Using a few of them and comparing the results would work I guess, thanks.
Hi, I am looking for tools to identify somatic mutations (SNPs, CNVs, indels, etc) from a tumour sample (RNA-seq) which has been aligned with the human reference genome using STAR. I noticed that there are many tools (MuTect, MuSE, VarDict, etc) for generating a vcf file. It looks like these tools require both reference mutations from dbSNP/COSMIC, and the aligned file for normal sample. On the other hand, samtools mpileup converts BAM to VCF without these requirements. So, is it recommended to go ahead with samtools or use the other tools?
Meanwhile, I am reading research papers about the use of Machine/Deep Learning Techniques for the identification of mutations. Is there any way I can incorporate these techniques into my analysis?
Thanks in advance.
1 answer
There are many published somatic variant callers but there is by best knowledge none that clearly and generally outperforms the others. Rather than that some may be preferred when depth is shallow or very high, some deal with contaminations better than others, well and some use ML-based approaches while others don't. Read the manuals and benchmarking papers and choose the one that seems to fit your data best. bcftools (former samtools mpileup) has a good reputation here at Biostars (from what I heard) for SNPs, while lumpy seems to be the jack-of-all-trades for structural variants.
Log in to answer this question.