Machine Learning for Detecting and Analyzing Chromoanagenesis Events
摘要
A comprehensive analysis of chromoanagenesis pan-cancer features is crucial for a broad and deep understanding of the phenomena. In this chapter, we describe a cancer-type agnostic machine-learning algorithm for detecting chromoanagenesis. We leveraged data from The Pan-Cancer Analysis of Whole Genome (PCAWG) and The Cancer Genome Atlas (TCGA) to construct and test a predictive algorithm for chromoanagenesis detection based on CNA data, with an accuracy of 86%. This algorithm was applied to analyze data from over 10,000 TCGA cancer patients. The analysis identified cancer-type specific chromoanagenesis characteristics and revealed distinct sets of genes impacted by chromoanagenesis versus non-chromoanagenesis tumorigenesis.