Evaluate & optimise
Start with a known compound and receive a composite scorecard across binding, cancer activity, safety, gene expression, ADMET, and structural docking.
Platform architecture
One ecosystem for therapeutic discovery intelligence.
This page explains the architecture behind Vecentra: the Gnosis modules, the Forge generation engine, and the decision layer that brings their signals together into one ranked output.
Vecentra
Decision Layer
GNS I
Target Binding
GNS II
Cancer Cells
GNS III
Safety
GNS IV
Gene Expression
GNS V
ADMET
FORGE
Generation
Gnosis intelligence suite
Six specialised models. One coherent decision.
Each module focuses on a distinct scientific question, then the ecosystem brings those signals together into a clearer view of candidate quality, risk, and next steps.
Gnosis I
Target Binding + 3D Docking
9 targets · auto 3D docking · pocket-resolved
Predicts binding across 9 oncology targets. Every top candidate is auto-docked into the 3D pocket of the relevant protein.
Gnosis II
Cancer vs Normal Cell Selectivity
700+ cancer cell lines · normal-cell selectivity index
Predicts IC50 and activity across cancer and normal cell lines. A cancer-vs-normal selectivity index downranks toxic non-selective compounds automatically.
Gnosis III
Clinical Safety
9 endpoints · multi-organ panel
Nine preclinical toxicity endpoints plus six organ-level clinical AE categories. Safety gates suppress favourable predictions when toxicity flags fire.
See it run live on doxorubicinGnosis IV
Downstream Omics
946 landmark genes · 49 Hallmark pathways · 76 cell-line contexts
Predicts the differentially expressed genes and Hallmark pathways a compound triggers, per cell line, dose, and timepoint. Drives MoA and resistance strategy.
Gnosis V
Drug Behaviour (ADMET)
12 models · 5-enzyme CYP panel
Twelve ADMET models covering absorption, distribution, metabolism, excretion, and toxicity.
Gnosis Forge
Generation Engine
50+ candidates per campaign · 142 mechanism categories
Every molecule Forge generates is scored in real time by all five Gnosis modules simultaneously. Binding, selectivity, safety (parent and metabolites), ADMET, and gene expression all feed back into the generation loop. Candidates are corrected during generation, not filtered out after it.
De novo generation
Start with a disease, target, or mechanism objective. Forge refines every generated candidate in real time against all five scoring dimensions at once, producing novel molecules that are potent, selective, safe, and metabolically sound from the first run.
How it works
One workflow, two modes, six intelligence modules.
01
Compound, target, or disease objective
Known molecule optimisation or de novo generation from a therapeutic goal.
02
Parallel scientific evaluation
Binding, selectivity, safety, gene expression, ADMET, and docking are evaluated together.
03
Multi-objective candidate generation
Flagged weaknesses become reward signals for improved candidate rounds.
04
Ranked shortlist with rationale
Outputs include scorecards, confidence, structural context, and next-experiment priorities.
Therapeutic scope
Cancer drug discovery first. Anti-neurodegenerative drug discovery next.
GNOSIS is built as a disease-agnostic architecture. Cancer is the lead program because it is where the platform has the deepest current data and target coverage. The learning patterns developed in cancer transfer into adjacent disease areas.
Cancer foundation
Current capability is cancer-focused, spanning oncology targets, cancer cell response, safety, pathway effects, and anti-metastasis biology.
9 live cancer targets700+ cancer cell lines57-gene anti-metastasis score
Next oncology layer
Planned expansion into KRAS G12C/G12D, BTK, EZH2, GPX4, PROTAC degrader modes, synthetic lethality, and covalent design.
KRAS roadmapPROTAC modesynthetic lethality scoring
Anti-neurodegenerative drug discovery next
The cancer ecosystem creates reusable learning patterns for target biology, safety, cell-state response, and candidate optimisation that can transfer into Parkinson's disease and broader anti-neurodegenerative drug discovery programs.
LRRK2alpha-synucleintransferable learning roadmap
Cross-module validation
The Coherence Engine.
Every prediction across the six modules is cross-checked by the Coherence Engine, a 13-rule contradiction detection layer that flags inconsistencies between modules before a result is surfaced to a researcher. Strong binding paired with reduced gene-expression engagement, favourable ADMET paired with adverse safety flags, high selectivity paired with off-target cytotoxicity: every combination is rule-checked. Contradictions are surfaced as low-confidence flags rather than hidden behind a single composite score.
One ecosystem. Cross-validated by design.