Translating cancer biology into therapeutic opportunities
We characterise tumours at molecular, cellular and spatial resolution, find the mechanisms and biomarkers worth acting on, follow how disease changes under treatment, and design therapeutics against what we find.
Selected research work
Tumour heterogeneity and the microenvironment
Multi-omics and spatial profiling
Resistance to immune checkpoint blockade
Tumour, immune and stromal profiling, with organoid models
Rare and single-patient cancers
Multi-modal profiling, variant structural modelling and a designed assay
Targets with no conventional active site
De novo design against allosteric pockets at protein–protein interfaces
ADCs, bispecifics and T-cell engagers
Antigen selection through to format developability
WHAT WE DO
An integrated solution for actionable therapeutic strategies
Omics analysis
2Longitudinal analysis
3Network modelling
4Structural analysis and target prioritisation
5Translational assay design
OMICS ANALYSIS
Multi-omics integration
Genomic, transcriptomic and proteomic data harmonised across studies and modalities to characterise tumour states and patient subgroups.
Spatial transcriptomics
Xenium, Visium and CosMx data analysed to map the spatial organisation of tumour, immune and stromal compartments.
Tumour immune states
Immune-landscape profiling and cell-state signatures from bulk and single-cell data, annotated against public reference atlases.
Tissue and circulating signals
Joint analysis of tumour tissue and blood-borne markers from the same patient.
LONGITUDINAL ANALYSIS
Clonal composition
Somatic and germline variant calling with clonality-aware analysis, tracking clonal expansion under treatment.
ctDNA, MRD and response
ctDNA analysis for minimal residual disease and molecular recurrence, tumour mutational burden models, and patient stratification on ctDNA profiles.
Resistance signatures
Molecular signatures associated with emerging resistance, including single-patient settings corroborated by clinical course.
Tumour–immune interaction
TCR repertoire profiling and longitudinal immune-state tracking.
NETWORK MODELLING
Network modelling
Pathway and network analysis linking molecular observations to signalling relationships and candidate targets, used for target prioritisation.
Tumour-microenvironment mechanisms
Characterisation of tumour–immune–stromal interactions and assessment of their druggability.
Drug repurposing
Systematic drug–target evidence integration against identified vulnerabilities, generating repurposing and combination candidates.
Biomarkers and endotypes
Molecular stratification of patient cohorts into subgroups, with candidate biomarkers for each.
STRUCTURAL ANALYSIS AND TARGET PRIORITISATION
Structural analysis of variants
Protein structure prediction, structural analysis and molecular dynamics to assess the functional consequences of a variant and its effect on drug binding.
Target prioritisation
A target shortlist ordered by the strength of the evidence behind each candidate, with the dossier that supports it.
TRANSLATIONAL ASSAY DESIGN
Organoid models
Patient-derived and tumour organoids for drug efficacy, resistance mechanisms and response biomarkers.
Image analysis
Whole-slide histopathology and high-content imaging: cell-type quantification, tissue classification, tumour grading and biomarker extraction.
Functional assay design and programme management
Assay design to test a defined mechanistic hypothesis, and integration of computational and experimental workstreams.
DRUG DESIGN
Design and derisk therapeutics
What follows once there is a strategy. Available on its own, and not part of the work above.
Rational therapeutic and combination design
Generative design of small molecules, degraders and covalent binders, optimised for activity together with synthesizability and ADMET.
Design constrained by the intellectual-property landscape as well as by biology, chemistry and pharmacology.
Targeted biologics, from antigen to format
Antigen selection, binder and epitope design, format developability and conjugation chemistry.
Binding affinity by quantum chemistry
Binding-affinity prediction using quantum-chemical and DFT methods, for candidates that empirical scoring functions cannot separate.
Developability and patient selection
In silico assessment of toxicity, permeability and solubility, and biomarker-guided patient selection for trial design.
EVIDENCE
Case studies
Biopharma R&D
Therapeutic design · De novo · Mitotic kinesin
Inhibitors designed from fragments for an allosteric pocket at a protein–protein interface
The conventional active site offered no useful way in. The opening worth designing against was an allosteric pocket where two proteins meet harder to model, and the reason the programme ran computationally first.
Characterise the pocket
Structure modelling and molecular dynamics of the complex, to establish the pocket and its sub-pockets
Build a fragment library
Fragments from known actives, approved drugs and generative chemistry, filtered for drug-likeness
Link and grow inside the pocket
Fragments docked by sub-pocket, then joined with generative linkers or grown in place
Triage the hits
Docking, molecular strain, off-target and toxicity filters, then molecular dynamics on what survived
Answered: which designed molecules hold the interactions the pocket needs, and which sub-pocket each one works through.
Lung cancer · immune checkpoint blockade
Mapping and modelling resistance to immune checkpoint blockade
Why do tumours stop responding to checkpoint inhibitors, and how can that resistance be reproduced in a model you can experiment on?
Tumour, immune and stromal profiling
Spatial transcriptomics
The microenvironment in its spatial arrangement
Patient-derived organoids
Resistance carried into a functional model
Answered: which microenvironment features travel with resistance, and whether they survive into an organoid.
The patient’s tumour profiled across all three compartments, carried into an organoid, then challenged and the question the design turns on: do the same features carry over?
Lymphoma · drug repurposing
Decoding therapeutic vulnerabilities for drug repurposing in lymphoma
Which molecular and cellular features separate the tumour states that matter clinically and does any existing drug already hit them?
Bulk and single-cell transcriptomics
Tumour states resolved at cell level
Surfaceome and proteomic profiling
What is actually presented on the cell surface
Immune-state characterisation and network biology
Phenotypes connected to signalling and process
Systematic drug–target evidence
Repurposing candidates ranked against the vulnerabilities found
Answered: which disease-associated phenotypes carry therapeutic vulnerabilities, and which existing drugs address them.
Tumour states resolved cell by cell, what the cells present on their surface, the immune states and pathways that connect them, and the existing drugs that reach them.
Precision medicine
Single-patient (N=1) workRare, heterogeneous tumour
Uncovering therapeutic vulnerabilities in a single patient with a rare cancer
What biological mechanism can be built from one patient’s multi-modal data, when the tumour is rare and heterogeneous and there is no cohort to learn from?
Multi-modal profiling
Tumour states resolved at cell level
Tumour–immune microenvironment
How the immune compartment is organised around the tumour
Structural modelling of a candidate variant
What the variant plausibly does to the protein
A functional assay, designed
Built to test that specific mechanism experimentally
Answered: which mechanism could guide the choice of an experimental therapy for this patient, and how to test it.
From one patient’s sample: molecular layers read together, the tumour and its microenvironment resolved, and a candidate variant modelled for the assay that tests it.
The platforms behind this work
EVIDENCE
DBTIPS™
AI-enabled retrieval and harmonisation of disease and target evidence literature, multi-omics, pathways, tractability, clinical and market context in one dossier.AI-enabled retrieval and harmonisation of disease and target evidence literature, multi-omics, pathways, tractability, clinical and market context in one dossier.
OMICS
DISTILL™
AI-powered single-cell and multi-omics platform for integration, tumour and immune profiling, and spatial analysis.
AGENTS
Igniva™
Agentic AI platform combining domain-specialised agents, models and workflows to run and reason over these analyses.
SEQUENCING
NGS DataWorks™
Sequencing data analytics for oncology, functional genomics and disease studies, on private cloud or HPC.
READ MORE
From our scientists
BLOG · DEGRADERS
De novo PROTAC design with generative AI and FBDD
Choosing the target, E3 ligase and linker together, with degradability models and MD of the ternary complex.
BLOG · COVALENT
Covalent binder design with generative AI and FBDD
Cysteine-targeting design that combines structure modelling, reinforcement learning and fragments.
Talk to a Scientist about your oncology programme
Bring the problem you are stuck on. We will tell you which of our methods and data apply, and what we have done that is close to it.