Skip to content

immutable=1 sqlite and a pool.join() - #377

Open
cariaso wants to merge 21 commits into
KarchinLab:masterfrom
cariaso:master
Open

cariaso wants to merge 21 commits into
KarchinLab:masterfrom
cariaso:master

Conversation

@cariaso

@cariaso cariaso commented Nov 9, 2025 •

Copy link
Copy Markdown

2 unrelated fixes.

immutable seems to help my performance, a lot.

pool.join() fixes hanging processes

@cariaso cariaso changed the title immutable=1 immutable=1 sqlite and a pool.join() Nov 23, 2025
@kmoad
kmoad force-pushed the master branch 2 times, most recently from f49e414 to 93d71bd Compare March 11, 2026 16:36
…input alt

Converter.liftover() only handled a strand flip (target reference equal to the
reverse complement of the input ref). When the assemblies simply disagree about
which allele is the reference, e.g. F5 rs6025 (hg19 ref T / hg38 ref C), the
hg19 ref/alt were passed through unchanged, so the variant was mapped and
annotated as a different (synonymous) change and the real one was missed.

Add orient_alleles_to_target_ref() covering same-strand and opposite-strand
swaps, and have liftover() return a swapped flag. When a converter supplies a
homozygous zygosity and the alleles were swapped, the sample is homozygous for
the new reference, so the row is dropped as a non-variant.
…verters

- orient_alleles_to_target_ref takes the chain strand for single-base lifts, which
  resolves the A/T and C/G strand-flip vs swap ambiguity.
- zygosity_after_liftover reinterprets converter zygosity after a swap: hom becomes
  a non-variant, and a new 'ref' value (homozygous for the input assembly's
  reference allele, emitted by converters at sites where the assemblies disagree)
  becomes hom. Missing zygosity is left alone.
… homozygous for an assembly-specific reference

Microarray converters only emit alleles that differ from the input assembly's
reference, so a call homozygous for that reference is never reported even where
the liftover target assembly has a different reference base (F5 rs6025 T/T:
homozygous Factor V Leiden). Provide a shared helper that returns a zygosity
'ref' row for those sites; the converter loop turns it into a hom variant once
liftover has swapped the alleles.

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant