[Clare72]

The scRNAseq data is moving, study by study, from the old FlyBase/SCEA loads (FBlc… ids) to the vfb-scRNAseq loads (VFBseq_… ids). Both id styles keep working, so this can merge before the switch.

The four transcriptomics queries need no change: the new files keep the old graph shape:

• Cluster / DataSet node labels;
• has_source, has_reference and composed_primarily_of edges;
• expresses edges with expression_level / expression_extent.

That holds provided the pipeline gives the new datasets the scRNAseq_DataSet / hasScRNAseq labels; see Outside this repo below.

Changes

• term_info_fallback: VFBseq_ Samples, Assays and Clusterings stay out of the anatomical-individual index, as their FBlc predecessors did. Datasets and Clusters are routed to their own indexers before this list is checked.
• combine namespace hints:
  • adds VFBseq_;
  • FBlc is now matched without an underscore. FBlc ids have none, so the old FBlc_ entry never matched.
• Transcriptomics tests (test_transcriptomics.py, test_query_performance.py):
  • each fixture keeps its FBlc id while that id still has data, and otherwise takes a live term that does (src/test/scrnaseq_fixtures.py);
  • cluster ids may start with FBlc or VFBseq_.
• Docs list both prefixes.

Tests

• tests/test_combine.py and src/test/test_term_info_fallback.py: 100 passed.
• Live src/test/test_transcriptomics.py (12 tests) and test_11_transcriptomics_queries: passed against production.
• With unknown ids, the fixture fallbacks return live terms.

Outside this repo (needed for the switch)

• vfb-pipeline-polishing finalStep.py: sets scRNAseq_DataSet / hasScRNAseq only on DataSets whose short_form starts with FBlc. VFBseq datasets need it too.
• VFB_json_schema_indexer: the anatomical-individual parameter query should exclude VFBseq_ as well, mirroring the change here.

🤖 Generated with Claude Code