Mike Lin

Nov-2025 embedding pipeline updates

1

Minor updates to the embedding pipelines in Nov-2025: - update to transcriptformer 0.6.0 - `census_contrib`: - update dependencies - adjust `roundHalfToEven` typing for numpy 2 - tune TileDB consolidation settings

Our regular procedure for generating Census scVI embeddings trains the scVI model with an up-to-date HVG h5ad, but doesn't revisit the [hyperparameters](https://github.com/chanzuckerberg/cellxgene-census/blob/main/tools/models/scvi/scvi-config.yaml) set a couple of years ago, since which...

[python] builder: validate uniqueness in census_info organisms table

1

Regression test for #1439

Follow-on to #1440 The builder has two different source files named `census_summary.py`. One computes the `census["census_info"]["summary"]` table stored in the tiledbsoma artifacts, and the other computes slightly different statistics shown...

These two procedures seem very similar, maybe one can use the other or we can factor out some shared core.

Configure instructions for this repository as documented in [Best practices for Copilot coding agent in your repository](https://gh.io/copilot-coding-agent-tips).