vdjmatch 0.0.1

First PyPI release of the Python rewrite of vdjmatch (built on the seqtree search core).

Single-chain annotator (Phase A):

• Fetch/cache the latest VDJdb release; parse to a canonical schema.
• AIRR rearrangement + paired (cell/TCRvdb) readers via polars.
• Fuzzy CDR3 search (seqtree) with the bundled TCR-specific VDJAM matrix; ranked hits with CIGAR + alignment + edit counts.
• Control-calibrated E-values (BLAST-style significance vs a background repertoire).
• Per-query best epitope call + epitope-level enrichment summary; pairwise sample overlap.
• CLI: vdjmatch update, vdjmatch match.

The legacy Java/Groovy tool is preserved on the legacy-java branch (tags 1.1.4–1.3.1).

Paired α/β scoring, a re-derived segment-aware VDJAM, and tool comparisons are in progress (see roadmap-iter-1.md).

1.3.1

Update VDJtools

1.2.2

License upd

1.2.1

jitpack tag

VDJmatch algorithm

Implemented TCR similarity metric and smart algorithms for VDJdb querying.

Java 9 compatibility

• Now should work with Java 9
• Other minor fixes

N.B Please ignore An illegal reflective access operation has occurred warnings occurring when running with Java 9, they have no effect on software performance/output whatsoever

Server compatibility release

1.1.6

Upd sh script

Better summary stats

Minor fixes, update VDJtools version, better annotation summary output

Cleanup release

Removed some experimental functions & fixed sever bugs

Scoring update

Critical fix for scoring system:

• Re-named and modified presets: hamming (hamming distance based); 5,2,2,7 search scope followed by high-recall (precision corresponding to recall of 80%), optimal (precision and recall corresponding to highest F1 measure)
• Scoring selection by precision and recall (selects nearest pre-trained preset matching specified value)
• Ensure that exact matches are always reported disregarding score threshold