Examples with StringProteinProteinInteractionObjectGenerator ubic.gemma.core.loader.protein.string.StringProteinProteinInteractionObjectGenerator used on opensource projects

Example 1 with StringProteinProteinInteractionObjectGenerator

use of ubic.gemma.core.loader.protein.string.StringProteinProteinInteractionObjectGenerator in project Gemma by PavlidisLab.

the class StringProteinInteractionLoader method load.

/**
 * Main method to load string protein protein interactions. Can either be supplied with files to load from or do
 * remote download. After files have been located/fetched the files are parsed and converted into value objects.
 * These value objects are then converted into GEMMA Gene2GeneProteinInteractions. Which are then loaded into the
 * database. Can be run on all eligable TAXA in gemma or on a supplied taxon.
 *
 * @param stringProteinFileNameLocal     The name of the string file on the local system
 * @param stringProteinFileNameRemote    The name of the string file on the remote system (just in case the string name
 *                                       proves to be too variable) - can be null
 * @param localEnsembl2EntrezMappingFile The name of the local biomart file - can be null?
 * @param taxa                           taxa to load data for. List of taxon to process
 * @throws IOException io problems
 */
public void load(File stringProteinFileNameLocal, String stringProteinFileNameRemote, File localEnsembl2EntrezMappingFile, Collection<Taxon> taxa) throws IOException {
    // very basic validation before any processing done
    this.validateLoadParameters(stringProteinFileNameLocal, taxa);
    // retrieve STRING protein protein interactions
    StringProteinProteinInteractionObjectGenerator stringProteinProteinInteractionObjectGenerator = new StringProteinProteinInteractionObjectGenerator(stringProteinFileNameLocal, stringProteinFileNameRemote);
    Map<Taxon, Collection<StringProteinProteinInteraction>> map = stringProteinProteinInteractionObjectGenerator.generate(taxa);
    /*
         * Get ENSEMBL to NCBI id mappings so we can store the STRING interactions
         */
    Map<String, Ensembl2NcbiValueObject> bioMartStringEntreGeneMapping = this.getIdMappings(localEnsembl2EntrezMappingFile, taxa);
    // To one taxon at a time to reduce memory use
    for (Taxon taxon : map.keySet()) {
        StringProteinInteractionLoader.log.debug("Loading for taxon " + taxon);
        Collection<StringProteinProteinInteraction> proteinInteractions = map.get(taxon);
        StringProteinInteractionLoader.log.info("Found " + proteinInteractions.size() + " STRING interactions for: " + taxon);
        this.loadOneTaxonAtATime(bioMartStringEntreGeneMapping, proteinInteractions);
    }
}