PolyLLM: polypharmacy side effect prediction via LLM-based SMILES encodings

Abstract

Polypharmacy, the concurrent use of multiple drugs, is a common approach to treating patients with complex diseases or multiple conditions. Although consuming a combination of drugs can be beneficial in some cases, it can lead to unintended drug-drug interactions (DDI) and increase the risk of adverse side effects. Predicting these adverse side effects using state-of-the-art models like Large Language Models (LLMs) can greatly assist clinicians. In this study, we assess the impact of using different LLMs to predict polypharmacy. First, the chemical structure of drugs is vectorized using several LLMs such as ChemBERTa, GPT, etc., and are then combined to obtain a single representation for each drug pair. The drug pair representation is then fed into two separate models including a Multilayer Perceptron (MLP) and a Graph Neural Network (GNN) to predict the side effects. Our experimental evaluations show that integrating the embeddings of Deepchem ChemBERTa with the GNN architecture yields more effective results than other methods. Additionally, we demonstrated that utilizing complex models like LLMs to predict polypharmacy side effects using only chemical structures of drugs can be highly effective, even without incorporating other entities such as proteins or cell lines, which is particularly advantageous in scenarios where these entities are not available.

Keywords: drug combination; graph neural networks; large language models; polypharmacy side effect; smiles.

FIGURE 1

(a) Distribution of SMILES string lengths for 645 drugs in the dataset. (b) Number of side effects associated with each drug pair in the reformatted dataset.

FIGURE 2

Pipeline of PolyLLM. (a) Drug-pair representations created by combining encoded drug features using language models (b) MLP approach: Multilayer perceptron to predict side effects from drug-pair representations. (c) Graph approach: Bipartite Graph constructed with drug-pair nodes and encoded side-effect features to predict whether or not a drug-pair is associated with side effects, considered as a binary link prediction task.

FIGURE 3

Overview of (a) the Multilayer Perceptron and (b) the Graph Neural Network approaches for predicting drug-pair side effects.

FIGURE 4

AUC scores for various fusion strategies across multiple encoders.