CancerLocator non invasive cancer diagnosis and tissue of origin prediction using methylation profiles of cell free DNA

• Brandon Grimes  |  Fengzhu Sun  |  Frank Alber  |  Gina Lee  |  Kostyantyn Krysan  |  Min Yu  |  Qingjiao Li  |  Quan Chen  |  Shuli Kang  |  Stacy Park  |  Steven M. Dubinett  |  Wei Wang  |  Wenyuan Li & Xianghong Jasmine Zhou  |  Yonggang Zhou

• Department of Pathology and Laboratory Medicine, David Geffen School of Medicine, University of California at Los Angeles, Los Angeles, CA, 90095, USA  |  Institute for Quantitative and Computational Biosciences, University of California at Los Angeles, Los Angeles, CA, 90095, USA  |  Molecular and Computational Biology, University of Southern California, Los Angeles, CA, 90089, USA  |  Shuli Kang, Qingjiao Li, Quan Chen, Frank Alber & Fengzhu Sun

A Java package for non-invasive cancer diagnosis using methylation profiles of cell-Free DNA. Cancer cells often display aberrant DNA methylation patterns, such as hypermethylation of the promoter regions of tumor suppressor genes and pervasive hypomethylation of intergenic regions. Therefore, DNA methylation is an ideal target for cancer diagnosis in clinical practice. The non-invasive nature of cfDNA methylation profiling makes it a promising strategy for general cancer screening. Usually, the differentially methylated marker genes are identified by comparing methylation profile data from patients with a certain cancer type to healthy controls. This approach takes into accout the methylation profile from the cfDNA of the patient and compares it with the methylation profiles of healthy as well as patients with different types of cancer, in order to accurately detect and identify the type of cancer associated with the patient. The approach used for modeling this is the maximum likelihood of occurence(error rate ~ 0.24) approach and out performs the RF(error rate ~ 0.804) and SVM(error rate ~ 0.816) in terms of error rate. The data under consideration is from "The Cancer Genome Atlas" which consists the methylation profiles of more than 11000 patients.

Input variables : Methylation values
Output Variables : Cancer Type

Visit Model : github.com

Additional links : genomebiology.biomedcentral.com