[BiO BB] predicting functional genomic neighborboods

Dave Roe roex0050 at umn.edu
Tue Jun 20 12:28:30 EDT 2006

I have a contig containing several genes from an otherwise unsequenced
bacteria.  I want to do comparative genomic analysis to check for evidence
that the genes in this contig are functionally related (i.e., in the same
pathway).  If the coexistance of these genes is conserved (especially in a
module/neighborhood/operon/etc.), it lends evidence for functional linkages
between them.

I have found several tools/databases that support this this type of
investigation: STRING[1], Prolinks[2], MaGe[3], IMG[4], MetaCyc[5], KEGG[6],
and others.  However, all these resources are made from fully sequenced
genomes and do not seem to allow direct analysis of user-inputted sequences.
Does anybody know of other resources I should look into, especially ones
that do allow analysis of user-provided sequences?

If not, I can think of two general approaches to take:
1. Using the above tools, find the closest homolog/family/COG/etc. that
   they have for each of my genes, and assume their analysis of them would
   apply to my genes as well.
2. Write my own code to do profiling by using BLAST, etc. to 
   find homologs for each of my genes and "union" them per genome 
   (preferably taking chromosomal proximity into account if possible).

Any criticisms and or advice on these approaches or about any others I
should consider?

  1. http://string.embl.de/
  3. http://www.genoscope.cns.fr/agc/mage
  4. http://img.jgi.doe.gov
  5. http://metacyc.org
  6. http://www.genome.jp/kegg/
Dave Roe		roex0050 at umn.edu

More information about the BBB mailing list