Variational Methods for Detecting Copy Number Alterations
Guillaume Alain · 2009
in numerous human diseases, including cancer. Various techniques exist to identify them, but the problem is far from solved and the prohibitive computational costs rule out many approaches. We show how the current state-of-the-art model hmmmix can be trained using variational methods with minimal assumptions. We show how this allows for partial patients clustering and how it partly addresses the difficult issue of determining the number of clusters to use. We compare our technique with the current best and show how it can handle very large datasets. iii Table of Contents