Bilkent University
Department of Computer Engineering
CS 590/690 SEMINAR
Integrating Multi-Modal Knowledge Graph-Based Embeddings for Improved Prediction of Drug Synergy
M. Buğra Kurnaz
Master Student
(Supervisor: Assoc.Prof.Ercüment Çiçek)
Computer Engineering Department
Bilkent University
Abstract: Accurately predicting drug synergy is vital for advancing combination therapies, particularly in complex diseases where single-drug treatments frequently prove inadequate. Building on existing approaches that employs knowledge graphs encompassing drugs, cell lines, tissues, and genes—this study will introduce an integrative framework that broadens the biomedical context. The proposed method will integrate the GenomicKB knowledge graph to study drug-drug interaction while including richer genomic information, such as copy number variations, chromosomal regions, and expanded cell line interactions. Additionally, the DrugBank dataset and textual descriptions detailing drug mechanisms of action will be utilized to produce detailed embeddings on drugs. These multi-modal embeddings are then fed into a neural network aimed at predicting synergistic drug combinations with higher accuracy. By merging diverse biological and pharmacological data sources, the approach seeks to reveal deeper insights into drug interactions, predicting whether drug pairs provide synergistic or antagonistic effect.
DATE: March 24, Monday @ 14:50 Place: EA 502