finding mutations in rna-seq transcriptome

Question

[genya35]

Hello,

Please suggest bioinformatics tools to identify indel and point mutations in RNA-seq transcriptome.

Thanks

Answer 1

[jared.andrews07]

You can use same tools used for finding them in whole genome sequencing. Be aware that most of your reads are only coming from one strand the majority of the time with RNA-seq, so het/homozygous calls are a lot less reliable.

GATK is kind of the standard, and objectively one of the most accurate when used properly. It's also under continuous development and has tons of options. VarScan2 is also quite good, though it's no longer actively updated. Avoid the samtools/bcftools method that you'll see in a lot of tutorials - it does not do well with low frequency mutations and generally isn't as accurate.

[genya35]

What do I use to convert fast-q RNA-seq to bam format?

[grant.hovhannisyan]

Convert is not a correct word here:) You need to map the reads (fastq file) to the reference genome (or transcriptome), if available, to obtain the bam file. To do it you can use for example STAR or Hisat2 software.

[genya35]

Is it possible to use rna-star on usegalaxy.org to do the alignment. I don't see two-pass mode option? It's not clear on the website.

[jared.andrews07]

It seems the two-pass mode isn't available on Galaxy, but STAR isn't too hard to set up locally if you've got enough RAM. It's kind of a hog, so you want at least 32 GB for it to run.

[genya35]

@jared.andrews07 is VarScan2 same as VarScan version 2? Should I Split'N'Trim and reassign mapping qualities with SplitNCigarReads before running samtools to generate mpileup, and finally run VarScan in somatic mode? Thanks

[jared.andrews07]

Yes, they are the same, though I'd somewhat recommend GATK over it, particularly for somatic calling. I imagine following GATK's best practices probably wouldn't hurt, and yes, you'll have to run in somatic mode.

Answer 2

[grant.hovhannisyan]

As already mentioned, you can use GATK. But variant calling based on RNAseq data has its tricks, so better follow the best practices of GATK specifically for RNAseq.

[cpad0112]

@yelekley7 : example bash script (from start to end) for GATK best practices in variant calling for RNAseq is available here.

[genya35]

I forgot to mention, that I have both tumor and normal fastqs and looking for somatic mutations. Is GATK best practices still appropriate? Thanks

[cpad0112]

I guess you can. But replace the variant caller with tumor specific variant caller (better somatic variant callers such as mutect) and subsequent steps