Life Specific Data and IODS (Intra-Organismal Data Symbiosis)

 IODS — Intra-Organismal Data Symbiosis

Biology's most productive metaphor is also its most misleading. We call genes "master regulators," frame CRISPR as "editing the code of life," and treat the gap between genotype and phenotype as a sampling problem. The metaphor of DNA-as-blueprint has driven enormous scientific progress — but the inference from experimental convenience to ontological primacy is an error, and one with real consequences for how we design research, interpret data, and allocate funding.

Intra-Organismal Data Symbiosis (IODS) is a new theoretical framework that takes a different position. Developed across two companion papers, IODS proposes that living organisms are organised through the structured interdependence of multiple data modalities — DNA, the epigenome, proteome, metabolome, microbiome, and behaviour — none of which occupies a permanent causal summit.

This is not the naïve claim that "everything matters equally." That would be empirically absurd. IODS draws a strict distinction between mutual constraint (the structural property that all modalities participate) and causal magnitude (the quantitative, context-dependent contribution of each). A genetic knockout is typically more dramatic than a temperature fluctuation — but the question is whether that asymmetry is a permanent fact about DNA's privileged status, or a context-dependent feature that shifts across timescales, environments, and developmental phases. IODS argues for the latter.

The stability gradient

The framework's core organising principle is a stability gradient: DNA is the most biochemically stable biological modality (persisting across generations), followed by the epigenome (days to years), the proteome (hours to days), and the metabolome (seconds to minutes). This generates a specific prediction: *each modality's causal magnitude peaks at the timescale matching its characteristic stability.* Stable modalities dominate long timescales; fast modalities dominate short ones.

This reframes what looks like genetic primacy as a context-selection effect. GWAS, knockout studies, and heritability analyses all preferentially sample contexts where DNA's magnitude is genuinely high — not because genes are universally primary, but because these experimental designs are tuned to the timescales where stable modalities shine.

Testable predictions

IODS produces differential predictions that distinguish it from both gene-centric models and prior distributed-causation frameworks:

Expression variance in isogenic populations is structured by developmental and epigenetic context — it's not noise.

Phenotype prediction improves nonlinearly** when contextual modalities are added to genetic data — step-function, not additive.

Heritability estimates for the same trait increase with measurement timescale — a novel prediction unique to IODS, directly derived from the stability gradient, and falsifiable.

Phenotypic ensembles encode recoverable information about genetic organisation — the decisive test, formalised as the bidirectional translation protocol.

The bidirectional translation test

The empirical heart of the framework: if biological modalities genuinely co-constrain, then the genotype-phenotype mapping should be readable in both directions. It should be possible not only to predict phenotype from genotype (the conventional direction) but also to recover structured information about genetic organisation from phenotypic data — exceeding what phylogenetic classification alone provides.

The companion paper formalises this as a computational protocol with three independently testable conditions:

C1 — Symmetric information: Cross-modal mutual information must be comparable in both directions.

C2 — Phylogenetic surplus: Prediction accuracy must exceed what taxonomy alone achieves.

C3 — Context sensitivity: Environmental context must improve accuracy by a measurable margin.

Failure of any condition has a specific diagnostic interpretation. If C1 fails, DNA is causally privileged and the gene-centric model is sufficient. If C2 fails, the correlation is phylogenetic, not mutual constraint. If C3 fails, a simpler model suffices. If all three fail, IODS is falsified as formalised.

Open-source implementation

The full reference implementation is available at https://github.com/alphanv/iods-framework under  AGPL-3.0 license. It includes modality-specific encoders (Transformer for DNA, ViT for images, CNN for audio), a context-dependent magnitude function, round-robin alignment that grants no modality architectural privilege, cross-attention fusion, graph attention networks for ecological context, and the complete C1–C3 validation suite with phylogenetic null models.

All training data comes from freely available biological databases — GenBank, iNaturalist, GBIF, xeno-canto, GLOBI, Open Tree of Life — with no new data collection required. Missing modalities are handled natively through modality dropout.

Where this stands

The framework is at Technology Readiness Level 2–3: concept formulated, mathematical architecture specified, reference implementation built, falsification conditions stated. What comes next is empirical validation on real cross-species data, metric quantification of the magnitude function, and scaling from individual organisms to planetary scope.

The papers commit to weak IODS — the epistemic claim that modality-coordination explanations will prove more empirically productive than gene-centric alternatives for the phenomena specified. Strong IODS — the ontological claim that living systems are constituted by symbiotic relations among modalities — remains plausible but is not yet required by the evidence.

The ultimate aim is not to replace molecular mechanism with semiotic theory, but to clarify how they relate: *how molecules become meaningful, how meaning becomes material, and how the transaction between the two sustains what we recognise as life.*

Alphan Vardarlı — Independent Researcher, Çanakkale, Türkiye

ORCID: https://orcid.org/0009-0007-1581-2190

GitHub: https://github.com/alphanv/iods-framework