Thread: BFAST published
View Single Post
Old 11-20-2009, 02:43 AM   #9
Michael.James.Clark
Senior Member
 
Location: Palo Alto

Join Date: Apr 2009
Posts: 213
Default

Congratulations, Nils!

Quote:
Originally Posted by lh3 View Post
I did not read into details, but something is a little bit odd in Figure 3. How can bowtie and bwa produce wrong alignments even if the read can be perfectly mapped? For perfectly aligned reads, both of them should always give correct results. Also note that 6% error is really a lot. If an aligner was that bad, it would not survive these days. Nonetheless, I agree that the performance of bwa degrades with higher error/mutation rate.
It makes sense to me that they would have higher false positive rates at zero errors because they identify significantly more true positives as well (as 3A is basically gauging true positives and 3B is gauging false positives). My guess is that they are aligning questionable reads that the other aligners are not aligning at all in order to have a higher yield of true positives. I would pose the question of whether one can get BWA and Bowtie to generate fewer false positives, and how that would thereby affect their true positive yield at zero errors. (I have not used them personally, so it's an open question.) Then again, I think it's rather a moot point. The figure is basically demonstrating to me that all of these aligners do quite well at zero-two nucleotide errors.

What I find more striking in that particular figure is how they perform with high error reads, but even more strikingly, reads containing deletions (even deletions of a relatively large size). This is important to me because a lot of studies, particularly whole genome, are searching for novel variants and deletions with relatively low coverage.

Anyway, we have had great success using BFAST on whole genome data from the SOLiD platform, so I am really happy to see it finally published and encourage others to try it.
Michael.James.Clark is offline   Reply With Quote