Equivariant Graph Attention Networks with Structural Motifs for Predicting Cell Line -Specific Synergistic Drug Combinations

Overview

Cancer is the second leading cause of death, behind heart disease, with chemotherapy as one of the primary forms of treatment. As a result, researchers are turning to drug combination therapy. However, current methods of screening such as in vivo and in vitro are inefficient due to time and monetary costs. In silico methods have become increasingly important, but current methods of screening drug combinations are inaccurate and generalize poorly. In this paper, I employ a geometric deep-learning model based on a rotational and translation equivariant graph attention network with structural motifs. Additionally, the gene expression of the cancer cell line is used as an input to a multi -layer perception to classify the synergistic drug combinations based on ea ch cancer cell line. I compared the proposed geometric deep learning framework to state -of-the-art methods and achieved greater performances on all 12 benchmark tasks performed on the DrugComb dataset. Specifically, the proposed framework performs superior to other state-of-the-art methods by greater than an accuracy of 28%. Based on these results, I believe that the equivariant graph attention network's capabilities of learning geometric data account for large performance improvements. The model's ability to generalize to foreign drugs is thought to be due to the structural motifs better representing the molecule. Overall, I believe that the proposed equivariant geometric deep-learning framework is an effective tool for virtually screening anticancer drug combinations for further validation in a wet lab environment.

Competition history

• JSHS 2024 Category not listed

Resources

• National JSHS abstract book ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.