Deep learning-based Integration of Multi-omics Data for Personalized Drug Response Prediction
- Abstract
- Individual variations in drug response pose a significant challenge to effective cancer treatment. Our study presents a predictive model that uses multi-omics data to describe drug responses according to individual patient profiles. The model’s main idea is the integration of individual gene sets for each cell line, capturing the distinct pathological signatures of different diseases. It integrates gene expression and DNA methylation data, taking into account the well-established inverse correlation between these factors (where methylation often down-regulates gene expression), to provide insight into the epigenetic mechanisms that influence drug response. Furthermore, we investigated the efficacy of integrating multi-omics data for drug response prediction by applying transformer models developed in multimodal tasks. Finally, our model outperformed state-of-the-art methods in predicting drug response, demonstrating the potential of our approach to contribute to personalised cancer therapy. This advance supports the goal of developing treatments that are specific to an individual’s genetic profile.
- Author(s)
- Juyoung Kang
- Issued Date
- 2024
- Type
- Thesis
- URI
- https://scholar.gist.ac.kr/handle/local/19062
- 공개 및 라이선스
-
- 파일 목록
-
Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.