Release 1.7 - Githubissues

iqbal-lab-org / gramtools

Genome inference from a population reference graph

MIT License

92 stars 15 forks source link

Release 1.7 #155

Closed bricoletc closed 3 years ago

bricoletc commented 4 years ago

As a sanity check and as for release 1.6, I ran 31eec32 on the Pf trio dataset: 20 samples on chromosomes 1 and 2, evaluating Mendelian error rates of gramtools multi-sample pipeline

Previous release 1.6, cf5cc20:

MendErr_rate_all

Proposed release, 31eec32: MendErr_rate_all

Main differences:

More calls in new release, probably due to allowing adjacent variant records
Slightly higher error rate without filter, but then same or lower past the first filter bin
More calls at higher GT_CONFs, probably again due to alleles not being combined

bricoletc commented 4 years ago

Logging: c4f21b8 which has breaking changes (from earlier commits), produces same results. Breaking change is that AlleleIDs now stored 0-based and -1 is used for ALLELE_UNKNOWN so build kmer indexes and coverage_Graph are different.