Benchmarking TB: build and quasimap

iqbal-lab-org / gramtools

Genome inference from a population reference graph

MIT License

92 stars 15 forks source link

Benchmarking TB: build and quasimap #132

Closed ffranr closed 4 years ago

ffranr commented 6 years ago

SNPs and small indels above frequency 0.01.
Combine the VCFs of 500 M. tuberculosis samples
genotype those 500 samples at the duplicated site-list
get a single VCF
exclude SNP/indels below 1% frequency.

bricoletc commented 4 years ago

I'm closing this issue as we have demonstrated build and quasimap works on 3K (Kerri) and 50K (CRyPTIC) samples in TB, using minos.

iqbal-lab commented 4 years ago

kerri''s 3k and the 50k dont ever build a whole genome prg with that many genomes do they?

bricoletc commented 4 years ago

true, we have found in both cases that the high amount of variation means we had to split the whole genome VCF for all samples and make and process PRGs of those splits.

In the future we will have such a TB variant dense dataset to benchmark changes to gramtools.