I-Corps: In Silico Prediction of Cancer Drug Susceptibility

Award ID: 2116886

The broader impact/commercial potential of this I-Corps project is the development of computational method that will accurately predict whether a particular genetic variant is susceptible to a particular cancer treatment drug. Drug resistance is the main reason a cancer drug treatment regimen fails to cure a patients cancer. With drug resistance occurring in 30-85% of cancers, it accounts for a major part of the approximately $200 billion spent on cancer care in 2020 in the United States. This proposed technology may help to improve the diagnosis and treatment of cancer patients by avoiding rounds of failed chemotherapy by tailoring treatment to the genetic profile of the cancer. By determining the correct drug treatment based on the genetic profile of the cancer, the chances of patient survival may increase, the treatment time may decrease, and the overall cost of treatment may decrease. In addition, the proposed technology may be applicable to the diagnosis and treatment of rare genetic diseases and other personalized medicine applications by identifying variants that lead to the disease phenotype. This I-Corps project is based on the development of a computational method that uses machine learning applied to feature sets derived from molecular simulation to predict the functional consequences of genetic variation. Depending upon the cancer type, cancer drug treatments fail 30-85% of the time because of drug resistant genetic variants. However, only a small number of these variants have been linked to a specific drug treatment. To address this problem, a method was developed that leverages machine learning applied to features of all atom molecular dynamics simulations to predict the specific functional effect and the disruptive severity of genetic variants. This technology may be applied to data extracted from the molecular simulation of the proteins that are cancer drug targets and used to predict the susceptibility of a particular genetic variant of the protein to a specific cancer drug. The proposed technology was used to quantitatively predict cancer drug resistance caused by variants of the oncogene BCL-2 to specific drugs such as Venetoclax. Specifically, the technology was shown to determine which cancer drug will work on which genetic variant leading to targeted therapy tailored to the cancer patients genetic profile. In addition, the proposed technology may be used to quantitatively classify newly found or unclassified variants as cancer causing or benign creating the potential of early and more accurate cancer variant diagnosis. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Program Director: Ruth Shuman