Scripps Translational Science Institute

Each cancer is characterized by numerous somatic mutations, of which only a subset contributes to the tumor’s progression. Scientists want to be able to distinguish these “driver” mutations from the preponderance of neutral “passenger” mutations that characterize each cancer, so that they can take advantage of the wealth of data produced by the large-scale tumor resequencing studies to advance understanding and treatment of cancer.

Scripps Translational Science Institutes (STSI) researchers developed a support vector machine (SVM)-based method that differentiates common, likely nonfunctional genetic variations from Mendelian disease-causing polymorphisms, specifically within the genes encoding the kinases, the protein family most frequently associated with cancer development and progression.

The researchers subsequently demonstrated that this specialized method of making highly precise predictions within the protein kinase gene family also accurrately and quickly identified known “driver” mutations in tumors. The scientists showed that their method generated data that aligned with previous statistical estimates of the frequency of “drivers,” and provided strong evidence that predicted “drivers” are under positive selection.

In multiple cancer data sets, the method identified particular positions in protein kinases that seemed to play a role in oncogenesis. It also enabled the researchers to compile a ranked list of candidate “driver” mutations. These “driver” mutations seemed to be very involved in nucleotide binding, possibly driven by resistance to inhibitors mimicking ATP, and regulatory functions, especially movements from the inactive to active conformation.

Accurate predictions of “drivers” in other protein families, such as transcription factors or phosphatases, build upon the data on kinases to help scientists draw a more ‘‘holistic’ picture of tumorigenesis and cancer treatment.