EMA split-read handling

[pdimens]

Hello,

I use haplotagging data with the EdHarry ema fork and have been trying to use the LEVIATHAN variant caller, which relies on split read information to call variants. My team noticed the ema bam files dont have any flags for supplementary (split-read) alignments (2048 or 0x800), whereas mapping with bwa did.

In the README, you propose mapping the noBC fastq using bwa mem -M, which looks like it changes the supplementary mapping flag to a secondary flag instead, so maybe the -M (or equivalent) is baked into ema? Is there a way for us to retrain this split-read information for use in LEVIATHAN?