Skip to content
Finding signal in non-linear combinations-min

Finding non-linear signal using combinations

Our combinatorial analytics approach reveals many more of the biological drivers of disease than any other method.

It uniquely captures the non-linear effects of interactions between genes and other epidemiological and environmental factors as they interact across complex genetic and metabolic networks. 

We generate much more signal from much smaller datasets than traditional Genome Wide Association Studies (GWAS) models, and we incorporate a wide range of multi-modal data including genomic, transcriptomic, clinical, environmental and epidemiological data. This allows us to better stratify patients according to the mechanistic origins of their disease.

GWAS misses signal in polygenic diseases

Because it looks to find the independent effects of single SNPs, GWAS misses much of the real disease signal in polygenic diseases, making it difficult to stratify patients.

Single gene approaches such as GWAS have transformed the study of cancer and rare disease but haven’t had nearly the same impact on more complex chronic diseases. This is because GWAS is an inherently limited and low-resolution tool best suited to relatively simple monogenic diseases.

GWAS finds single SNPs that occur more often in a patient population than in healthy controls. It assumes the effects of those SNPs on a disease are independent, additive, and the same across all patients, ignoring subgroups and interactions between SNPs and genes. It focuses on rarer variants often not shared between different populations, especially those with different ethnic backgrounds.

GWAS misses signal in polygenic diseases
5. Endometriosis-01-min

Insights that power innovation

Machine learning and polygenic risk models built on the results of GWAS cannot reconstruct the critical non-linear interactions and so are also incomplete and lack predictive power in chronic diseases, especially across populations with different origins.

Combinatorial analytics reveals many more valid genetic associations than GWAS and generates deeper disease insights from much smaller patient datasets than alternative AI drug discovery approaches.

As well as case:control studies we also apply our analytics to examine the features associated with a continuous phenotypic trait – for example retrospectively analyzing patients' responses to a drug candidate in a clinical trial to identify biomarkers that are predictive of drug response.

An essential tool for the delivery of precision medicine

You might also be interested in

DiseaseBank icon white


The deepest repository of insights and IP into the biological mechanisms of chronic diseases

Learn more
Time to insight icon white

Time to insight

Our analysis of chronic diseases rapidly reveals insights that would usually take years of investigation

Learn more
Deployment platform icon white

Deployment platform

We work with collaborators worldwide and protect their most confidential and sensitive data

Learn more