PAV-aware evidence tiers for UK Biobank Pharma Proteomics Project plasma cis-pQTLs: resource, code and analysis plans (v1.3.1)

v1.3.1 corrects an outdated statement about platform concordance and MS support in the README and field dictionary; data values unchanged. In the two independent MS cohorts (Generation Scotland, HOLBAEK), platform concordance was positively associated with MS support (combined OR 2.92); the association was weak and not significant only in the two Seer-study layers. The resource TSV and Parquet are byte-identical to v1.3. This archive accompanies the paper Protein-altering variants inflate cross-platform replication of plasma protein QTLs (Ni, Zhang and Zhang; in submission). What it contains. A table of all 14,472 cis-pQTL credible sets (1,879 genes) from the UK Biobank Pharma Proteomics Project (UKB-PPP) Olink study, as fine-mapped in Open Targets Platform 26.09. Each credible set is annotated with: whether it contains a protein-altering variant (PAV) of the measured protein's gene; whether it colocalises with an eQTL for that gene (transcript support); replication on SomaScan (Sun et al. 2018; deCODE, ARIC and Fenland); mass-spectrometry (MS) labels from four layers: MSPA, Seer full summary statistics, Generation Scotland and HOLBAEK; replication in non-European UKB-PPP ancestries. Tiers. The four evidence tiers combine the PAV and transcript-support annotations by fixed rules from a pre-specified analysis plan. They are annotations, not a validated reliability ranking: transcript support did not predict MS confirmation. Credible sets on chromosome X have no MS data, because no MS layer releases chrX summary statistics. Platform concordance. Concordance across affinity platforms is higher for PAV-containing signals, so for these signals it is not orthogonal validation. Overall, concordance was positively associated with MS support in Generation Scotland and HOLBAEK (combined OR 2.92) but only weakly, and not significantly, in the two Seer layers. Files. Resource v2.3 (TSV and Parquet; 101 fields), its README and field dictionaries (English and Chinese), the analysis and figure code (MIT), the analysis plans (main plan and addenda A1–A4, each with its SHA-256 checksum), the resource package as one archive (C10plus_evidence_tiers_resource_v2.3.tar.gz) and CHECKSUMS.sha256 for every file. What changed in v1.3. Resource v2.3 adds 25 fields from addendum A4 (Generation Scotland and HOLBAEK MS layers); the 76 v2.2 fields are unchanged. The release also adds the A4 code, post hoc analyses L1/L2 and an English field dictionary. Online lookup portal: https://wcs52113140123-pixel.github.io/c10plus-pqtl-evidence-tiers/ Code repository: https://github.com/wcs52113140123-pixel/c10plus-pqtl-evidence-tiers Data use. Only derived per-credible-set statistics are included; no third-party summary-statistic files are redistributed. Please also cite the original data sources listed in DATA_LICENSE.md (inside the code archive) and in the resource README.