Limiting variant calls to amplicon target regions?

Question

[drollix]

What's the best way to limit variant calls from amplicon sequencing to desired target regions: Fastq trimming, soft clipping bam files or filtering vcf files?

• Fastq trimming with amplicon primer sequences can take a long time (for large panels like CCP) and requires the primer sequences
• Are there any tools that can soft clip bam files based on target regions?
• Filtering vcf files seems to be the easiest option, but I imagine that upstream modifications will have an influence on downstream variant calls.

This seems like it should be a solved problem, but I couldn't find a clear solution in my internet/biostars search.

Thanks!

[Pierre Lindenbaum]

you meant you want to 'cut-out' the reads (soft-clipping) if they go beyound the amplified region ?

[drollix]

Either trim the reads, soft clip the alignments in bam file or filter out the variants based on the target region. Looking for standard procedure for filtering out possible false positive variant calls at the edge of amplicon target regions due the the presence of residual primer sequences.

[stephen.r.atkinson]

Did you ever find an answer to this question? I am facing the same issue. The Novoalign tool has an option to do this but is only available in the commercially licensed version...

Answer 1

[Pierre Lindenbaum]

I just wrote a tool to clip the reads in a BAM according to a collection of PCR fragments: https://github.com/lindenb/jvarkit/wiki/PcrClipReads

Example:

echo  "seq2\t1100\t1200" > test.bed
java -jar dist-1.133/pcrclipreads.jar -B test.bed -b  samtools-0.1.19/examples/ex1.bam  |\
    samtools  view -q 1 -F 4 -bu  -  |\
    samtools  sort - clipped && samtools index clipped.bam

samtools tview -p seq2:1100  clipped.bam  samtools-0.1.19/examples/ex1.fa

Answer 2

[Tommy Au]

Primer trimming at FASTQ level (1) is computationally expensive, (2) incorrectly handles nested PCR amplicons, (3) makes indels harder to detect by conventional variant calling. BAMClipper is demonstrated to remove PCR primers by soft-clipping BAM alignments (Scientific Reports 7:1567), solving these challenges.

Answer 3

[Pierre Lindenbaum]

update 2021: there is now samtools ampliconclip samtools-ampliconclip http://www.htslib.org/doc/samtools-ampliconclip.html

Clip reads in a SAM compatible file based on data from a BED file.

Answer 4

[Jorjial]

From my experience, the best way to do it is to trim the primers after alignment. In this way, the amplicon primers help the aligner to align well the indels at the edge of the amplicons. After the alignment, you can trim the primers if you know the sequence or the start and end position in the genome. Here is a paper that can help you.

I hope this helps.

[drollix]

Thanks, the same paper motivated my question. Is there a tool that can soft clip/trim alignments using the interval list of primers (or target regions)?. In the paper, they refer to custom scripts and my attempts to modify CIGAR, MD tags etc for soft clipping have been futile so far.

[Jorjial]

I do not know any tool. I guess you can try Pierre Lindenbaum's one.

Answer 5

[User 59]

Most genotypers will allow you to supply a bed formatted 'regions of interest' file so that you only call variants in that region. A little simpler than your proposed methods I think.

[Jorjial]

This could be a quick solution. But, when you have a region covered by several amplicons, if you do not trim the primers, the variant fraction of the variants overlapping with primers can be underestimated and you can have some false negatives. If you trim the primers, the variant fraction will be the correct one.

[User 59]

Right, this is why hybridisation capture is preferable in my eyes! I didn't fathom that was the requirement from the question!

[drollix]

Yes, that is the issue I am facing. Trimming lots of primers from all reads is a very expensive process.

Answer 6

[George Carvalho]

To solve the soft-clip problem after the alignment I using Katana, u cold try it.

https://github.com/umich-brcf-bioinf/Katana