.jpeg)
Omni Genomic Research · Horizon programme
From genomic sequence
to living system.
A long-horizon research programme investigating how transcriptomic, epigenomic, proteomic, spatial and temporal signals can be integrated into dynamic biological graphs—then tested against physical constraints, uncertainty and experimental evidence.
The research question
Can biology be modelled
as a changing system?
Genes, proteins, metabolites, cells and tissues do not operate as isolated lists. The programme asks whether their interactions can be represented as evidence-linked graphs that preserve time, spatial context, uncertainty and biological constraints.
Multi-omic integration
Combine transcriptomic, epigenomic, proteomic, metabolomic and spatial measurements without erasing modality-specific uncertainty.
Dynamic biological graphs
Model regulatory relationships, cell states and molecular interactions as evolving networks rather than static reference tables.
Physical and experimental tests
Constrain computational predictions using spatial evidence, molecular dynamics, known biological rules and independent laboratory validation.
From shape to system
Structure matters.
Dynamics determine function.
The long-term direction moves beyond predicting a static molecular structure toward understanding how molecular, cellular and tissue-level states change, interact and respond to intervention.
Kinetic behaviour
Investigate stability, conformational change, binding and solvent-dependent behaviour rather than relying on a single predicted fold.
State transitions
Track how regulatory, metabolic and signalling changes move cells between healthy, stressed, adaptive and disease-associated states.
Spatial organisation
Connect cell state with physical neighbourhood, tissue architecture, mechanical context and local signalling environments.
Morphogenetic models
Explore how collective cellular rules may produce, maintain or repair large-scale biological form.
Architecture pathway
Harden the science
before expanding autonomy.
The programme begins with reproducibility, provenance and computational integrity. Automation and physical actuation remain later-stage possibilities, conditional on validated models, governance and suitable containment.
01 · Reproducible genomic infrastructure
Containerised workflows, version-locked references, cryptographic file integrity, isolated sample processing and auditable data lineage.
02 · Graph and multi-modal representation
Interoperable single-cell and multi-omic data structures, variation-aware references, entity resolution and evidence-linked biological graphs.
03 · Dynamic and uncertainty-aware modelling
Graph learning, sequence models, differentiable molecular simulation and probabilistic methods with explicit calibration and precision auditing.
04 · Human-guided experimental loops
Only after validation: carefully governed laboratory integration where model suggestions are reviewed, bounded and tested through approved experimental protocols.
Evidence boundaries
Separate what exists
from what may become possible.
This is a research programme, not a clinical diagnostic service, autonomous laboratory or validated biological control platform.
Computational foundations
Single-cell analysis, multi-omic integration, workflow containerisation, molecular simulation and graph-based biological analysis are active scientific capabilities.
Dynamic system inference
Reliable regulatory-network recovery, cross-scale modelling and generalisable prediction of biological state transitions remain difficult and context-dependent.
Morphogenetic operating systems
Closed-loop biological design, autonomous experimentation and programmable tissue-scale outcomes remain conditional future directions requiring extensive validation and governance.
Failure modes
Scientific velocity without
silent biological failure.
The programme treats reproducibility and falsification as core capabilities rather than administrative overhead.
Numerical drift
Accelerated computing can suppress rare signals. Critical outputs require precision-aware validation and reproducible reference calculations.
Implausible relationships
Predicted interactions must be tested against spatial, mechanistic and experimental constraints rather than accepted from model confidence alone.
State contamination
Samples, references, intermediate artefacts and model versions must remain isolated, hashed and traceable throughout the pipeline.
Automation ahead of control
Laboratory actuation must remain human-authorised, bounded, logged and subject to biosafety and institutional review.