Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Drug Selection via Joint Push and Learning to Rank

Yicheng He, Junfeng Liu, Xia Ning

TL;DR

This work tackles cancer drug selection by casting it as a ranking problem and introducing pLETORg, a genomics-regularized joint push-and-learning-to-rank method. It learns latent representations for cell lines and drugs to score and rank drugs within each cell line, explicitly pushing sensitive drugs to the top and enforcing correct ordering among sensitive drugs, while leveraging genomics to regularize cell-line latent factors. The approach outperforms strong baselines on key metrics for prioritizing sensitive drugs, especially under small-sample or transductive scenarios, and provides insights into the interpretable structure of drug and cell-line latent vectors. These results suggest significant potential for improving drug prioritization in precision oncology, with avenues for personalization and localized feature selection in future work.

Abstract

Selecting the right drugs for the right patients is a primary goal of precision medicine. In this manuscript, we consider the problem of cancer drug selection in a learning-to-rank framework. We have formulated the cancer drug selection problem as to accurately predicting 1). the ranking positions of sensitive drugs and 2). the ranking orders among sensitive drugs in cancer cell lines based on their responses to cancer drugs. We have developed a new learning-to-rank method, denoted as pLETORg , that predicts drug ranking structures in each cell line via using drug latent vectors and cell line latent vectors. The pLETORg method learns such latent vectors through explicitly enforcing that, in the drug ranking list of each cell line, the sensitive drugs are pushed above insensitive drugs, and meanwhile the ranking orders among sensitive drugs are correct. Genomics information on cell lines is leveraged in learning the latent vectors. Our experimental results on a benchmark cell line-drug response dataset demonstrate that the new pLETORg significantly outperforms the state-of-the-art method in prioritizing new sensitive drugs.

Drug Selection via Joint Push and Learning to Rank

TL;DR

This work tackles cancer drug selection by casting it as a ranking problem and introducing pLETORg, a genomics-regularized joint push-and-learning-to-rank method. It learns latent representations for cell lines and drugs to score and rank drugs within each cell line, explicitly pushing sensitive drugs to the top and enforcing correct ordering among sensitive drugs, while leveraging genomics to regularize cell-line latent factors. The approach outperforms strong baselines on key metrics for prioritizing sensitive drugs, especially under small-sample or transductive scenarios, and provides insights into the interpretable structure of drug and cell-line latent vectors. These results suggest significant potential for improving drug prioritization in precision oncology, with avenues for personalization and localized feature selection in future work.

Abstract

Selecting the right drugs for the right patients is a primary goal of precision medicine. In this manuscript, we consider the problem of cancer drug selection in a learning-to-rank framework. We have formulated the cancer drug selection problem as to accurately predicting 1). the ranking positions of sensitive drugs and 2). the ranking orders among sensitive drugs in cancer cell lines based on their responses to cancer drugs. We have developed a new learning-to-rank method, denoted as pLETORg , that predicts drug ranking structures in each cell line via using drug latent vectors and cell line latent vectors. The pLETORg method learns such latent vectors through explicitly enforcing that, in the drug ranking list of each cell line, the sensitive drugs are pushed above insensitive drugs, and meanwhile the ranking orders among sensitive drugs are correct. Genomics information on cell lines is leveraged in learning the latent vectors. Our experimental results on a benchmark cell line-drug response dataset demonstrate that the new pLETORg significantly outperforms the state-of-the-art method in prioritizing new sensitive drugs.

Paper Structure

This paper contains 42 sections, 14 equations, 9 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Literature Review on Learning to Rank
  3. Methods
  4. Drug Scoring
  5. Pushing up Sensitive Drugs
  6. Ranking among Sensitive Drugs
  7. Overall Optimization Problem for $\mathop{\mathtt{pLETORg}}\limits$
  8. Materials
  9. Dataset and Experimental Protocol
  10. Experimental Setting
  11. N-Fold Cross Validation
  12. Leave-Out Validation
  13. Sensitivity Labeling Scheme
  14. Labeling Scheme for N-Fold Cross Validation
  15. Labeling Scheme for Testing New Cell Lines
  16. ...and 27 more sections

Figures (9)

  • Figure 1: $\mathop{\mathtt{pLETORg}}\limits$ scheme overview
  • Figure 2: Exemplar cell line response score distribution
  • Figure 3: Data split for 5-fold cross validation
  • Figure 4: Data split for testing new cell lines
  • Figure 5: Performance of $\mathop{\mathtt{pLETORg}}\limits$ w.r.t. the Push Parameter $\alpha$
  • ...and 4 more figures