Skip to main content


Phase III of III
Visual representation of data flowing from one to many, over top cellular texture.

Predictive Transcriptome Analysis

Instead of focusing on a single hypothesis for disease, Gordian’s Pythia examines the entire transcriptome to see how each therapy affects dozens of cellular pathways in the living animal. This allows Gordian to observe changes across multiple areas of cellular activity, providing a broader understanding of the effects of these therapies. In other words, with most predictive transcriptome analysis, drug developers must decide in advance what they want to measure. With Pythia, our approach is unbiased.

Pythia further predicts how therapies will work in clinical trials by combining transcriptomics data from Mosaic Screening with measurements of real-life clinical outcomes. We use samples from human patients to make paired datasets that link the transcriptome to clinically relevant readouts. This lets us translate transcriptomic data from Mosaic Screening to predictions about physiological human outcomes.

Pythia builds on itself by training machine-learning models on our ever-growing number of in vivo data sets across indications. As a result, each screen we conduct enhances our ability to predict the effectiveness of potential treatments.

Gene therapy allows us to write arbitrary changes to the genome in living cells. Single-cell sequencing allows us to read the effects of those changes at the cellular level. Machine learning allows us to relate these cellular changes to clinically relevant outcomes for the whole organism. Together, these read / write / learn technologies allow us to manipulate and understand the genetic code driving the disease of interest in a living animal and directly discover therapeutics that will modify that code and restore healthy function.

Ian Driver Head of Bioinformatics
A portrait of Ian Driver.