GB Discovery

GB Discovery here we can leverage the work in our Health and Community solutions to deliver the next generation of new algorithms and pharmaceutical targets.

Ancestry as a Service

We offer state-of-the-art genetic ancestry algorithms as a service (AaaS). Here are some of our proprietary algorithms:

Global Ancestry:
Neural admixture is a neural net adaptation of the well known ADMIXTURE algorithm that allows us to infer the proportions of ancestry from each individual with greater accuracy. Neural admixture models can be developed for custom ancestries and also they can be stored for posterior use in a federated learning approach.

Local Ancestry:
Gnomix is a local ancestry inference tool powered by XGboost, a modern machine learning method which produces high accuracy estimations at the chromosome level of inference.

Customized ancestry solutions

We help you build a tailored pipeline that fits your use case. Who is using our ancestry solutions?

We provide the backend solution to power several ancestry products: Global Ancestry, European Breakdown, Native American Report, Asian Breakdown, and Hispanic Reports.

We provide free global ancestry estimation to our research participants, through continental-level analysis that can be easily consumed through our interface.

Health Risk Scoring

Polygenic Risk Score (PRS)

We deploy novel algorithms focused on the aggregation of effects from multiple genetic variants identified from GWASes to be associated with a given disorder. The use of PRS models can inform the risk of developing a given disease during a person’s lifetime and this knowledge may contribute to the implementation of early detection and prevention strategies.

Automated Chart Review (ACRe)

Our solution maps clinical features extracted from the electronic health record (EHR) with features associated with Mendelian disorders. ACRe quantifies the degree by which a patient’s clinical records resembles a Mendelian disorder’s phenotype. ACRe models can help us to uncover rare-variant associations with Mendelian disorders that otherwise would require the stratification of enormous sample sizes. ACRe may also allow us to identify subsets of patients with undiagnosed Mendelian disorders.