SCT files record the Spearman s correlation coefficient and corresponding signed logp b Clinical relevance based on CRC tumor tissue All the clinical relevant data tracks are included in the Category 02 Clinical relevance red box in Figure 33 These tracks are divided into four categories 01 Clinical relevance snapshot 02 Survival including Markers for overall survival and Markers for disease free survival 03 Stage IV vs Stage I including Signature genes for Stage IV vs Stage I and Signature proteins for Stage IV vs Stage I 04 MSI vs MSS including Signature genes for MSI vs MSS and Signature proteins for MSI vs MSS 01 Genomic and proteomic alterations EQ ce nicas b mi 01 Clinical relevance snapshot 23 02_Survival 1 01_Markers for overall survival
54. s are based on the Fisher s exact test and the KS test for SBTs and SCTs respectively o 1000 X 2000 3 000 4 000 5 000 6 000 7 000 8 000 9 Main Help Zoom to a Gene hide 0 2 500 5 000 7 500 nrichment Results e 1 Sabates Belver Normal Adenomas_siggene CR T A E E UL ALL IU Original Track TIT TT I rra 4 Network Analysis Sabates Bellver Normal Adenomas Gene Set Enrichment Enriched Tracks in go bp Node xi Total 256 G Node Find related functional tracks Track p adjust ER Presei 2 cellmap go bp go cc cell cycle phase 1 41e 14 2 54 go mf humancyc kegg nci M phase 1 24e 13 O Track reactome cell cycle process 2 92e 13 t e Expori Select FDR cutoff cell proliferation 340e 43 2 Delete 0 001 0 005 0 01 mitotic cell cycle 4 20e 12 E 0 05 0 1 0 25 2 00e 11 mitoti 2 00e 11 545e 11 1 Search 1 53e 9 immune response 2 12e 9 10 Entries Per Page Page 1 of 26 Figure 16 Gene set enrichment analysis in NetGestalt c Subtrack annotation For composite tracks CCTs and CBTs containing multiple samples i e subtracks using the Subtrack Annotation feature in the drop down menu accessible from the track name users can visualize sample information as a sample heat map with black to green colors for binary data e g Tissue categorical data e g Location and continuous data e g Size Figure 15 The order of the sample features can be rearranged and the samp
55. s at least 5 values for EACH category c gt 90 of values for the feature are NA c Statistical correlation results Statistical association tests see section II 6 c were conducted for all of the CCTs CBTs in the portal both from CPTAC and TCGA and separate SCTs were generated for the resulting p values q values and test statistics see Figure 41 An SBT of significant genes with a q value lt 0 05 is also generated For each CCT CBT gene level association tests are conducted for each subtrack annotation feature see sections 11 2 c 111 and IIL3 c While all subtrack annotation features for a given track are statistically associated with the CCT CBT data tracks are generated and available in the correlations subfolder for a given subtrack annotation feature only if at least one gene has a q value lt 0 05 In the example given in Figure 42 only a single clinical feature gender red box was found to have at least one significantly associated gene for the precursor AUC unshared peptides data from the CPTAC Colorectal proteomic profile data available by click on profile in purple box while five different molecular subtype features were found with at least one significant result green box 44 ec BI proteome ms global ec shared y correlations 1 3 copy number alteration 1 3 molecular subtype amp CJ mutation H unshared 1 3 Bl phosphoproteome ms global 1 2 TCGA hypermethylations H O
56. s in the new binary track are colored corresponding to the colors in Venn diagram Le xata Wre grana paste ey iil vet kapej p mm p E Suntan Siri hi al Ach bas hara fs paret Figure 31 Track comparison in NetGestalt k Track co visualization NetGestalt allows users to co visualize two single tracks SBT or SCT in a node link graph using the border and fill colors of the nodes to represent data in the two tracks respectively First the user should select the tracks to be co visualized SBT or SCT and the node attributes border or fill colors associated with each track in the Track Co visualization section located in the left panel of the page brown box in Figure 30 After clicking the G or g buttons a node link graph which contains edges between all genes in the visible range or all present genes in the visible range will be displayed left bottom and right bottom plots in Figure 30 I Switch between different views The users can visualize the same set of tracks in different network views by making changes using the Views Select menu as shown in the Figure 2 This feature allows users to explore the same data sets in the different biological contexts 29 Main np Lon 000 s A a 1 i SQ No ervichment anatua poctormed N GA OM tap ece Agora 07 3 Ravel Bega H Ti YCOA BM xpOsee Agen V om mani LN s I I l M
57. s value If a gene level test is selected a Fisher s exact test will be conducted If a module level test is selected users must choose between an Anova or Kruskal Wallis test Next users must select whether or not to perform an FDR adjustment Next users must select which results to output as new SCTs which is currently limited to the p values log10 transformed orange box Finally the user should click Go green box The aov kruskal test or fisher text R functions are used to calculate the results vi Binary annotation data and CCT statistical testing When users select a categorical annotation feature e g Tissue etc See section 2 c 111 Track sample information TSI file in a CCT they first select the direction of the two values e g adenoma normal vs normal adenoma see purple box in Figure 20 Next users must choose between a t test or Wilcoxon rank sum test red box Next users must select whether or not to perform an FDR adjustment blue box Next users must select which results to output as new SCTs which is currently limited to the signed p values log10 transformed with same sign as test statistics and or the test statistic orange box Finally the user should click Go green box The t test and wilcox test R functions are used to calculate the results vii Binary annotation data and CBT statistical testing When users select a categorical annotation feature e g Tiss
58. scription a CCT file 1s a data matrix in which each row represents a gene and each column represents a sample The first column lists the gene ids e g gene symbols and the first row lists the sample names Two or more data columns sub tracks are required Columns must be separated by tab Each cell in the matrix 1s a continuous value for corresponding gene and sample Missing values are represented by NA Data for different samples must be comparable 1 e properly normalized Duplicated row names or column names are not allowed No special characters for row or column names Example GeneSymbol Samplel Sample2 Sample3 Sample4 Genel 0 025 0 55 1 0 095 Gene2 0 077 0 069 0 64 0 18 Gene3 0 47 l 0 87 0 88 Gene4 0 71 0 19 0 33 0 45 11 Composite binary track CBT file Definition a CBT file cbt is a tab delimited text file that contains data for a composite track vvith multiple related sub tracks vvith binary data e g mutation status for genes in multiple samples The file name will be used as the track name File format description same as the CCT file except that each cell in the matrix is a binary value 0 or 1 for corresponding gene and sample iii Track sample information TSI file Definition a TSI file ts1 1s a tab delimited text file that contains the sample information for a CCT or CBT This file is an optional sample annotation file for the matching CCT CBT file and it shoul
59. st red box Users must then select whether or not to perform an FDR adjustment blue box and which results to output as new SCTs which is currently limited to the signed p values log10 transformed with same sign as test statistics and or the test statistic orange box Next users must select whether or not to perform an FDR adjustment blue box and which test results to output as new SCTs which is currently limited to the signed p values log10 transformed with same sign as test statistics and or the test statistic orange box Finally the user should click Go green box The t test and wilcox test R functions are used to calculate the results The t test or wilcox test R functions are used to calculate the results If a module level test 1s selected they must choose between the Spearman or Pearson correlation tests as describe above in see 11 6 d 11 iv Categorical annotation data and CCT statistical testing When users select a categorical annotation feature e g tumor location etc See section 2 c 111 Track sample information TSI file in a CCT they must choose between an Anova or Kruskal Wallis test see red box in Figure 19 Next users must select whether or not to perform an FDR adjustment blue box Next users must select which results to output as new SCTs which is currently limited to the p values log10 transformed orange box Finally the user should click Go green box T
60. tracks including Cell Map pathways GO Biological Processes GO Cellular 3l Components and GO Molecular Functions HumanCyc pathways KEGG pathways NCI pathways and Reactome pathways Expand all Collapse all 5 g Data Tracks E lt 01 Main Data Colorectal cancer CRC 01 Primary Tumor Tissue C 01_Genomic and proteomic alter 4C 02 Clinical relevance m 02 Cell Lines cj y 02 Functional Tracks Cell Map 3 Gene Ontology Y HumanCyc KEGG 0 NCI Reactome Figure 33 Data tracks in NetGestalt CRC portal a Genomic and proteomic alterations in the TCGA CRC tumor cohort Tracks derived from the multidimensional omics data on the TCGA CRC tumor cohort are included in the Category 01 Genomic and proteomic alterations red box in Figure 32 These tracks are divided into nine sub categories 01 Omics snapshot 02 Somatic mutation 03 Somatic copy number alteration SCNA 04 Epigenetic silencing 05 Differential mRNA expression 06 Differential protein expression 07 Correlation between mRNA and protein 08 Correlation between SCNA and mRNA and 09 Correlation between SCNA and protein 22 01_Primary Tumor Tissue ET ki 01 Omics snapshot E E 02 Somatic mutation 03_Somatic copy number alteration Y 04 Epigenetic silencing 05 Differential mRNA expression E 06 Differential protein expression UJ 07 Correlation between mRNA and protein 08 Correlation between SCNA and mRNA E
61. ue etc See section 2 c 111 Track sample information TSI file in a CBT the statistical test used depends on whether or not the test is to be conducted as the gene level in which case the CBT data will be binary or at the module level in which case the module level CBT data will be a continuous value If a gene level test is selected a Fisher s exact test will be conducted If a module level test is selected users must choose between a t test or Wilcoxon rank sum test If a module level test 1s selected users must first select the direction of the two values e g adenoma normal vs normal adenoma see purple box in Figure 20 Next users must choose between a t test or Wilcoxon rank sum test red box Users must then select whether or not to perform an FDR adjustment blue box and which results to output as new SCTs which is currently limited to the signed p values log10 transformed with same sign as test statistics and or the test statistic orange box If a gene level test is selected users must select whether or not to perform an FDR adjustment blue box Next users must select which results to output as new SCTs which is currently limited to the signed p values log10 transformed Finally the user should click Go green box The t test wilcox test or fisher text R functions are used to calculate the results 21 E ET 2000 3 000 PO 6000 7000 8 000 EX Saas ia a S ux m s E _ rr yo gt
62. ue to red The first track in Figure 9 green box is a CCT track representing a gene expression data containing 32 CRC samples and 32 normal mucosa samples When hovering the mouse over the heat map plot NetGestalt will show the gene 1d sample index and sample name at the corresponding position b CBTs NetGestalt visualizes a CBT track with a heat map of two colors red and grey The fourth track in Figure 9 yellow box shows a CBT track representing a TCGA GBM somatic mutation data containing 148 CRC samples Hovering the mouse over the heat map plot will show the gene symbol sample index and sample name at the corresponding position c SCTs NetGestalt visualizes a SCT track with a bar plot The second track in Figure 9 purple box is an SCT track containing log p value of gene expression between CRC samples and normal samples based on Sabates Bellver dataset GSE8671 Hovering the mouse over the bar plot NetGestalt will display the gene symbol and statistic value at the corresponding position d SBTs NetGestalt visualizes a SBT track with a barcode plot The third track in Figure 9 black box is an SBT track representing significantly genes based on Sabates Bellver dataset Hovering the mouse over the plot will display gene symbol at the corresponding position 12 9 00 o 500 1000 2 000 2 500 3 000 3 500 4 000 14500 5000 5500 6 000 16 500 7 000 7 500 8 000 8 500 q Node link Graph Visible Range kul D
63. value annotation feature e g Tumor size age etc See section 2 c 111 Track sample information TSI file for a CCT they must choose between the Spearman or Pearson correlation tests see red box in Figure 18 Next users must select whether or not to perform an FDR adjustment see blue box Finally users must select which results to output as new SCTs the signed p values logl0 transformed with same sign as corresponding test statistic and or the correlation test statistics orange box and click Go green box The cor R function is used to calculate the Pearson s r or Spearman s rho values iii Continuous annotation data and CBT statistical testing When users select a continuous value annotation feature e g Tumor size age etc See section 2 c 111 Track sample information TSI file for a CBT the statistical test used depends on whether or not the test is to be conducted as the gene level in which case the CBT data will be binary or at the module level in which case the module level CBT data will be a continuous value If a gene level test is selected users must choose between a t test or Wilcoxon rank sum test 19 red box where the mutation status is used to dichotomize the continue feature data If a gene level test is selected users must first select the direction of the two mutation status values 0 1 vs 1 0 see purple box in Figure 20 Next users must choose between a t test or Wilcoxon rank sum te
64. y clicking the Add all related modules link at the top of the table see blue box in Figure 13 In this composite track each row represents a hierarchical level of the network enriched modules are colored in light red and genes in the original binary track are colored in red The enrichment section can be hidden shown by toggling the hide show button see brown box in Figure 13 15 vy y v e I f Ja o es 000017000 fes fs s me TI Figure 15 Output for netyvork analysis in NetGestalt For an SCT NetGestalt uses the Kolmogorov Smirnov test KS test to identify enriched modules When clicking an entry in the table or Add all related modules NetGestalt will add the leading edge genes in the enriched module as a new binary track see Subramanian et al PNAS 102 43 15545 15550 2005 for the definition of leading edge genes Clicking the Add all related modules link at the top of the table will add a new composite track in which each row represents a hierarchical level of the network enriched modules are colored in light red and the leading edge genes are colored in red 11 Network expansion To expand genes in an SBT ie the seed genes to include other related genes in the network two options are available in the Section Network expansion under Network Analysis green box in Figure 12 The all neighbors option works for SBTs containing 10 or

NetGestalt User Manual

Contents

Download Pdf Manuals

Related Search

Related Contents