Add a global phosphosite reference builder

[jjacobson95]

Pipeline Hardening/Debugging PR # 6

Add a global phosphosite reference builder

Adds a new reference builder that produces phosphosites.csv, a master list of human phosphorylation sites. It works like the existing gene reference (genes.csv): dataset omics pipelines use it to map raw phosphoproteomics measurements to stable, consistent site identifiers. It is currently used by the cNF dataset, but is built as a standalone reference so any future dataset can use it.

What it builds (00-buildPhosphositeFile.py, 589 lines)

• Merges phosphosites from three sources in priority order, so a site already defined by a higher-priority source is never overwritten by a lower one (de-duplicated on the site identifier):
  1. Ochoa et al. (2020), Nat Biotechnol 38:365-373, Supplementary Table S3: ~116k curated human phosphosites (downloaded from Springer).
  2. UniProt PTM annotations via the UniProt REST API: ~12k additional sites not present in Ochoa.
  3. Synapse supplement(s): raw phosphoproteomics files on Synapse (default syn70078415, the cNF raw phospho data): ~188 additional experiment-specific sites not yet in either database.
• Looks up gene symbols for UniProt accessions through the UniProt REST API, and maps to Entrez IDs using the project genes.csv.

Stable identifiers across builds

• Each site gets a phosphosite_id. Passing the previous run's file with --prev preserves existing IDs so downstream identifiers stay stable from build to build.
• Output columns: phosphosite_id, entrez_id, gene_symbol, residue, position, modification, other_id. The other_id column stores the human-readable site string (for example AAAS-S495s) and is the join key for raw phospho data.

Configuration and packaging

• Source URLs, the Synapse supplement IDs, the site-column name, and the previous-file path are all command-line options, so sources can be updated without code changes.
• Adds the container image (Dockerfile.phosphosites), the build entrypoint (build_phosphosites.sh), and a README.md documenting the sources, site notation, and output columns.

Scope: 4 files, all new. Base: genes-ensembl-resilience. The genes step must run before this so gene-to-Entrez mapping is available.

[sgosline]

I thought we weren't doing phosphosites this build?