aDNA graph construction and MAPQ defaults

[CormacKinsella]

Graph construction aDNA

• current best is k 15, w 5 k 15, w 3. Here we underperform vs bwa aln only at 30 bp, with good performance at 40+ bp

• TODO on this: go further to optimise 30bp, without sacrificing accuracy across all lengths, or increasing run time to dealbreaking levels

Graph MAPQ default

• Set to 30

• While MAPQ 20 does very well on accuracy for endogenous content, we get a little too much microbial contamination mapping

• TODO: Thus we need proper, benchmarked filtering via kraken2 prefiltering #34. The aim would be to enable a lower MAPQ default of 25 (20 at a reach of if we can optimise)