Explore colorectal cancer molecular landscape

The scatter plot shows a 2D embedding of the tumors based on multi-omics fingerprints, which are composite molecular features capturing patterns across gene expression, copy number variation, and mutations. Each dot represents a patient tumor. The patient groups represent groups tumor samples which are similar in their multi-omics fingerprints.

The box plot above shows the tumor mutational burden in mutations per megabase of the genome (log scale). Mutational burden is associated with immunotherapy response

The bar chart above shows the percentage of patients by microsatellite instability status.

The box plot above shows the CD8 effector scores by patient group. The CD8 effector score is shown to be associated with response to immune therapy.

The bar chart above (activated when a patient group is selected) shows the top three drugs for that group. The drug response is inferred from similarity of primary tumors to disease models.

The bar chart above shows the most common markers for known drugs which are present in each group. These markers are obtained from public database and some of them are FDA approved diagnostics for certain cancers.

Inferred Targets

The tables above show the number of matched disease models per patient group, and the known/novel cancer targets for each patient group based on these disease model studies. Matching is done using multi-omic fingerprints obtained via our deep learning platform that focuses on molecular similarities between tumor models and primary tumors.