Which Snp Caller / Method To Use After Aligning Rna-Seq With Tophat

Question

[William]

Which SNP caller / method can / should I use after aligning RNA-seq data with TopHat?

For genomic data I use GATK, but supposedly it is not just as easy as running GATK on the TopHat RNA-seq data. The team from Broad has no information / documentation on how to use GATK for RNA-seq data.

I don't have any variants yet from DNA re-sequencing.

Answer 1

[Chris Miller]

VarScan will happily make calls on RNAseq data with a matched normal (from RNAseq or exome). Be sure to consider all of the parameters carefully, especially minimum coverage levels and such.

[Nicolas Rosewick]

Do you have some advices for strand-specific RNA-Seq (2x50bp) ?

[Chris Miller]

Don't have a lot of experience with this, but some tools have steps with a directional filter (variants only appearing in one strand are removed as artifacts). Make sure that any such option is off.

Answer 2

[yyaobo]

I haven't looked at GATK on RNA-seq data, instead I used samtools/BCFtools, it's probably not the best, but it will get you started.

Answer 3

[jockbanan]

Current GATK (2.8-1) seems not to be a good option for reads mapped with TopHat (and any other splice-aware aligners). When running on .bam file generated by TopHat, you end up with an error saying you should remove (or ignore) all reads with 'N' cigar string (spliced reads). You can force GATK to process these reads, but the authors say that this may result in something not really meaningful.