View Single Post
Old 08-18-2010, 08:55 AM   #19
Senior Member
Location: Boston

Join Date: Feb 2008
Posts: 693

I gradually recall the decision on choosing the parameters for blat. My focus was more on >=500bp reads. And for these reads, blat -fastMap is similar to blat deault in accuracy but tens of times faster. However, for shorter reads which you are focusing on, blat default is much more accurate than blat -fastMap (still much slower, though). Your table would largely agree with mine for blat default.

Actually for 454, I would highly recommend ssaha2. Ssaha2 is designed for mapping sequencing data and calling SNPs from the first day and has been thoroughly validated. Blat, although being one of the best tools for mapping ESTs, is not for SNP finding initially and is not heavily evaluated. From what I have heard, blat does not refine the final alignment, which may make gaps positioned suboptimally and pose problems to indel finding. The default blat mode is also much slower and less accurate than ssaha2. In my view, it is a common mistake to overlook the superiority of ssaha2 for longer reads. The 1000 genomes project chooses every program for a reason.

Last edited by lh3; 08-18-2010 at 08:58 AM.
lh3 is offline   Reply With Quote