Sometimes we'll encounter entries which have a $gt_str which has
characters not contained in the $ids map. This results in malformed gt
entries, like /0 or 1/ or /. When these entries occur,
downstream calls like merge_vcf (in the case of immuno workflow) will
print an error and silently continue on their merry way. I have no
idea why they do this, but it's something we noticed when testing
conversion over to WDL with HCC1395 sample.
We address this issue by checking the results of @gt_ids, and if any
are undefined we eject early, returning undefined. At the call site we
handle this case separate by printing to STDERR, dropping that line,
and moving on to the remainder.
An alternative approach of a more strict failure is an option, but
discussion with Chris Miller and Malachi Griffith landed on decision
that we just drop and move on.
Sometimes we'll encounter entries which have a $gt_str which has characters not contained in the $ids map. This results in malformed gt entries, like
/0or1/or/. When these entries occur, downstream calls like merge_vcf (in the case of immuno workflow) will print an error and silently continue on their merry way. I have no idea why they do this, but it's something we noticed when testing conversion over to WDL with HCC1395 sample.We address this issue by checking the results of @gt_ids, and if any are undefined we eject early, returning undefined. At the call site we handle this case separate by printing to STDERR, dropping that line, and moving on to the remainder.
An alternative approach of a more strict failure is an option, but discussion with Chris Miller and Malachi Griffith landed on decision that we just drop and move on.