iqbal-lab
commented
2 years ago Suggest you turn a catalogue into three catalogues
Part A: nucleotide matches Use VCF definition of catalogue, and match Drprg vcf with catalogue vcf. Some trickiness with indel normalisation.
Part B: amino variants First translate the pandora-inferred sequence into amino sequence Then,
- catalogue has two dictionaries. Both have key which is amino position /index in the reference amino sequence. First dictionary has value which is array of amino acids which cause resistance. Second dictionary has value which is array of amino acids which definitely don't cause resistance. WHO and others starting to provide these "definitely not a resistance variant" lists, could be useful to report.
Finally notice if there is
- stop codon
- Part of the Pandora sequence has zero coverage, signifying truncated gene
Part C
- rule about whether frameshifts or stop codon or truncations in a gene are meant to cause resistance. Frameshift detection by aligning amino sequence seems most reliable as you incorporate all variants at the same time. Otherwise you have multiple nucleotide positions nearby affecting the same codon
The
argmatchfunction currentlty does what it is supposed to; it returns whether a panel variant matches with a variant in the pandora VCF. But there can be false positives which lead to FP resistance calls. For example(I've removed some VARIDs for brevity)
This variant simplifies to
The two variants
gid_A205Eandgid_A205*do overlap this VCF record, but they're not technically a match.This position, 715, maps to codon 205, which in the reference is
GCA. So this VCF position is the last base in the codon.Changing the last base to
G, as this record calls, would make the codonGCA->GCGwhich in amino space isA->A- i.e., synonymous. However, whatargmatchdoes (by design) is look in the panel VCF and see that aGat this position matches with those other two variants, which it does, but it's complicated....Here are the panel VCF records for those two variants
Extracting just position 715 from these variants does indeed provide
A->Gas an option. But it ignores the fact that you would also need to change positions 713 and/or 714, which our running example at the top does not.I need to think about the cleanest way to handle this... Any thoughts are most welcome though.