Difference between revisions of "InterMine Presentation"

Revision as of 21:16, 8 February 2007

This Wiki page is an edited version of Gos's presentation

Background

InterMine was developed as the generic underpinnings of the FlyMine Project

• Team of 7 FTE
  • 5 developers, one sys admin,
  • 1 biologist/ bioinformatician
• Java/ postgreSQL
• SVN repository: 125,000 lines of code + 57,000 lines of tests
• Under development since 2002
• In use by others in Cambridge, Edinburgh, Vienna… + modENCODE DCC if funded
• modENCODE/ Chado

Technical Overview

• Data model --> Java classes, relational schema, mappings through automatic code generation
• Custom Java object/relational system
  • When we started, couldn’t select from multiple classes at one time using hibernate.
• Optimised for read-only performance
• Designed for big, complex queries
• Performance optimisation
  • Transparent query re-writing
• Web application - Struts/JSP/Ajax

Test Problem

• Used SOFA as core data model - similar to Chado.
• Added Gene.description (absent from model), compiled, loaded data (here XML + FASTA), released webapp.

Loading Data

• Read-only in production environment
• Load data from InterMine XML
• Parsers from standard formats
  • e.g. UniProt, GFF3, PSI, FASTA
• Powerful integration system

Example InterMine XML

<xml> <items>

  <item id="0_3" class=”” implements="http://www.flymine.org/model/genomic#Gene">
     <attribute name="identifier" value="xfile" />
     <attribute name="description" value="A test gene for GMOD meeting" />
     <reference name="organism" ref_id="0_1" />
     <collection name="transcripts">
        <reference ref_id="0_9" />
     </collection>
  </item>
  <item id="0_1" class="" implements="http://www.flymine.org/model/genomic#Organism">
     <attribute name="taxonId" value="7227" />
  </item>
  ...

</xml>

Resulting webapp object page

thumb|xfile Gene details page

Quicksearch

Java API

<java>

 Query q = new Query();
 QueryClass qcObj = new QueryClass(Gene.class);
 q.addFrom(qcObj);
 q.addToSelect(qcObj);

 QueryField qf = new QueryField(qcObj, "identifier");

 SimpleConstraint sc = new SimpleConstraint(qf, ConstraintOp.MATCHES, new QueryValue("x-%"));
 q.setConstraint(sc);

</java>

IQL

<sql>

 SELECT DISTINCT a1_.identifier AS a2_ FROM org.flymine.model.genomic.Gene AS a1_ WHERE a1_.identifier LIKE 'x-%'

</sql>

Perl API

<perl>

 my $genes = InterMine::Gene::Manager->get_genes(query => [
                            identifier => { like => 'x-%' },],);

</perl>

Bake-Off code

<java> public class BakeOff {

   public static void main(String[] args) throws Exception {
       // code to get the "xfile" gene
       ObjectStore os = ObjectStoreFactory.getObjectStore("os.production");
       Query q = new Query();
       QueryClass qcObj = new QueryClass(Gene.class);
       q.addFrom(qcObj);
       QueryField qf = new QueryField(qcObj, "identifier");
       q.addToSelect(qf);
       SimpleConstraint sc = new SimpleConstraint(qf, ConstraintOp.EQUALS, new QueryValue("xfile"));
       q.setConstraint(sc);
       System.err.println("query: " + q);
       Results res = os.execute(q);

       // a Results object is a List of Lists
       List rr = (List) res.get(0);
       Gene gene = (Gene) rr.get(0);

       System.err.println ("symbol: " + gene.getIdentifier());

       // a BioEntity in FlyMine has a collection of Synonym objects -
       // we need Synonym.value for each Synonym
       System.err.print ("synonyms: ");
       Iterator synIter = gene.getSynonyms().iterator();
       while (synIter.hasNext()) {
           Synonym syn = (Synonym) synIter.next();
           System.err.print (syn.getValue() + ' ');
       }

       System.err.println ("description: " + gene.getDescription());

       // get the class name, but we already know that the gene is a Gene
       System.err.println ("type: " + gene.getClass().getName());

       // make a List from a the Set of exons for this Gene
       List exons = new ArrayList(gene.getExons());
       Exon exon1 = (Exon) exons.get(0);
       Exon exon2 = (Exon) exons.get(1);

       // get the start and end via the Location object
       System.err.println ("exon1 start: " + exon1.getChromosomeLocation().getStart());
       System.err.println ("exon1 end: " + exon1.getChromosomeLocation().getEnd());
       System.err.println ("exon2 start: " + exon2.getChromosomeLocation().getStart());
       System.err.println ("exon2 end: " + exon2.getChromosomeLocation().getEnd());

       // write out the first cds
       List cdss = new ArrayList(gene.getCDSs());
       FlyMineSequence flymineSequence = FlyMineSequenceFactory.make((CDS) cdss.get(0));

       // use BioJava to output the sequence
       Annotation annotation = flymineSequence.getAnnotation();
       annotation.setProperty(FastaFormat.PROPERTY_DESCRIPTIONLINE,
                              gene.getIdentifier() + " cds");
       SeqIOTools.writeFasta(System.err, flymineSequence);
   }

}

</java>