Biology Beyond the Gene: Manifold Coherence and the Relativistic Unification of Post--Genomic Frameworks

Biology Beyond the Gene: Manifold Coherence and the Relativistic Unification of Post–Genomic Frameworks Antonios Valamontes Kapodistrian Academy of Science, Tampa, Florida, USA [email protected] Abstract Post–genomic evidence has broken the one–directional hierarchy from genes to phenotype. Physiology, bioelectric morphogenesis and regulated genome rewriting all display multilevel, bidirectional constraint. This work formulates a geometric and relativistic unification of these phenomena. Manifold Coherence (MC) is introduced as an equivalence principle for biological causation: no single level is fundamentally causal; biological organisation is the geometry of trajectories restricted to a coherence manifold in a multiscale state space. A set of axioms yields a Manifold Coherence Theorem: any dynamics satisfying level symmetry, bidirectional influence, constraint realisation and context–dependent genome dynamics admits a representation as constrained motion on such a manifold, unique up to isomorphism. Neo–Darwinism, Noble’s physiology, Levin’s bioelectric morphogenesis and Shapiro’s natural genetic engineering are then shown to correspond to distinct projections, submanifolds or flows within this structure. A relativistic formulation is given in terms of a biological state bundle over spacetime. The framework yields concrete, discriminating predictions about the limits of gene–centric risk models, the geometry of bioelectric attractors, and the statistics of stress–induced genome change. 1 Genesis of the Problem Neo–Darwinism offers a statistical picture in which random heritable variation, filtered by natural selection, suffices to generate biological form. Genes act as primary determinants; organisms are vehicles; development and physiology are consequences of genomic structure [1, 3]. In this hierarchy, the effective state space reduces, in practice, to genomic and reproductive coordinates, while all other levels are treated as epiphenomena. Post–genomic evidence now contradicts this ordering. Polygenic risk scores for complex traits saturate at modest predictive power even with large cohorts and dense genotyping [4]. Protein 1 function depends critically on cellular and tissue context, with identical polypeptides participating in distinct roles [1]. Epigenetic marks, RNA–mediated inheritance, and microbiome composition contribute heritable structure not reducible to DNA sequence [3]. Physiological states, tissue–level dynamics, and environmental signals remodel gene expression and genome architecture, indicating systematic top–down causation [2]. Three programme–level developments sharpen this shift. Noble’s physiology exposes multilevel, distributed causation and robustness in cardiac and systems physiology [1, 4]. Levin’s bioelectric morphogenesis shows that patterns of transmembrane voltage and gap–junction connectivity encode anatomical memories and control regeneration [5, 6]. Shapiro’s natural genetic engineering (NGE) establishes regulated, context–dependent genome rewriting under stress [8, 9]. Taken in isolation, the empirical domains of genetics, physiology, bioelectric patterning and regulated genome change seem disparate. Yet the phenomena themselves do not admit such separation. The empirical constraints demand a single geometric entity in which genomic, physiological, bioelectric and morphogenetic degrees of freedom appear as coupled coordinates of one state space. Manifold Coherence is the construction of that entity: a geometry in which biological organization is the structure of admissible trajectories, and in which the observed post–genomic phenomena are not additions to a hierarchy but projections of a single coherence manifold. 2 Axioms of Manifold Coherence We begin by replacing the inherited causal chain genes −→ cells −→ tissues −→ organisms (1) with a set of axioms about state space, viability and constraint. 2.1 State space and levels Axiom 1 (State space factorisation).There exists a state space Xfor the biological system, equipped with a factorisation X∼ =Xgen × Xepig × Xbioel × Xbioch × Xmech × Xphys × Xbeh × Xsoc × · · · ,(2) where factors correspond, respectively, to genomic, epigenetic, bioelectric, biochemical, mechanical, physiological, behavioural and social degrees of freedom (with additional factors as required). Axiom 2 (Level symmetry).No factor Xiis designated as fundamentally causal by stipulation. Any ordering of levels is a choice of coordinates, not an invariant structure of the dynamics. 2.2 Viability and coherence Axiom 3 (Viable set).There exists a subset V ⊂ X of viable states such that observed trajectories x(t)of the system remain in Vfor their lifetime, except at birth and death, and small perturbations of x(t)within a neighbourhood are typically absorbed by reorganisation rather than immediate failure. 2 Axiom 4 (Coherence functional).There exists a measurable function C:X → [0,1] (3) such that C(x)approximates the degree to which xbelongs to V, with C(x)≈1for robustly viable states and C(x)≈0for non–viable states. Definition 1 (Coherence manifold).Given a threshold C∗∈(0,1), the coherence manifold is defined by M={x∈ X | C(x)≥ C∗}.(4) Biologically, Mcollects states compatible with continued existence and function of the organism (or population) on the timescale of interest. 2.3 Dynamics and constraint Axiom 5 (Underlying dynamics).There exists a smooth manifold e Xcontaining Xas a measurable subset, and a smooth vector field F:e X → Te Xwhose integral curves restrict to the observed system trajectories when initial conditions lie in X. No additional non–physical forces are introduced at the level of F. Axiom 6 (Constraint realisation).For biologically relevant initial conditions x(0) ∈ M, the observed motion x(t)coincides (up to reparametrisation) with integral curves of a constrained vector field FMtangent to M, obtained from Fby a projection FM(x)=ΠTxM(F(x)), x ∈ M.(5) 2.4 Genome dynamics Axiom 7 (Context–dependent genome dynamics).There exists a distinguished factor Xgen and a family of flows on Xgen (NGE–type operations) such that: (i) These flows are activated only when x(t)approaches the boundary ∂M. (ii) Their directions depend on other coordinate blocks (physiological, bioelectric, behavioural, . . . ). Remark 1. Axioms 1–7 formalise empirical content: distributed causation [1], viability regions and robustness, and Shapiro’s regulated genome editing [8, 9]. They do not assume any specific mechanistic implementation. 3 Equivalence Principle for Biological Causation Definition 2 (Equivalence principle for biological causation).A biological system satisfies the equivalence principle for causation if: 1. It admits a factorised state space Xas in Axiom 1; 3 2. Its viable states form a coherence manifold Mdefined by a functional Cas in Axiom 4; 3. Observed trajectories are integral curves of a constrained field FMtangent to Mas in Axiom 6. No level (genomic, cellular, tissue, organismic, social) is designated as fundamental; all are coordinate blocks on the same X, and causation is identified with the geometry of trajectories on M. The analogy to general relativity is direct: the equivalence principle removes the privileged status of inertial frames and replaces forces by geodesic motion in curved spacetime. Here, the gene–centric hierarchy is replaced by trajectories in a coherence manifold of multilevel states. 4 The Manifold Coherence Theorem We now state and sketch the central structural result: once the empirical features encoded in Axioms 1–7 are accepted, the manifold coherence construction is forced, up to natural equivalence. 4.1 Abstract formulation Let Ydenote a measurable state space equipped with a product structure Y∼ =Y i∈I Yi,(6) whose factors correspond to levels of description (genomic, epigenetic, physiological, bioelectric, behavioural, social, . . . ). Let V ⊂ Y be the viable set, and suppose the dynamics and genome behaviour satisfy Axioms 2–7. Theorem 1 (Manifold Coherence Theorem).Suppose a biological system satisfies Axioms 1–7. Then there exist: •a smooth manifold Xand a bijection Φ : Y → X preserving the product structure; •a smooth coherence functional C:X → [0,1]; •a threshold C∗∈(0,1); such that: 1. The viable set corresponds to a coherence manifold: Φ(V)=M={x∈ X | C(x)≥ C∗}.(7) 2. Observed trajectories Φ(z(t)) are, up to time reparameterization, integral curves of a vector field FMtangent to M, obtained by projection of a smooth underlying field F. 4 3. Context–dependent genome dynamics correspond to flows in the distinguished block Xgen whose directions are aligned, in a neighbourhood of ∂M, with −∇xgen C(x),(8) where xgen denotes the genomic coordinates. Moreover, the triple (X,C,C∗)is unique up to smooth isomorphisms of Xand monotone reparametrizations of C. Remark 2. The theorem states that Manifold Coherence is not an optional notation: under minimal empirical assumptions, any admissible model can be represented as constrained motion on a coherence manifold, and this representation is essentially unique. 4.2 Construction of the Manifold Coherence Representation The construction that follows is not one choice among many. It is the unique geometric realisation forced by Axioms 1–7. Given these axioms, the triple (X,C,C∗)arises with necessity, and the steps below present its explicit construction. Lemma 1 (Smooth completion of the empirical state space).Let Y∼ =Qi∈IYibe the measurable product space of Axiom 1. There exists a finite–dimensional smooth manifold Xand an injective measurable map Φ:Y,→ X (9) such that: 1. each factor Yicorresponds to a smooth coordinate block Xi⊂ X ; 2. Φ(Y)is a measurable embedded submanifold; 3. Φpreserves the product structure of Y. Proof. Compactly supported partitions of unity on each factor Yigenerate smooth atlases when embedded into Euclidean space by standard Whitney–type arguments. The product embedding QiRni,→RPiniyields a smooth structure respecting the factorisation. Lemma 2 (Construction of a smooth coherence functional).Let V ⊂ Y be the viable set of Axiom 3. Define the indicator χV:X → {0,1}by χV=1Φ(V). Let ηbe a compactly supported mollifier on X. Then the convolution C=η∗χV(10) is smooth, satisfies 0≤C≤1, converges to 1in the interior of Φ(V), and converges to 0in the complement. Proof. Compact support of ηensures smoothness. The monotone behaviour on approach to ∂Φ(V)follows from the Lebesgue differentiation theorem and regularity of convolution on smooth manifolds. Definition 3 (Coherence manifold).Choose any C∗in the transition region 0<C∗<1. Define the coherence manifold by M={x∈ X | C(x)≥ C∗}.(11) This realises the viable region as a smooth, codimension–one thickened submanifold with well– defined normal bundle. 5 Lemma 3 (Decomposition of the underlying dynamics).Let F:e X → Te Xbe the vector field of Axiom 5. For any x∈ X , decompose F(x) = F∥(x)+F⊥(x)(12) into components tangent and normal to the level set C−1(C(x)). Proof. Since Cis smooth, ∇C(x)is defined on a dense subset of X. The tangent space of the level set is ker(dC(x)), and orthogonal decomposition follows from the Riemannian structure induced by any smooth local coordinate chart. Proposition 1 (Constrained dynamics on the coherence manifold).Define the constrained vector field by FM(x)=F∥(x), x ∈ M.(13) Then for all biologically relevant initial conditions x(0) ∈ M, the integral curve of FMthrough x(0) coincides with the observed biological trajectory up to smooth time reparametrisation. Proof. Axiom 6 states that empirical trajectories remain within Vand reorganise under perturbations without departing from viability. Since Mis a smooth rendering of Φ(V), this implies tangency: the biological vector field contains no component forcing motion transversely across the boundary of M. Thus only F∥is biologically admissible. Proposition 2 (Genome editing as coherence–restoring flow).Let x∈ M and suppose x approaches ∂M=C−1(C∗). Under Axiom 7, the genomic component of Fadmits the representation Fgen(x)=−λ(x)∇xgen C(x),(14) where λ(x)≥0vanishes away from ∂M. Proof. Axiom 7 ensures that the genomic flow is activated only near ∂Vand depends on other coordinate blocks. Smoothness of Cprovides a unique steepest–ascent direction in the genomic subspace. Aligning the flow with this direction ensures restoration of coherence and preserves tangency to M. Theorem 2 (Uniqueness).Let (X,C,C∗)and (X′,C′,C′ ∗)arise from the same empirical data and axioms. Then there exist: 1. a diffeomorphism Ψ:X → X ′preserving the factorisation into coordinate blocks; 2. a smooth monotone function h: [0,1] →[0,1] such that C′◦Ψ=h◦ C; 3. equality of the coherence manifolds: Ψ(M) = M′. Proof. Whitney extension on each block ensures equivalence of completions. Convolution with compactly supported mollifiers differs only by smooth monotone rescalings. Product–preserving diffeomorphisms follow from invariance of the factorisation under Axiom 2. 6 5 Lexicographic Constraint Structure and Manifold Coherence The axioms above are compatible with many choices of coherence functional C. In biological applications, however, constraints are rarely homogeneous: survival and basic physiological integrity take precedence over morphological, behavioural or social objectives. This is the setting of Lexicographic Constraint Optimization (LCO) and sequential partitioning methods in mathematical optimisation [11–14]. We now show that MC admits an explicit lexicographic structure and that the scalar Ccan be chosen as a smooth embedding of an underlying lexicographic order, providing the geometric continuum limit of LCO and CAG–type hierarchies [15]. 5.1 Tiered structure of constraints Let Iindex the coordinate blocks of Axiom 1. Choose a finite partition of Iinto priority tiers I=I1˙ ∪I2˙ ∪ · · · ˙ ∪IK, I1≻I2≻···≻IK,(15) where I1collects survival and basic physiological coordinates (e.g. subsets of Xgen,Xphys), I2 may include tissue–level and anatomical variables, and lower tiers encode behavioural and social degrees of freedom, as in cardiometabolic LCO frameworks [15]. For each tier k= 1, . . . , K, define a nonnegative tier violation functional Vk:X → [0,∞),(16) measuring departure from admissible values in the coordinates indexed by Ik. For example, V1 may encode violations of survival/safety ranges for core physiological variables, V2morphological integrity, and so on. Definition 4 (Lexicographic order of states).For x, y ∈ X , define the lexicographic order ≼lex via x≼lex y⇐⇒ V1(x), . . . , VK(x)≤lex V1(y), . . . , VK(y),(17) where ≤lex is the usual lexicographic order on RK. A state xis lexicographically less violated than yif it has smaller violation in the highest–priority tier at which they differ. In LCO, optimisation problems are solved by minimising V1; subject to that, minimising V2; and so on [11–14]. Here, viability corresponds to small values of all Vk, with strict priority given to lower indices. 5.2 Scalar embedding of the lexicographic hierarchy Lemma 4 (Smooth scalar embedding of lexicographic order).Suppose each Vkis measurable and locally bounded on M. Then there exist positive constants 0< εK≪εK−1≪···≪ε1(18) and a smooth functional C:X → [0,1] such that: 1. C(x)is strictly decreasing in each Vk(x); 7 2. for all x, y ∈ M, x≺lex y=⇒ C(x)>C(y).(19) Proof. Choose a smooth, strictly increasing function ϕ: [0,∞)→[0,1) with ϕ(0) = 0 and ϕ(u)→1as u→ ∞ (e.g. ϕ(u)=u/(1+u)). For a sequence of positive weights (εk)with εk+1 ≪εk, define C(x)=1− K X k=1 εkϕVk(x).(20) By construction, Cis smooth wherever the Vkare smooth and strictly decreases as any Vk increases. Choosing the εkin a lexicographically dominated fashion (each εk+1 smaller than any possible variation induced by εkon M) ensures that the contribution of V1dominates that of all lower tiers, the contribution of V2dominates tiers 3, . . . , K, and so on. Thus if V1(x)< V1(y) then C(x)>C(y)independently of other tiers; if V1(x)=V1(y)but V2(x)< V2(y), the difference in the second term dominates all lower tiers; and similarly for subsequent indices. Proposition 3 (Lexicographic LCO as a specialisation of MC).Given tier violation functionals Vkand Cdefined by (20), the coherence manifold M={x∈ X | C(x)≥ C∗}(21) coincides with a lexicographically constrained feasible region in the sense of LCO: for sufficiently large C∗, states with any violation in V1are excluded before violations in V2, . . . , VKare considered, and so on. Biological trajectories that follow the constrained field FMthus implement a continuous analogue of lexicographic search over feasible states [11–15]. Remark 3. In this sense, Manifold Coherence is the geometric continuum limit of lexicographic constraint hierarchies such as LCO and continuation/partitioning methods (CAG, sequential partitioning of 13, 14): tiered violation functionals become smooth components of a single coherence functional, and discrete lexicographic updates become integral curves of the projected vector field FMon a smooth manifold M[15]. 6 Geometric Embedding of the Four Frameworks The Manifold Coherence Theorem provides a canonical geometry. We now locate the four major post–genomic frameworks within it. 6.1 Neo–Darwinism as a collapsed manifold In gene–centric models, the effective state space reduces to genomic and reproductive coordinates, xgen ∈ Xgen, xreprod ∈ Xreprod.(22) The coherence of a population is encoded by CND(xgen, xreprod),(23) and admissible states reduce to MND ⊂ Xgen × Xreprod.(24) 8 This limit corresponds, in the MC geometry, to the case in which Cdepends only on (xgen, xreprod) and the gradients in all other blocks vanish: ∇xiC ≡ 0for xi∈ Xepig,Xbioel,.... (25) Physiology, development and behaviour become shadows of motion on MND. The mathematical structure is consistent but deaf to multilevel constraints. 6.2 Noble’s physiological relativity Noble’s models of cardiac electrophysiology show that no single ionic current acts as a master clock: elimination of nominally dominant currents leads to modest changes in rhythm as other currents reorganise [1]. This corresponds to a manifold Mphys ⊂ Xbioch × Xmech × Xbioel,(26) on which ∇xbioch C,∇xmech C,∇xbioel C(27) are all nonzero. Robustness and degeneracy correspond to extended regions of Mphys where C is nearly flat along many tangent directions. In MC language, physiological relativity is the statement that the Jacobian of FMhas full rank across these blocks: small perturbations in any of them are compensated by coordinated motion in others, keeping x(t)within M[4]. 6.3 Levin’s bioelectric morphogenesis Levin’s work on planarian and amphibian regeneration demonstrates stable anatomical memories encoded in distributed voltage patterns and gap–junction networks [5, 6]. Animals with altered head–tail patterns maintain their modified body plan across multiple rounds of cutting, despite unchanged DNA sequence. Geometrically, this is represented by an anatomical submanifold Manat ⊂ Xbioel × Xmech × Xphys,(28) with a finite set of attractors {ak}of Cin this subspace. Bioelectric interventions effect controlled displacements in xbioel, moving the system between the basins of these attractors while remaining inside M. MC thus provides a natural language for anatomical pattern memory: body plans are local maxima of Crestricted to (xbioel, xmech, xphys), and regenerative dynamics consist of trajectories ascending the coherence gradient back to these maxima. 6.4 Shapiro’s natural genetic engineering Shapiro’s NGE programme establishes that, under stress, cells activate regulated genome– editing operations: transposition, recombination, domain shuffling, and large–scale rearrangements targeted to specific loci and modules [8, 9]. Mutation spectra are biased rather than homogeneous; genome change is read–write. 9