Robust and accurate deconvolution of tumor populations uncovers evolutionary mechanisms of breast cancer metastasis

De novo Prediction of Cell-Drug Sensitivities Using Deep Learning-based Graph Regularized Matrix Factorization

Application of artificial intelligence (AI) in precision oncology typically involves predicting whether the cancer cells of a patient (previously unseen by AI models) will respond to any of a set of existing anticancer drugs, based on responses of previous training cell samples to those drugs. To expand the repertoire of anticancer drugs, AI has also been used to repurpose drugs that have not been tested in an anticancer setting, i.e., predicting the anticancer effects of a new drug on previously unseen cancer cells de novo. Here, we report a computational model that addresses both of the above tasks in a unified AI framework. Our model, referred to as deep learning-based graph regularized matrix factorization (DeepGRMF), integrates neural networks, graph models, and matrix-factorization techniques to utilize diverse information from drug chemical structures, their impact on cellular signaling systems, and cancer cell cellular states to predict cell response to drugs. DeepGRMF learns embeddings of drugs so that drugs sharing similar structures and mechanisms of action (MOAs) are closely related in the embedding space. Similarly, DeepGRMF also learns representation embeddings of cells such that cells sharing similar cellular states and drug responses are closely related. Evaluation of DeepGRMF and competing models on Genomics of Drug Sensitivity in Cancer (GDSC) and Cancer Cell Line Encyclopedia (CCLE) datasets show its superiority in prediction performance. Finally, we show that the model is capable of predicting effectiveness of a chemotherapy regimen on patient outcomes for the lung cancer patients in The Cancer Genome Atlas (TCGA) dataset*.

OUP accepted manuscript

Motivation

Identifying cell types and their abundances and how these evolve during tumor progression is critical to understanding the mechanisms of metastasis and identifying predictors of metastatic potential that can guide the development of new diagnostics or therapeutics. Single-cell RNA sequencing (scRNA-seq) has been especially promising in resolving heterogeneity of expression programs at the single-cell level, but is not always feasible, e.g. for large cohort studies or longitudinal analysis of archived samples. In such cases, clonal subpopulations may still be inferred via genomic deconvolution, but deconvolution methods have limited ability to resolve fine clonal structure and may require reference cell type profiles that are missing or imprecise. Prior methods can eliminate the need for reference profiles but show unstable performance when few bulk samples are available.

Results

In this work, we develop a new method using reference scRNA-seq to interpret sample collections for which only bulk RNA-seq is available for some samples, e.g. clonally resolving archived primary tissues using scRNA-seq from metastases. By integrating such information in a Quadratic Programming framework, our method can recover more accurate cell types and corresponding cell type abundances in bulk samples. Application to a breast tumor bone metastases dataset confirms the power of scRNA-seq data to improve cell type inference and quantification in same-patient bulk samples.

Availability and implementation

Source code is available on Github at https://github.com/CMUSchwartzLab/RADs.

Assessing the contribution of tumor mutational phenotypes to cancer progression risk

Cancer occurs via an accumulation of somatic genomic alterations in a process of clonal evolution. There has been intensive study of potential causal mutations driving cancer development and progression. However, much recent evidence suggests that tumor evolution is normally driven by a variety of mechanisms of somatic hypermutability, which act in different combinations or degrees in different cancers. These variations in mutability phenotypes are predictive of progression outcomes independent of the specific mutations they have produced to date. Here we explore the question of how and to what degree these differences in mutational phenotypes act in a cancer to predict its future progression. We develop a computational paradigm using evolutionary tree inference (tumor phylogeny) algorithms to derive features quantifying single-tumor mutational phenotypes, followed by a machine learning framework to identify key features predictive of progression. Analyses of breast invasive carcinoma and lung carcinoma demonstrate that a large fraction of the risk of future clinical outcomes of cancer progression-overall survival and disease-free survival-can be explained solely from mutational phenotype features derived from the phylogenetic analysis. We further show that mutational phenotypes have additional predictive power even after accounting for traditional clinical and driver gene-centric genomic predictors of progression. These results confirm the importance of mutational phenotypes in contributing to cancer progression risk and suggest strategies for enhancing the predictive power of conventional clinical data or driver-centric biomarkers.

Computational deconvolution to estimate cell type-specific gene expression from bulk data

Computational deconvolution is a time and cost-efficient approach to obtain cell type-specific information from bulk gene expression of heterogeneous tissues like blood. Deconvolution can aim to either estimate cell type proportions or abundances in samples, or estimate how strongly each present cell type expresses different genes, or both tasks simultaneously. Among the two separate goals, the estimation of cell type proportions/abundances is widely studied, but less attention has been paid on defining the cell type-specific expression profiles. Here, we address this gap by introducing a novel method Rodeo and empirically evaluating it and the other available tools from multiple perspectives utilizing diverse datasets.