Common Parent Between Go Terms

Question

[Yogesh Pandit]

Hello,

How to programmatically (Python/Java) find out the common parent between 2 GO IDs (gene ontology)? And How to calculate the distance between each GO ID and the common parent?

Thanks

Answer 1

[Chris Mungall]

If you would like an alternative method that doesn't require a relational database, here is a link to a java library we are using within the GO consortium:

http://code.google.com/p/owltools/

It wraps the OWL API and is intended to work over any obo format or owl ontology.

Here is an example for finding the least common ancestor between any two nodes in the GO graph:

import owltools.graph.*;
import owltools.sim.*;
import owltools.io.*;

OWLGraphWrapper  g =  getOntologyWrapper("file:go.obo");

// negative regulation of type B pancreatic cell apoptosis
OWLObject a = g.getOWLObjectByIdentifier("GO:2000675");

// caspase activator activity
OWLObject a = g.getOWLObjectByIdentifier("GO:0008656");

SimEngine se = new SimEngine(g);
for (OWLObject c : se.getLeastCommonSubsumers(a,b)) {
  System.out.println(c);

  // there may be different paths with different distances between {a,b}
  // and c - just exhaustively list them all. if we like, we could
  // average them, find the max, etc
  for (OWLGraphEdge e : se.getEdgesBetween(a,c)) {
    System.out.println("  Edge:" +e +" distance: "+e.getDistance());
  }
  for (OWLGraphEdge e : se.getEdgesBetween(b,c)) {
    System.out.println("  Edge:" +e +" distance: "+e.getDistance());
  }
}

For more, see:

http://code.google.com/p/owltools/wiki/FindingCommonAncestors

Answer 2

[Pierre Lindenbaum]

Here is my java solution:

import java.sql.*;
import java.util.*;

public class GoDistance
    {
    private Connection connection;
    private GoDistance() {}

    /** returns a map (go-id/distance) */
    private Map<String,Integer> ancestors(String go) throws SQLException
        {
        Map<String,Integer> acn2distance=new HashMap<String,Integer>();
        //put self
        acn2distance.put(go,0);
        PreparedStatement stmt=null;
        ResultSet row=null;
        try 
            {
            //search all the parents
            stmt=connection.prepareStatement(
            "SELECT DISTINCT "+
            "   graph_path.distance,"+
            "   ancestor.acc "+
            " FROM "+
            "   term "+
            " INNER JOIN graph_path ON term.id=graph_path.term2_id) "+
            " INNER JOIN term AS ancestor ON ancestor.id=graph_path.term1_id) "+
            " WHERE term.acc=?"
            );
            stmt.setString(1,go);
            row=stmt.executeQuery();
            while(row.next())
                {
                int distance=row.getInt(1);
                String acn=row.getString(2);
                Integer prev=acn2distance.get(acn);
                if(prev==null || prev>distance)
                    {
                    acn2distance.put(acn,distance);
                    }
                }
            return acn2distance;
            }
        finally
            {
            if(row!=null) row.close();
            if(stmt!=null) stmt.close();
            }
        }

    private void run(String go1,String go2) throws Exception
        {
        Class.forName("com.mysql.jdbc.Driver");

        connection=DriverManager.getConnection(
                "jdbc:mysql://mysql.ebi.ac.uk:4085/go_latest"+
                "?user=go_select&password=amigo"
                );
        //get all parents of go1
        Map<String,Integer> acn2dist1=ancestors(go1);
        //get all parents of go2
        Map<String,Integer> acn2dist2=ancestors(go2);
        connection.close();

        //common terms
        Set<String> acns=new HashSet<String>(acn2dist1.keySet());
        acns.retainAll(acn2dist2.keySet());
        if(acns.isEmpty()) return;
        //find the minimal distance to a common term
        Integer bestDist=null;
        String bestTerm=null;
        for(String acn:acns)
            {
            int d= acn2dist1.get(acn)+acn2dist2.get(acn);
            if(bestDist==null || bestDist>d)
                {
                bestDist=d;
                bestTerm=acn;
                }
            }
        if(bestDist==null) return;
        //print result
        System.out.println(bestTerm+"\t"+(acn2dist1.get(bestTerm)+acn2dist2.get(bestTerm)));
        }

    public static void main(String args[])
        throws Exception
        {
        if(args.length!=2) return;
        new GoDistance().run(args[0],args[1]);
        }
    }

Compilation:

$ javac -Xlint GoDistance.java

Execution:

$ java -cp path/to/mysql-connector.jar:. GoDistance "GO:0001578" "GO:0030036"
GO:0007010  3

[Chris Mungall]

Thanks Pierre

(extra points for anyone who can do the "least" part directly in SQL using some fiendish aggregate operators...)

For fans of Hibernate, there is a Hibernate API for the GO mysql database. It would probably be over-the-top for this example anyway.

We also hope to have an alternative more modern database schema available for querying soon (with default DBMS of postgresql). This also has its own Hibernate layer. An announcement will be sent to gofriends when we have something ready for testing.

[Pierre Lindenbaum]

@Chris, I 'solved' the mysql problem a few years ago by creating an UDF function: http://plindenbaum.blogspot.com/2010/02/mysql-user-defined-function-udf-for.html

[Yogesh Pandit]

perfect. I was not sure about the query. thank you so much

Answer 3

[Nicolas Rochette]

I wanted to query GO with python, this page helped me a lot ! Thanks.

The mySQLdb module is needed, you can install it through aptitude (python-mysqldb)

import mySQLdb
go_db=MySQLdb.connect("mysql.ebi.ac.uk", port=4085, db="go_latest", user="go_select",passwd="amigo")
go_db_cursor=go_db.cursor()

query="..."
go_db_cursor.execute(query)
results=go_db_cursor.fetchall()

The MySQLdb user guide : http://mysql-python.sourceforge.net/MySQLdb.html GO request howtos : http://wiki.geneontology.org/index.php/Example_Queries