Bulk RNA sequencing (Bulk RNA-seq) and single-cell RNA sequencing (scRNA-seq) techniques are practical tools for measuring gene expression levels from biopsy samples. The scRNA-seq quantifies the gene expression levels of individual cells, while the bulk RNA-seq measures the average transcriptional profiling of a population of cells. The inference of drug response sensitivities for individual cells can provide detailed and novel insights to understand the mechanism of heterogenous anti-cancer responses and drug resistance at cellular resolution. Unfortunately, the pharmacogenomic information associated with their corresponding single cell is often limited. In contrast, cancer cell lines are still the most widely applied laboratory models for oncology study and anti-cancer drug discovery so far. Therefore, we proposed a transfer learning framework for single-cell drug response inference. This framework integrates domain adaptation strategy to learn cell line bulk pharmacogenomics and transfers the learned knowledge to infer drug response sensitivities at single cells prior to treatment. We have also adopted a widely used model interpretability algorithm (for machine learning models) to infer the subset of genes that contribute the most to drug sensitivity prediction. In addition, our model offers a new perspective on anti-cancer drug combinations. The identified drug sensitivity biomarkers reveal insights into the tumor heterogeneity and treatment at cellular resolution.

Speaker: Mr Zetian ZHENG
Date: 8 January 2024 (Mon)
Time: 3:00pm - 04:00pm
Poster: Click Here