Transcriptional memory: prediction and identifiability — analysis release

<p dir="ltr">Analysis code, retained inputs, derived measurements and numerical outputs accompanying the study</p><p dir="ltr">"Calibrated interventions, not residual expression, identify signaling-independent transcriptional</p><p dir="ltr">memory" (Yu, Desterke & Mata-Garrido).</p><p><br></p><p dir="ltr">The study separates two questions that the transcriptional-memory and trained-immunity literatures</p><p dir="ltr">usually treat as one: whether a primed transcriptional response carries transferable predictive</p><p dir="ltr">information, and whether that information identifies the mechanism maintaining it.

It evaluates</p><p dir="ltr">gene-specific response transfer under strict information restrictions in public RNA-seq datasets,</p><p dir="ltr">reserving whole donors, whole challenges and whole ligand pairs from calibration. It then states</p><p dir="ltr">and proves a sharp lower bound on the signaling-independent component of a persistent response,</p><p dir="ltr">and shows that the bound is informative only when the efficacy of the maintenance intervention is</p><p dir="ltr">calibrated on the same scale as the response itself.</p><p><br></p><p dir="ltr">Contents:</p><p dir="ltr">- Analysis code for every reported evaluation</p><p dir="ltr">- Retained input tables and derived measurements</p><p dir="ltr">- Every primary reserved prediction, with its control definitions and uncertainty calculations</p><p dir="ltr">- The frozen local analysis specifications, with SHA-256 hashes and recorded access timestamps</p><p dir="ltr">- Per-panel source mappings (SOURCE_DATA_MAP.csv) linking each manuscript figure to its numerical table</p><p dir="ltr">- Environment information and file checksums</p><p><br></p><p dir="ltr">Unfavourable and domain-incompatible observations are retained rather than filtered.