Reader Comments
Post a new comment on this article
Post Your Discussion Comment
Please follow our guidelines for comments and review our competing interests policy. Comments that do not conform to our guidelines will be promptly removed and the user account disabled. The following must be avoided:
- Remarks that could be interpreted as allegations of misconduct
- Unsupported assertions or statements
- Inflammatory or insulting language
Thank You!
Thank you for taking the time to flag this posting; we review flagged postings on a regular basis.
closeInteresting candidate gene generator
Posted by Heather on 19 Jun 2008 at 12:29 GMT
I would like to know if the authors validated the utility of the candidate gene list they derived. One approach could be to make an artificial (if small) list of what once were candidate loci for a number of conditions, before the causative gene ended up being identified, and then using their RCG approach to see if they can pick out the validated causative gene from these intervals.
Also, some loci are broader or better determined than others. Are the loci simply defined by the chromosomal band, or by the markers with highest LOD score, or LOD>3...? Were physical positions assigned to the band, and in the case of a borderline locus, perhaps an arbitrary distance into the neighboring band might be appropriate?
RE: Interesting candidate gene generator
provero replied to Heather on 24 Jun 2008 at 08:31 GMT
Thanks for the interesting comments.
In this work we limited the systematic evaluation of the performance of the algorithm to the predicted functional annotations and did not perform it for candidate disease genes. In a related paper whose main focus is instead disease gene identification (and which is based on phylogenetically conserved coexpression) we did perform an evaluation using a method similar to the one you propose: we constructed artificial loci around genes already known to cause diseases and evaluated the percentage of cases in which our algorithm was able to recover the correct gene, with satisfactory results. The paper is Ala et al., PLoS Comput Biol. 2008 Mar 28;4(3):e1000043.
Regarding the definition of loci: we simply used the chromosomal bands as given by OMIM. No doubt the use of LOD scores would improve the characterization of the loci, but we are not aware of databases in which these are systematically reported for all OMIM diseases, so their use in a phenome-wide study such as this one would be problematic. On the contrary, it is possible and certainly advisable to use a LOD-based definition of the loci when applying the method to specific cases.