Does anyone know if BWA Mem -M flag (mark shorter split hits as secondary) will have a detrimental effect on structural variation calling using split reads with Delly?
I orignially added it to my BWA mapping pipeline for compatibility with Picard/GATK but I believe that issue has been resolved in GATK4. Now wondering if it's even necessary and if it will have an effect on SV calls that use split reads.
Input to this pipeline may include paired, single end, interleaved or split data from illumina, ion torrent, or nanopore.
2 answers
I've never used -M option for structural variant calling. Based on this BWA thread, you may lose supplementary mapping information by using -M option, because -M may label split reads as duplicates and strip the secondary/supplementary mapping information. I believe at least some SV tools use split read information for variant calling.
https://github.com/lh3/bwa/issues/163
The best way might be to try running a small experiment of calling SVs with and without -M option to see the differences.
Log in to answer this question.