Drug sensitivity prediction across cancer types using graph isomorphism networks and biological pathway features: A dual-branch deep learning approach
PLOS One · 11.08.2026 · 21 (8) · e0354669
DOI: 10.1371/journal.pone.0354669 ↗Kurz gefasst
Fachartikel in der Zeitschrift PLOS One, online erschienen am 11.08.2026. Verfasst von Shuang Li und 6 weiteren Autor:innen. Beteiligte Einrichtung: Institute of Medical Technology, Luoyang Polytechnic, Luoyang, Henan, China. Hauptthemen laut MEDLINE: Deep Learning, Neoplasms, Antineoplastic Agents, Drug Resistance, Neoplasm. Die Arbeit nennt 25 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Drug sensitivity prediction is an important issue within the precision medicine field. IC50, which is the molar drug dose needed to decrease the viability of cells by half compared to the drug-free control, is the main pharmacodynamics parameter used for drug sensitivity analysis in large-scale pharmacogenomics screenings. Computational estimation of IC50s based on molecular and genomic factors significantly reduces costs associated with experiments for measuring cell viability and allows for accelerating the process of drug discovery. Traditional methods of IC50 calculation do not allow integrating the three-dimensional chemical structure of drugs and the biological context of particular cell lines, resulting in suboptimal model performance when using different pharmacogenomics data sources. In this work, we propose an innovative dual-branch approach based on Graph Isomorphism Network (GIN) drug representations coupled with a Multilayer Perceptron (MLP) for 50-dimensional ssGSEA pathway activities calculated from CCLE gene expression. After training on cell-line-drug pair combinations from the Genomics of Drug Sensitivity in Cancer 2 (GDSC2) dataset across various cancers, the proposed GIN+Pathway MLP model attains an R 2 of 0.8553 and a Pearson Correlation Coefficient (PCC) of 0.9249 on the testing split of the same dataset. In a variant ablation study of six variants, we find that eliminating the pathway MLP component lowers the R 2 value by more than 0.15, thus proving the importance of biological features in the two-branch model. The performance of our proposed model exceeds benchmark scores for models such as GraphDRP (PCC = 0.870, R 2 = 0.756) and DeepCDR (PCC = 0.847, R 2 = 0.720) when tested on the same GDSC2 dataset.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Substanzen
Autor:innen und Institutionen
- Institute of Medical Technology, Luoyang Polytechnic, Luoyang, Henan, China.
- Zhongyuan Scholars Workstation of Henan, Luoyang Polytechnic, Luoyang, Henan, China.
- Henan Engineering Research Center, Key Immunological Biomaterials, Luoyang Polytechnic, Luoyang, Henan, China.
- Luoyang Key Laboratory of POCT Diagnosis Technology, Luoyang Polytechnic, Luoyang, Henan, China.
ORCID-Profile: Weikai Zhang
Zitieren
APA
Li, S., Yang, Q., Shen, F., Chen, W., Zhang, S., Dong, X., & Zhang, W. (2026). Drug sensitivity prediction across cancer types using graph isomorphism networks and biological pathway features: A dual-branch deep learning approach. PLOS One, 21(8), e0354669. https://doi.org/10.1371/journal.pone.0354669
Vancouver
Li S, Yang Q, Shen F, Chen W, Zhang S, Dong X, et al. Drug sensitivity prediction across cancer types using graph isomorphism networks and biological pathway features: A dual-branch deep learning approach. PLOS One. 2026;21(8):e0354669. doi:10.1371/journal.pone.0354669
BibTeX
@article{Li2026,
author = {Li, Shuang and Yang, Quanzhong and Shen, Feifei and Chen, Wei and Zhang, Shuya and Dong, Xinyi and Zhang, Weikai},
title = {Drug sensitivity prediction across cancer types using graph isomorphism networks and biological pathway features: A dual-branch deep learning approach},
journal = {PLOS One},
year = {2026},
volume = {21},
number = {8},
pages = {e0354669},
doi = {10.1371/journal.pone.0354669}
}
EUVIMED