There is a newer version of the record available.

Published June 18, 2026 | Version v2

Coarse Graining, Sampling Bias, and Emergent Dynamics: How Discretization Choices, Network Topology, and Stoichiometric Constraints Jointly Shape Inference in Biological Systems

  • 1. Saluca LLC

Description

Version 2 — revised in response to an external structural review and an automated critique pass. See "Response to Review" appendix in the PDF for the change log.

A recurring structural problem cuts across several recent preprints in molecular network biology, population genetics, and genomics: the inference tools we deploy to characterize biological systems introduce systematic distortions that are not random noise but are instead architectural—embedded in the discretization schemes, sampling distributions, or representational formalisms chosen at the outset. This paper synthesizes six findings from the q-bio corpus to argue that a coherent pattern is *visible* across scales—though not formally derivable from a single shared structure: (1) Boolean discretization of gene regulatory networks systematically suppresses intermediate dynamical behaviors including higher-order multistability and stable periodic orbits [corpus:arxiv:2606.14925]; (2) uniform sampling of canalizing Boolean functions over parameters rather than over distinct functions exponentially suppresses high-sensitivity functions, biasing conclusions about network robustness and attractor structure [corpus:arxiv:2606.05196]; (3) autocatalytic formalisms that appear mathematically incompatible—RAF sets and stoichiometric autocatalysis—share a common stoichiometric matrix representation, and under mild conditions any RAF is stoichiometrically autocatalytic, suggesting the apparent theoretical gap is at least partly an artifact of representational choice [corpus:arxiv:2605.25523]; (4) a transformer-based foundation model for m6A RNA methylation demonstrates that reformulating the input representation (peak-derived priors rather than adenosine-centered windows) substantially reduces false positives and improves precision-recall performance, though a PR-AUC of 0.635 indicates meaningful false positives remain [corpus:arxiv:2606.12219]; (5) spatial context is a non-ignorable variable in cell-level gene expression inference, and treating cells as i.i.d. introduces counterfactual errors correctable by explicit disentanglement of intrinsic state from neighbor context [corpus:arxiv:2606.08493]; and (6) elemental stoichiometry across metabolomes appears to occupy a statistically distinct region of chemical space relative to synthetic and planetary chemistry samples—though this distinction depends on standardized data-collection methods—suggesting that the *statistical envelope* of molecular composition may be a candidate biosignature [corpus:arxiv:2605.19252]. This is a heuristic reading, not a derivation: the six findings do not share a single formal structure, but they share a common inferential failure mode—conclusions that depend on representation are being treated as conclusions about biology. The primary falsification path is stated per claim. Sources are drawn from q-bio.MN, q-bio.GN, q-bio.BM, and q-bio.PE preprints from May–June 2026. ---

Authorship: Saluca Agentic AI Research Team (Saluca LLC). AI-drafted from arXiv preprint corpus on the date in the filename.

Cited arXiv preprints: 2602.02840, 2605.19252, 2605.21945, 2605.25523, 2605.29958, 2606.03071, 2606.05196, 2606.07372, 2606.08493, 2606.12219, 2606.12573, 2606.12712, 2606.14925

AI disclosure. This work was produced with an agentic AI research apparatus operated by Saluca Labs. The apparatus drafted, searched and analysed under direction. Cristian Ruvalcaba is the human author and is accountable for the content. No AI system is listed as an author or contributor, because authorship entails accountability that a model cannot hold; this disclosure is the credit, and it is deliberately the whole of it.

Notes

This paper was AI-drafted by an internal multi-persona research agent over a curated arXiv corpus. It is not peer-reviewed. All cited works are listed by arXiv ID; readers should follow those links to verify claims against the primary preprints.

Files

20260618_beast-boy_representation-artifacts-biological-network-inference_v2.pdf

Files (78.1 kB)