Engineered gut graph: constrain model-inferred interactions and diet-dependent evenness claims #1454

Engineered_Gut_Amino_Acid_CrossFeeding_Consortium.yaml, PMID:31409662, doi:10.1128/mSystems.00352-19.

Growth rescue is unequal; B. fragilis is not rescued in the tested supernatants although it grows in coculture. Do not interpret the design as measured complete all-to-all exchange. Beneficial interaction edges were inferred from in-vitro qPCR time series using a Bayesian dynamical model; reduced evidence for negative edges is not proof that antagonism disappeared. Remove community-wide MUTUALISM and transporter-specific annotation. Preserve three nodes and two PARTIAL arrows, with canonical four-member rosters. In-vivo evenness is a day-10 fecal qPCR endpoint in a four-member gnotobiotic community, not longitudinal stability, clinical benefit or direct flux. Low-protein comparisons were significant; the standard-diet trend was not. Model-inferred in-vitro interactions do not establish the same mediation in the mouse gut. No new species-pair mechanisms or engineering procedures are needed.

Review scope: whole records and cached primary main bodies, tables and captions read. No figure images, separate supplements or raw data reviewed. Selected Emiliania published peer-review discussion was also read, not the entire exchange. Deduplication searched 772 all-state issue titles/bodies and ignored/hidden-inclusive local content; #583/#529 and #1091/#1094 bodies reviewed. No exact duplicate within that scope; not every issue comment or PR reviewed. This is source-based self-adversarial review, not independent approval.