Abstract:
|
One central theme of cancer genome research is to identify the genes associated with the initiation and progression of cancer. With silent somatic mutations as the control group, cancer-associated genes can be detected through the disparity between the observed number of non-silent mutations and its expectation under the null hypothesis that no selection pressure exists for non-silent mutations favoring carcinogenesis. Instead of treating selection pressure homogeneously across subjects as previous methods, we investigate the selection pressures on the individual level. The new method implements one-sided score tests for category specific and subject specific selection pressure parameters, which enables us to identify selection pressure heterogeneity either across mutation category or between subjects. Applied to the lung squamous cell carcinoma of The Cancer Genome Atlas (TCGA) datasets, the proposed method is able to properly control the false positive rate, identify several known cancer genes and find two extra age-dependent cancer genes.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.