openvax / pyensembl / 25810373798

85%

master: 83%

Pull #356

github

Fix #335 (part 1): wire GENCODE_BIOTYPE_ALIASES into read_gtf call

GENCODE GTFs use gene_type / transcript_type where Ensembl uses
gene_biotype / transcript_biotype. Before this fix, pointing pyensembl
at a vanilla GENCODE GTF produced a sqlite database with empty biotype
columns, so Transcript.is_protein_coding returned False for every
transcript and downstream tools like varcode reported NoncodingTranscript
for every variant effect (the original #335 user repro).

gtfparse 2.7.0 introduced an attribute_aliases kwarg on read_gtf that
renames source columns onto canonical names at parse time, plus a
GENCODE_BIOTYPE_ALIASES constant for the two GENCODE->Ensembl renames.
pyensembl now passes that constant unconditionally:

* Pure Ensembl GTFs: alias sources don't exist, the rename is a no-op
  (gtfparse skips missing alias source columns).
* GENCODE GTFs: gene_type / transcript_type get renamed to
  gene_biotype / transcript_biotype before they hit the sqlite schema.
* Mixed (both column variants present): canonical wins, gtfparse warns.

Bumps gtfparse dep floor to >=2.7.0.

Companion to #350 / v2.9.6 (versioned protein-ID FASTA matching). With
both shipped, the original #335 GENCODE genome example finally produces
Substitution effects instead of NoncodingTranscript.

Bump version to 2.9.8.

2 of 2 new or added lines in 2 files covered. (100.0%)

1600 of 1883 relevant lines covered (84.97%)

3.4 hits per line