Understanding the Evolution of Tumours Using Hybrid Deep Generative Models

Abstract

Understanding both the evolutionary dynamics and subpopulation or subclonal structure that impacts tumour progression has important clinical implications for patients. However, deconvoluting subclonal structure and performing evolutionary parameter inference have largely been treated as two independent or step-wise tasks. Here, we show that combining stochastic simulations with hybrid deep generative models enables joint inference of subclonal structure and evolutionary parameters. Ultimately, by jointly learning these two problems, we show that our proposed approach leads to improved performance across a multitude of cancer evolution tasks including, but not limited to, detecting subclones, quantifying subclone frequency, and estimating mutation rate. As an additional benefit, we also show that hybrid deep generative models also provide substantial reductions in inference time relative to existing methods.