PPIA Molecular Evidence Atlas: Literature Corpus, Evidence Vectors, and Mechanistic Adjudications
Description
This dataset provides an author-curated molecular evidence atlas of peptidyl-prolyl isomerase A (PPIA/Cyclophilin A) biology and pharmacology. It was developed to support a critical mechanistic review and to make the underlying literature corpus, screening decisions, evidence-depth labels, mechanistic classifications, and claim-level adjudications transparent and reusable. The dataset contains 3,631 candidate records, including 1,178 retained original-study records and 81 core reviews. It also includes 266 reference-gene or sample-quality methodology records and 183 non-mammalian or microbial PpiA/CypA homolog records. Verification depth is recorded as Q0 (full-text or claim-level audit; n = 32), Q1 (abstract-level evidence extraction; n = 1,129), or Q3 (title-level or information-insufficient evidence; n = 17). Mechanistic evidence is classified using four independent, non-additive axes: directness of the PPIA–object relationship (D), target-specific peptidyl-prolyl cis/trans isomerase evidence (C), functional causality (F), and in vivo or clinical validation (V). Thirteen manuscript-defining mechanistic objects, pharmacology cases, and evidence-boundary definitions underwent focused author-team adjudication. This resource is an author-curated critical-review dataset and should not be interpreted as a PRISMA-compliant systematic review or a whole-corpus dual-review consensus dataset. The deposited files contain bibliographic metadata, public PubMed and DOI links, author-generated evidence summaries, scoring rules, and classification records. They do not contain article PDFs, reproduced figures or tables, full article text, patient-level data, personal contact information, or confidential experimental data.
Files
Steps to reproduce
The dataset was constructed from a PubMed-scoped literature corpus focused on human and mammalian PPIA/Cyclophilin A biology and pathogen studies in which host PPIA was the relevant species. Candidate records were identified through title- and abstract-based searches and were assigned to the original-study corpus, review collections, methodology appendix, homolog appendix, or exclusion categories according to predefined scope rules. Evidence-verification depth was recorded as Q0, Q1, or Q3. Q0 denotes full-text or claim-level source verification; Q1 denotes abstract-level evidence extraction; and Q3 denotes title-level assessment or insufficient information requiring additional verification. Selected mechanistically informative objects were evaluated using four independent and non-additive evidence axes: directness of the PPIA–object relationship (D), target-specific PPIase catalysis (C), functional causality (F), and in vivo or clinical validation (V). Each axis was scored from 0 to 3 according to the definitions provided in the Evidence_scoring and Record_status_dictionary worksheets. Thirteen manuscript-defining mechanistic objects, pharmacology cases, and evidence-boundary definitions underwent focused author-team adjudication. Final classifications and explicit inference boundaries are recorded in the Object_classification, Key_claim_crosscheck, and Classification_adjudication worksheets. To reproduce the summary counts, filter the Screening_audit worksheet by Final destination and verification status, and compare the retained records against the Original_studies, Core_reviews, Methods_appendix, and Homolog_appendix worksheets. Figure-level source counts and duplicate checks are provided as frozen public-release snapshots. No proprietary software or executable analysis code is required; the dataset can be reviewed using standard spreadsheet software.