TMAP versus other aligners

[quantrix]

Hi there,
I am working on some new data from Ion Proton and PGM as well. I was wondering if anyone had any experience comparing the alignment metrics for TMAP aligners (Company provided) versus other aligners? I am mainly interested in BWA, Bowtie2, Novoalign and SOAP2. Initial estimates show that nearly 95% of reads are mapped using TMAP versus SOAP2 which seems to aling only 35% of the reads (eek!). I am beginning to wonder if it is even worthwhile to spend time trying to setup an alternative pipeline and instead just use the company provided one?

I would be glad to hear other people's experiences. Can I tune the parameters on the other aligners to bring it up to 95% + mapping?

Thanks in advance

[MrGuy]

TMAP works very well. Bowtie pretty bad. CLC (commercial) worked just as good as TMAP. Mosaic worked fine, too. I generally find if you follow the manufacturers recommendations things seem to work pretty well. Our strategy is looking to be TMAP for on "torrent" server analysis and CLC for off server custom analysis. I haven't really tested anything more since everything is working just fine and compares well with previous work we've done.

[scarl]

I can agree that Bowtie works pretty badly. I think that in general, aligners that are designed for short, fixed-length reads (e.g. Illumina-type reads), like Bowtie, are going to have problems with Ion Torrent data. I haven't done any extensive testing, but I suspect that a major part of the issue is how they assign penalties for mismatches and that sort of thing. I imagine that you'd have similar problems with bwa for the same reason.

I've had good success with smalt (http://www.sanger.ac.uk/resources/software/smalt/) for reads from a PGM, and a colleague says he's gotten good alignment by setting up a local blast instance and using that (though this may not be the most time-efficient method). Overall, though, I've been pretty happy with TMAP's performance, so I haven't spent a lot of time looking for alternative aligners.

[oiiio]

There is a free website tool called GCAT (Genome Comparison and Analytic Testing) that is designed to answer these types of questions without qualitative speculation.

For example, to compare TMAP pipeline variant calls on Ion Torrent exome data versus other pipelines: http://www.bioplanet.com/gcat/report...x/tmap-gatk_hc

If you want to see the raw ability of the aligner using read simulations, here is an example of that type of report: http://www.bioplanet.com/gcat/report...l-indel/bowtie

You can also download any of the data sets as FASTQs, then upload your BAM or VCF to the app for the same kind of privately viewable report to compare. There is also a forum for questions as well

[quantrix]

Thanks all! I did see Bioplanet data yesterday and was planning to check it out. But the rest of the tips are useful as well. I will post once I am done with the analysis my findings.