Feature Allocation Models for Tumor Heterogeneity

• Activity Number: 668
• Type: Invited
• Date/Time: Thursday, August 13, 2015 : 10:30 AM to 12:20 PM
• Sponsor: IMS
• Author(s): Juhee Lee and Peter Mueller* and Subhajit Sengupta and Kamalakar Gulukota and Yuan Ji
• Companies: UC Santa Cruz and The University of Texas at Austin and Northshore University HealthSystem and Northshore University HealthSystem and NorthShore University Health System
• Keywords: Feature allocation ; Indian buffet process ; nonparametric Bayes

Abstract

We propose a feature allocation (mixed membership) model for inference on tumor heterogeneity. We implement model-based Bayesian posterior inference. We jointly model mutations, recorded as variant allele fractions, and copy number variations, using next-generation sequencing data. We demonstrate how the augmented model with both outcomes improves inference over a model for only variant allele fractions. Joint modeling of both. We discuss efficient posterior inference in the proposed models. The proposed models are based on variations of the Indian buffet process and alternatively an implementation of determinantal point processes.