To borrow code lines for seqr loader automated validation.
To perform basic concordance, use the functions defined in the beginning:
extract NA12878 from the mt,
run hl.split_multi(mt) on both mt and the truth mt table,
run filter_highconf(mt) to filter out LCR on each matrix table,
call hl.concordance(mt, truth_mt) to generate concordance matrix,
and parse the result to calculate metrics (print_conc_summary)
Most of everything else in the notebook is attempt to perform more fancy binned concordance, to plot different VQSR score bins and figure out the best VQSLOD threshold.
To borrow code lines for seqr loader automated validation.
To perform basic concordance, use the functions defined in the beginning:
hl.split_multi(mt)on both mt and the truth mt table,filter_highconf(mt)to filter out LCR on each matrix table,hl.concordance(mt, truth_mt)to generate concordance matrix,print_conc_summary)Most of everything else in the notebook is attempt to perform more fancy binned concordance, to plot different VQSR score bins and figure out the best VQSLOD threshold.