SWAN-MPO machine-readable supplementary tables

This archive contains the CSV files corresponding to Supplementary Tables S1–S23 described in the Supplementary Information, together with Supplementary_Table_File_Manifest.csv containing row counts, column counts, and SHA-256 checksums for the table CSV files.

The retrospective boundary-stress tables use the frozen 15-compound benchmark throughout. File names correspond directly to the table descriptions in the Supplementary Information.

For ChEMBL-derived ranking metrics with exact score ties, BEDROC and enrichment factors are evaluated independently of input row order by averaging the expected contribution across positions within each tied score block. Table S13A was regenerated from Table S13B using this tie-aware policy.

Table S3 contains the full 10,620-task expanded-library deployment ledger. Per-grid affinity records are retained for 10,241 tasks and runtime metadata for 8,961 tasks. Complete 20-grid target-level summaries are preserved for all 531 ligand-target blocks in Table S17D and in the archived reproduction inputs; missing ledger-level metadata therefore do not imply missing target-level scoring summaries.
