Should you retain all the alignments after primer clipping using Katana for variant calling or just consider the clipped ones?

Question

[bioinfo89]

Hello All,

I am working with amplicon sequencing data for ~200 samples. For the same, I did soft clipping of gene specific primers using Katana.

Katana provides clipped.bam file which only includes the alignments where it finds the matching primer pairs.

If I use --preserve_all_alignments parameter, I get a BAM file which also includes the non matching ones (X4:Z:UNMATCHED_PRIMER_PAIR, X4:Z:INVALID_CIGAR).

I wanted to know which type of clipped BAM file I should use further for variant calling. Any suggestions will be appreciated.

Thanks!