MosaicDM Research · Tulane University · 2026

Disease Modeling
using Multi-dimensional Modeling
& Computation with Holomorphic
Geometric Frameworks

Psychiatric and neurodegenerative diseases are not isolated pathologies — they are distinct temporal projections of a single shared latent manifold. This framework unifies their genetics, proteomics, and pathway biology.

Ωshared  →  { Psychiatric,  Structural,  Neurodegenerative }  at  t1 < t2 < t3
25
GWAS inputs across 3 disease domains
722
Deep human brain proteomes profiled
13
Bridge proteins shared across disease families
2.6×
Excess protein–protein interactions vs. chance
30%
Neurodegen. causal proteins shared with psychiatry
Scroll
The Evidential Basis

The holomorphic modeling approach is grounded in landmark multi-omics findings from Wingo et al. (2022), which systematically mapped shared genetic and molecular architecture across the major brain disease categories.

Nature Communications · 2022 · 13:4314
Shared mechanisms across the major psychiatric and neurodegenerative diseases
Wingo TS, Liu Y, Gerasimov ES, Vattathil SM, Wynne ME, Liu J, Lori A, Faundez V, Bennett DA, Seyfried NT, Levey AI & Wingo AP · Emory University
The Holomorphic Disease Field

Each disease is modeled as a 6-dimensional state projected from a shared latent substrate Ω. Different diseases occupy different regions and temporal windows of the same underlying manifold — not separate systems.

Ω shared manifold PSYCHIATRIC MDD · SCZ · BD PTSD · ANX · INS NEURODEGEN. AD · PD · ALS FTD · LBD GENOME Gᵢ · GWAS / LDSC PROTEIN Pᵢ · PWAS / PPI STRUCTURE Bᵢ · Brain MRI TRANSCRIPT Tᵢ · TWAS t₁ → t₂ → t₃

Disease State Space 𝔻ᵢ

G
Genome-wide polygenic liability
25 GWAS · LD score regression across disease families
P
Causal protein field
651 cis + 12 trans causal proteins · PWAS · SMR/HEIDI · COLOC
T
Causal transcript field
615 psychiatric + 64 neurodegen. causal mRNAs · TWAS
B
Brain structural endophenotype
11 MRI traits — cortical thickness, regional volumes, WMH
C
Cell-type / region expression field
Excitatory · inhibitory · oligodendrocytes · astrocytes
τ
Temporal manifestation window
Early adulthood → subclinical → late-life neurodegeneration

Holomorphic Projection Map

// Each disease is a projection from Ω
𝔻i = Πi(Ω)

// Psychiatric disorders → early projections
// Neurodegen. diseases → late projections
// Brain structures → geometric intermediaries

Ω(t) {
  psychiatric,  t₁
  structural,  t₂
  neurodegen, t₃
}

Cross-domain Genetic Overlap

Compositional Analysis Pipeline

The holomorphic pipeline carries structure forward through each representational layer — from genotype to protein to transcript to network to biology — without collapsing relational information.

🧬

GWAS Input

25 genome-wide association studies across 3 disease families. Up to 807,553 participants.

Ω → Rg
📐

LDSC Regression

LD score regression quantifies pairwise genetic correlations across all 24 brain traits (300+ pairs).

Rg structure
🔬

PWAS

Proteome-wide association: 722 brain proteomes. 25 independent PWAS identify 839 cis-regulated proteins.

Pc causal proteins
⚖️

SMR/HEIDI + COLOC

Mendelian randomization disambiguates pleiotropy from LD. Bayesian colocalization confirms causal variants.

651 cis + 12 trans
🧬

TWAS

888 human brain transcriptomes. 615 psychiatric + 64 neurodegen. causal mRNAs identified.

Tc causal mRNAs
🕸️

PPI + GSEA

BioGRID physical PPIs map network topology. Gene set enrichment extracts dominant pathway basins.

M shared mechanisms
Three Projection Domains

The unified manifold maps across 24 brain traits in three groups — each a distinct temporal and mechanistic projection of the shared substrate Ω.

🧠

Psychiatric Disorders

8 traits · Early-to-mid adulthood · t₁
MDD Bipolar Disorder Schizophrenia Anxiety PTSD Alcoholism Neuroticism Insomnia
🔗

Neurodegenerative Diseases

5 diseases · Late life · t₃
Alzheimer's (AD1) Alzheimer's (AD2) Parkinson's (PD) ALS FTD Lewy Body (LBD)
🏗️

Brain Structural Traits

11 traits · Intermediary geometry · t₂
Cortical thickness Cortical surface area WMH Hippocampus Putamen Brainstem Caudate Thalamus Pallidum Amygdala Nucleus accumbens
Disease as Time-Propagating Field

The disease is not changing type — it is changing projection regime. Treating shared mechanisms early may alter late-life neurodegenerative risk.

🌱
Genetic Susceptibility
Polygenic liability present from birth. Shared Ω determines long-term trajectory.
🧠
t₁ · Psychiatric Onset
Early/mid-adulthood. MDD, SCZ, BD, PTSD emerge as Ω projects onto psychiatric manifold.
🔭
t₂ · Structural Changes
Subclinical brain morphology shifts. Measurable by MRI — cortical thinning, volume loss.
t₃ · Neurodegeneration
Late life. AD, PD, ALS manifest. Same causal proteins — different temporal window.
🎯
Intervention Window
Targeting bridge proteins at t₁ or t₂ may modify the neurodegenerative projection at t₃.

Key Genetic Correlations Discovered

AD ↔ Neuroticism — significant positive genetic correlation (FDR p < 0.05)
AD ↔ MDD — significant positive genetic correlation
AD ↔ Bipolar Disorder — cross-domain correlation
AD ↔ PTSD, Alcoholism, Insomnia — polygenic overlap
LBD ↔ Anxiety — novel significant correlation
ALS ↔ Anxiety symptoms — novel neurodegen–psychiatric link
AD ↔ PD ↔ LBD — shared susceptibility within neurodegen. group
Genetic correlation predicts protein sharing — Spearman ρ = 0.39, p = 0.01
The 13 Bridge Proteins 𝓑PN

These 13 proteins are singular intersection points where the psychiatric and neurodegenerative manifold sheets meet. They represent 30% of all neurodegenerative causal proteins, and are the highest-priority targets for cross-disease therapeutic development. The two "dual-layer" proteins (ADAM10, CCDC6) are stable across both protein and mRNA representational levels.

ACE
Angiotensin-converting enzyme · vascular & neuro signaling
ADAM10
★ Dual-layer bridge (protein + mRNA) · amyloid & synaptic
ADK
Adenosine kinase · inhibitory neuron enriched
AKT3
Serine/threonine kinase · PI3K/AKT survival pathway · notably low expression in locus coeruleus
CCDC6
★ Dual-layer bridge (protein + mRNA) · mitochondrial fission
DOC2A
Calcium-regulated exocytosis · excitatory neuron enriched
HSDL1
Lipid metabolism · inhibitory neuron enriched · locus coeruleus
MAPT
Tau protein · shared across PD, putamen, brainstem, cortex, psychiatric
MTSS1L
Actin dynamics · synaptic structure
SPATA2
TNF signaling · excitatory neuron enriched
STX1B
Syntaxin · SNARE complex · synaptic vesicle fusion
STXBP3
Syntaxin binding · oligodendrocyte enriched · vesicle docking
VKORC1
Vitamin K epoxide reductase · coagulation & neuronal function
🕸️ Network Topology: 118 Shared & Interacting Causal Proteins

Beyond the 13 directly shared proteins, physical PPI data (BioGRID) reveals 120 physical interactions between 99 psychiatric and 30 neurodegenerative causal proteins — yielding 118 total shared/interacting proteins. This is 2.6-fold more than expected by chance (bootstrap p = 0.003). The manifold is not just connected at isolated points; it has network curvature. Notable interaction hubs: PDHA1 · MAPT · LACTB · SCFD1 · CCDC6 · STX6

Three Master Holomorphic Axes

Gene set enrichment analysis (GSEA) condenses the shared causal protein network into three dominant pathway basins — the dual-core engine of the shared disease field.

Synaptic Transmission
102 of 118 interacting proteins are synaptic (2.2-fold enrichment). Processes include neurotransmitter secretion, vesicle-mediated transport, synaptic vesicle recycling, and SNAP receptor activity.
2.2× enriched · p < 9.5×10⁻²⁷
🔋
Mitochondrial Regulation
24 of 118 interacting proteins are mitochondrial (3.7-fold enrichment). These form a 66-protein interactome in mitochondrial proximity space, spanning matrix, envelope, and inner membrane compartments.
3.7× enriched · p < 2.3×10⁻⁶
🛡️
Immune & Epigenetic Regulation
Myeloid leukocyte activation enriched at protein level (FDR p < 0.05). At the mRNA level, the regulatory shell broadens: chromatin remodeling, epigenetic regulation of gene expression, histone methyltransferase complex, RNA binding, RNA metabolism, and protein ubiquitination. The 171 causal mRNA PPIs (145 shared proteins) show 2.0-fold enrichment (p = 0.008), mirroring the protein-level network topology.
Protein: myeloid activation FDR p < 0.05 · mRNA PPI: 2.0× enriched (p = 0.008)
🎛️ Dual-Core Engine of the Shared Disease Field
𝓔shared = 𝓢synapse 𝓜mito     // synaptic: information flow layer
                                  // mitochondrial: energy & stability layer

The cross-disease field is not random pleiotropy — it is a conserved bioenergetic-synaptic control architecture, mirrored at both protein and transcript levels. The transcript layer adds chromatin/RNA regulation as an upstream control shell: 𝓣shared → 𝓟shared → 𝓜shared. Notably, only 2 of the 13 bridge proteins (ADAM10 & CCDC6) are validated at both molecular layers — making them the most robust mechanistic anchors in the model.

Project Goals & Research Roadmap

The multi-dimensional holomorphic geometric modeling framework enables a new generation of cross-disease research — from early intervention design to precision therapeutic targets.

Goal 01 · Framework

Map the Full Shared Disease Manifold Ω

Extend the holomorphic state-space representation to cover additional disease domains, incorporate longitudinal biobank data, and formally characterize the projection operators Πᵢ connecting Ω to each observable phenotype.

Multi-omics Integration
Goal 02 · Targets

Characterize the 13 Bridge Proteins as Therapeutic Anchors

Perform deep functional characterization of 𝓑PN — particularly ADAM10, CCDC6, MAPT, and AKT3 — across cell types, brain regions, and disease stages to identify tractable intervention points.

Goal 03 · Temporal

Model the t₁→t₂→t₃ Disease Trajectory

Use longitudinal cohort data to empirically validate the temporal holomorphic model — testing whether psychiatric phenotype severity at t₁ predicts structural changes at t₂ and neurodegeneration risk at t₃ through shared protein expression.

Goal 04 · Network

Expand the PPI Network Curvature Analysis

Map the full 118-protein interacting causal network using high-resolution proximity proteomics and spatial transcriptomics — particularly resolving the 66-protein mitochondrial interactome and its synaptic coupling.

Goal 05 · Translation

Design Cross-Disease Early Intervention Strategies

Leverage the shared mechanism kernel to design therapeutic strategies targeting the bioenergetic-synaptic dual core (𝓢synapse ⊗ 𝓜mito) — potentially modifying both psychiatric symptoms and long-term neurodegeneration risk simultaneously.

Goal 06 · Modeling

Build Predictive Holomorphic Simulation Tools

Develop computational tools that simulate disease field propagation — allowing researchers to perturb individual nodes in the shared manifold and predict downstream phenotypic consequences across psychiatric and neurodegenerative outcome measures.