Modules 6 and 7: Genboree and Epigenome Comparison Aleksandar Milosavljevic Epigenomics Data Analysis and Coordination Center (EDACC) Presented at the 2 nd Epigenome Informatics Workshop Houston, Texas, September 10t h 2010
Diversity of Epigenomes
Comparing Epigenomes
Modules 3, 4, 5
Modules 6, 7
Tools Genboree Workbench and Atlas Gene Browser Genome-centric databases Users Groups Roles Access Projects Genboree Workbench Atlas Gene Browser
Genboree Workbench Genboree Tools>>Workbench
Genboree Workbench
Preliminary Exercises 1. Go to Genboree Tools>>Workbench 2. Expand and shrink Data Selector tree in the Workbench 2. Examine properties in “Details” window by clicking on objects in Data Selector 3. Read about required tool inputs and outputs by clicking on inactive tool icons 4. Activate tool icons by selecting required Input and Output
Comparison of “broad” peaks
Exercise #1: Pairwise epigenome signal comparison Go to Genboree Workbench Drag the following tracks into Input box: Group: EDACC Workshop 2010 Database: EDACC Workshop 2010 Class: 0.6 HDHV_Data Input Track: meCG:h1_bar Input Track: meCG:imr90_bar Class: GeneModel Input Track: GeneModel:5’UTR Drag your database into Output box. Run Signal Comparison: Analysis>Epigenomics>SignalComparison Check for results
Two types of Genboree annotation tracks 1.HDHV (High Density - High Volume) Example: go to Genboree Browser and select Group: EDACC Workshop 2010 Database: EDACC Workshop 2010 Class: 0.6 HDHV_Data Input Track: meCG:h1_bar Input Track: meCG:imr90_bar 2. Regular tracks Example: Class: GeneModel Input Track: GeneModel:5’UTR
Using attribute-values for annotations Go to Genboree Browser Group: EDACC Workshop 2010 Database: EDACC Workshop 2010 Class: GeneModel Input Track: GeneModel:5’UTR Check out many available attributes using Tabular View
Genboree Help and LFF format Click on Genboree “Help?” button in the top right area. Check “Help” entries. Check out “LFF format”. View the documentation of attribute-value pairs.
Exercise #2: Building a query Go to Genboree Workbench Drag your database into output box. Build “KEGG_PATHWAY-repair-5UTR” query:
Exercise #2 (cont’d): Running a query Running a KEGG_PATHWAY-5UTR query: Group: Epigenome Atlas Database: Your Database Drag into Input Data: All Annotations in Database and query “KEGG_PATHWAY-5UTR” Drag into Output Targets: Your Database Run Query
Exercise #3: Atlas similarity search Go to Genboree Workbench Drag the following tracks into Input Data box: Group: Epigenomics Roadmap Repository Database: Data Freeze 1 – Full Repo Class: 06.HDHV Data Input Track: SD:H1 H4Kme3 Drag your database into Output Targets box. Run Atlas Similarity Search: Analysis>Epigenomics>Epigenome Atlas Search Repeat with Input Track: H1 H3K9ac Check for results (results in about 1 hour)
Exercise #3 (cont’d): Atlas sim. search Hello Aleksandar Milosavljevic, Your searchAtlasSim tool run has completed successfully. Job Summary: JobID: wbJob _ Input Track: SD:H1 H3K4me3 The number of targets : 247 Result Summary: The top 10 target track matches: Track Name Pearson correlation SD:H1 H3K4me3 1.0 BI:iPS18c H3K4me BI:iPS11a H3K4me SF:H1 H3K4me BI:ESI3 H3K9ac BI:H1 H3K4me SF:H1 H3K4me BI:ESI3 H3K4me BI:iPS15b H3K4me SD:H1 H3K4me
Exercise #3 (cnt’d): Atlas sim. search Hello Aleksandar Milosavljevic, Your searchAtlasSim tool run has completed successfully. Job Summary: JobID: wbJob _ Input Track: SD:H1 H3K9ac The number of targets : 247 Result Summary: The top 10 target track matches: Track Name Pearson correlation SD:H1 H3K9ac 1.0 BI:H1 H3K9ac BI:H1 H3K9ac BI:iPS15b H3K9ac BI:iPS18c H3K9ac SF:H1 H3K9ac BI:iPS20b H3K9ac BI:H1 H3K4me BI:Pancreas H3K9ac BI:ESI3 H3K9ac
Exercise #4: Atlas Gene Browser
Exercise #1 (continued): Epigenome comparison Check output track reporting results of comparison in 5’UTR methylation in IMR90 relative to H1. Sort the track by score – note largest decrease in IMR90 relative to H1 (largest negative score) in ODF1 Visualize output track in Genboree browser. Search for ELF3 and visualize increase in IMR90 relative to H1.
Querying to the output of comparisons Example: Find all repair-related genes whose 5’UTR regions are much less methylated in IMR90 than in H1
Genboree and scripting Everything that can be done interactively using the workbench can also be done programmatically, typically using short scripts in programming languages like Perl, Python, or Ruby. This is accomplished through Genboree HTTP-based REST APIs (Application Programming Interfaces). Proof: the Genboree Workbench itself was written using Genboree REST APIs. To learn more, check Genboree Help ( under REST APIs ) – the help is correct but incomplete -- does not document all available functionality. Interested to start using Genboree REST APIs? Attend Andrew Jackson’s talk ( Module 8 )