Trained on more than 30,000 tumors, the UC San Diego model looks for combinations of genetic alterations where conventional biomarkers look at only a few genes.
Researchers at the University of California's San Diego School of Medicine presented MutationProjector, a model of genomic data from over 30,000 tumours spread across ten solid cancers. Its principle: instead of relying on a handful of known biomarkers, analyze wider combinations of genetic alterations placed in the molecular networks of the cell.
On several independent cohorts — bladder, lung, melanoma — the model equals or exceeds existing methods for predicting response to immunotherapy and chemotherapy. The challenge is concrete: today, only about 8% of patients see their genetic profile leading to FDA-approved treatment.
The work, led by Trey Ideker with JungHo Kong as first author, was published in the American Association for Cancer Research's Cancer Discovery magazine.
Read the full article on the original source
Go to the source →