dnanexus-rnd / GLnexus / 1710

93%

push

travis-ci

Unify some non-left-aligned indels (#128)

Unifier recognizes equivalent indels represented with different positions (i.e. at least one version isn't left-aligned), when the representations are overlapping, abutting, or spanned by some third allele. They are collapsed into the leftmost representation observed. The genotyper is adjusted slightly to ensure it retrieves such non-aligned indel records from the database, as their positions might not even overlap the unified site.

Limitations:

1. Doesn't recognize non-left-aligned indels other than the aforementioned cases. Theoretically, equivalent representations can be far apart (in homopolymer or tandem repeats) which is difficult to handle at this stage.
2. Doesn't necessarily produce the most left-aligned possible representation; rather, collapses into the most left-aligned version actually seen.

So, we'll continue to strongly recommend the upstream pipeline produce gVCF with left-aligned variants.

95 of 95 new or added lines in 3 files covered. (100.0%)

4110 of 4448 relevant lines covered (92.4%)

251820.54 hits per line