v2.3.1.4

What's Changed

• skani updates by @dpark01 in #44
  • First, this changes all skani invocations to use -m 15 -c 10 as the new default parameters for the marker k-mer compression factor (-m) and the k-mer subsampling rate (-c). This is down from our previous default values of -m 30 -c 20 which is in turn lower than the skani-recommended settings for viruses of -m 200 -c 30 which is lower than the skani bacterial-inspired default values of -m 1000 -c 125. The new default values were determined through empirical testing of diverse viral taxa, and -m 15 -c 10 succeeds in clustering (finding non-zero ANI between) all of rhinovirus & enterovirus together as well as all of Lassa virus (our old default values only partially succeeded on these taxa).
  • Second, this PR reformats skani TSV output to be written out in sorted order of descending ANI * Total_bases_covered (instead of the default sort order of descending ANI). This prevents our reference selection code from favoring reference genomes with higher ANI for short stretches of sequence (which happened with some rhino/enteros) and now favors longer matches of high identity. This specific metric (the product of ANI and match length) is inspired by ReferenceSeeker (publication, GitHub). Unit tests added to confirm proper sort order.

Full Changelog: v2.3.1.3...v2.3.1.4