[BioC] SemSim returns Lin similiarity greater than 1 (01-06-2009).

dyu at purdue.edu dyu at purdue.edu
Tue Jan 6 20:26:47 CET 2009

To whom it may concern,

I try to calculate Lin similarity for GO terms by SemSim package (published in
12/10/2007) and obtained some results greater than 1 since December, 2008. (e.g.
termSim("GO:0045944", "GO:0007329","Lin","yeast") returns 1.232)

I sent this problem to the author (xguo at jcvi.org) but the delivery was failed. 

After reading through the internal code, I find that the "get" function in
SemSim connects to "AnnotationDbi" (published in 11/22/2008) and SemSim depends
on the ".gocount" internal data which is produced before 12/10/2007. So, one of
the reasons to get Lin similarity greater than 1 might be that ".gocount" did
not count the common ancestors identified between 12/10/2007 and 11/22/2008.

I appreciate you very much for any kind help on solving such a problem.


Danni Yu,
Ph.D. student
Department of Statistics
Purdue University

More information about the Bioconductor mailing list