Dr.VAE: improving drug response prediction via modeling of drug perturbation effects

Abstract

Motivation Individualized drug response prediction is a fundamental part of personalized medicine for cancer. Great effort has been made to discover biomarkers or to develop machine learning methods for accurate drug response prediction in cancers. Incorporating prior knowledge of biological systems into these methods is a promising avenue to improve prediction performance. High-throughput cell line assays of drug-induced transcriptomic perturbation effects are a prior knowledge that has not been fully incorporated into a drug response prediction model yet. Results We introduce a unified probabilistic approach, Drug Response Variational Autoencoder (Dr.VAE), that simultaneously models both drug response in terms of viability and transcriptomic perturbations. Dr.VAE is a deep generative model based on variational autoencoders. Our experimental results showed Dr.VAE to do as well or outperform standard classification methods for 23 out of 26 tested Food and Drug Administration-approved drugs. In a series of ablation experiments we showed that the observed improvement of Dr.VAE can be credited to the incorporation of drug-induced perturbation effects with joint modeling of treatment sensitivity. Availability and implementation Processed data and software implementation using PyTorch (Paszke et al., 2017) are available at: https://github.com/rampasek/DrVAE. Supplementary information Supplementary data are available at Bioinformatics online.

Post-perturbational transcriptional signatures of cancer cell line vulnerabilities

“…A number of studies have presented broad characterizations of the expression-viability relationship by examining responses to drugs with various mechanisms, revealing patterns in gene expression that are associated with fitness loss [18,24]. Another line of work has focused on building predictive models to reliably estimate a cell's future viability response given the transcriptional profile [20]. While these studies have provided beneficial insights for a handful of compounds and cell contexts, there remains a need for a detailed characterization of the landscape of viability-related transcriptional responses across many chemical and genetic reagents.…”

Deriving disease modules from the compressed transcriptional space embedded in a deep autoencoder

“…Chen et al used cancer data and showed that starting from pathways represented as a priori defined hidden nodes, allowed the investigators to explain 88% of variance, which in turn produced an interpretable representation 18 . Recently a few authors have shown that unbiased data-driven compression can learn meaningful representations from unlabeled data, which predicted labeled data of single-cells RNA-seq 19,20 and drug responses 21,22 . These results demonstrate that AEs can use predefined functional representations, and can learn such representations from input data that can be used for other purposes in transfer learning approaches.…”

Enhancing scientific discoveries in molecular biology with deep generative models

“…In a more clinical application, Rampášek et al () learned a VAE on bulk gene expression data sets from cancer cell lines. A small fraction of the data has paired pretreatment and post‐treatment measurements as well as drug response information.…”