v2.3.0

Phasing analysis updates

• added phasing threshold: superclusters are only considered phased if one phasing is an X% improvement over the other in terms of edit distance (this reduces false positive supercluster phasing flip errors that are actually variant calling errors), default 0.6
• added phasing summary TSV (phasing-summary.tsv) that reports total flip errors, switch errors, phaseblock NG50, switch NGC50, and switchflip NGC50
• add switchflip TSV (switchflips.tsv) that reports flip range, type, supercluster, and phase block
• phase blocks are now computed from input phase sets, not backtracking, and per-phaseblock switch/flip errors were added to phase-blocks.tsv

Partial credit replaced with credit threshold

• partial credit calculation is less intuitive and complicates matters more than necessary; I replaced this with a partial credit threshold where passing variants are counted as TP, default 0.7
• I think that counting mostly-correct calls with a user-defined credit threshold is better

Runtime improvements: skip alignment distance and writing

• alignment distance calculation is now skipped by default (I now think stratifying precision-recall curves by INDEL size may be more useful), can be turned on with -d, --distance
• original and realigned truth/query VCFs are only written if --realign selected

Added new analyses

• added analysis-v2 directory for upcoming paper figures

Minor fixes

• GA4GH output VCF no longer always outputs gm: now it uses gm for TP, lm for PP, and . for FP/FN