msigdb-to-biopax

Originated from https://bitbucket.org/armish/gsoc14 and will continue here (ToDo).

MSigDB (TRANSFAC) to BioPAX Level3 data converter

This type of data sets are available from different resources, such as TRANSFAC, JASPAR and ENCODE projects; but converting the data provided by these databases will require additional processing—e.g. mapping ChIP-Seq peaks to near-by genes or finding genes by binding motifs.

The easiest way, however, to import this data type is to use MSigDB, where transcription factors and their predicted targets are present in GSEA format provided by TRANSFAC. These associations can be converted to BioPAX by transcription events regulated by corresponding transcription factors.

Data source

• Home page: http://www.broadinstitute.org/gsea/msigdb/collections.jsp
• Type: Transcription factor - target
• Format: XML (MSigDB format)
• License: Free for academic use (also see MSigDB license)

Implementation details

MSigDB provides data in various formats, the most commonly used being a tab-delimited format, GMT or GSEA, to describe a gene set. It also provides downloadable files in XML format where full description of gene sets can be obtained in a single file. This converter, with the help of the Java-based GSEA software as a library (MIT proprietary), parses the XML to obtain information about gene sets that describe transcription factors and their targets as gene sets. For this, we are specifically extracting information from the C3 category, specifically the TFT subcategory of it.

The genes in the gene sets are identified either via their symbols or Entrez Gene IDs. Although unstructured, the name of the transcription factor associated with each gene set is present in the description of the gene set. This converter has a built-in HGNC utility that helps better map the gene identifiers to full BioPAX RnaReferences. For each TFT gene set, it creates a separate pathway identified by the unique name, and within this pathway, the transcription factor positively regulates the transcription of targets listed in the corresponding gene set. This information is captured via TemplateReactions where target Rnas are being produced, and the reaction itself is being regulated by the transcription factor via TemplateReactionRegulation.

Usage

This project uses the MIT GSEA java library (gsea2-2*.jar), which is not open source (as well as MSigDB data); see GSEA/MSigDB licence and this wiki page . So, we downloaded the software at http://software.broadinstitute.org/gsea/downloads.jsp and added to the lib folder (we also manually removed the junit package from that jar).

Check out (git clone) the repository, change to:

$ cd msigdb-to-biopax

To compile the code and create an executable JAR file, run:

$ mvn clean package

You can then run the converter as follows:

$ java -jar target/msigdb-to-biopax.jar 
Usage: MsigdbToBiopax /path/to/msigdb_v5.2.xml /path/to/output.owl

For the conversion, you need to download the MSigDB database as an XML file: msigdb_v5.2.xml. Once downloaded, you can convert this into BioPAX as follows:

$ java -jar msigdb-to-biopax.jar msigdb_v5.2.xml msigdb_v5.2.owl