MASLD-FN: Data, code, and reproducibility materials for a systematic evidence-structure analysis of microbiome-targeted randomized trials

This data and code release accompanies the study “From Class-Level Benefit to Formulation-Level Uncertainty in MASLD: A Systematic Evidence-Structure Analysis of Microbiome-Targeted Randomized Trials.” It provides versioned, machine-readable evidence authorities and reproducibility materials for a systematic evidence-structure analysis of microbiome-targeted randomized trials in adults with metabolic dysfunction-associated steatotic liver disease (MASLD), including studies historically reported under NAFLD/NASH/MAFLD terminology. The final evidence set comprises 136 reports mapped to 78 adult eligible randomized trials. Of these, 65 were ALT candidates and 56 trials contributed 121 selected arms and 65 harmonized ALT mean-difference contrasts. The release includes report-to-trial reconciliation and study-flow authorities; trial- and arm-level ALT and formulation data; ALT harmonization and contrast provenance; formulation/component structural-identifiability outputs; result-specific Cochrane RoB 2 authorities; public-safe title/abstract screening decisions; exact archived database search queries; the frozen six-trial recurrent seven-species-plus-fructooligosaccharide frequentist and Bayesian prior-sensitivity/posterior-predictive analyses; R scripts; reproducibility run outputs; session information; and SHA-256 checksum manifests. The package is designed to reproduce and audit the reported evidence structure rather than redistribute source publications. Primary article PDFs, raw licensed bibliographic-database exports, and verbatim article abstracts are therefore intentionally excluded. The associated Open Science Framework record (DOI: 10.17605/OSF.IO/CT32D) is a retrospective, time-stamped record of protocol evolution, methodological amendments, decision gates, stopping rules, provenance rules, and the frozen analytical framework; it is not a prospective preregistration.