sARDS-VIBench: software for auditing virtual cell evidence in sepsis associated ARDS

Résumé fourni par la source

sARDS-VIBench provides analysis code for evaluating virtual cell predictions together with the biological units, comparators and claims they can support in sepsis associated ARDS. The study includes a held out perturbation benchmark, selective reliability analyses, three transport questions and three intervention profiles. In the Alector macrophage CRISPRa analysis, the selected model produced the same off target response vector for all 31 held out targets within each donor after target genes were masked. None of 27 prespecified selectors accepted a held out unit. The three transport analyses measured different biological operations, and all intervention profiles retained unresolved evidence requirements. This is a software only release. It does not include third party datasets, project derived result tables or Alector derived outputs. Redistribution conditions and permanent exclusions are recorded in LICENSE-DATA.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.