valency-anndata

To analyze Polis conversation data using AnnData, this skill orchestrates an end-to-end workflow that loads vote matrices, preprocesses the data, runs recipe_polis, and generates embedding-based visualizations. It structures the output into standard AnnData fields like X, obs, and var.

For clustering Polis participants, you can compute PaCMAP, LocalMAP, UMAP, or t-SNE projections. The workflow applies per-embedding clustering to these bases and generates corresponding perspective plots for visual comparison.

Yes, the workflow fully satisfies AnnData structure requirements by mapping Polis data into X, obs, var, layers, obsm, and uns fields. It utilizes API namespaces for datasets, preprocessing, tools, and viz to ensure comprehensive data handling.

To run the recipe_polis pipeline, invoke the skill and follow the prompts to select your desired projection maps. The pipeline automatically calculates QC metrics, preprocesses the vote matrices, and generates embeddings for downstream analysis.

The best way to visualize Polis embeddings with different projection bases is to compute PaCMAP, UMAP, or t-SNE maps and generate per-embedding clustering plots. This approach allows you to directly compare participant groupings across various projection bases.