MPRDR: A Multi-Path Relational Drug Repurposing Framework Grounded in Graph-Theoretic Principles
International Journal of Drug Delivery Technology · 23.04.2026 · 16 (17s)
DOI: 10.25258/ijddt.16.17s.48 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Drug Delivery Technology, online erschienen am 23.04.2026. Verfasst von Pharsana Parveen M und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Current GNN-based drug repurposing algorithms rate drug-disease relationships using learned embedding proximity, which identifies the patterns of co-occurrence but is neither mechanistically interpretable nor cold-start. We introduce MPRDR (Multi-Path Relational Drug Repurposing), a GNN architecture based on three graph theoretical concepts: typed path algebra [1] to support multi-hop mechanistic evidence, Turan theorems cohesion of drug modules [2], and effective resistance structural robustness [3,4]. In contrast to TxGNN [5] and GDRnet [6], MPRDR precomputes three interpretable scores of topology of PrimeKG [7] and trains a GNN to combine these structural prioritizations with learned representations and replacing uncalibrated sigmoid outputs with a graph-theoretic confidence score. MPRDR performs competitively or better than previous experiments when evaluated using well-known standard test sets such as TxGNN and GDRnet. MPRDR demonstrates additional features compared to existing databases by being able to predict cold-start results, generate drug combination synergy scores, and provide mechanistic explanations.
Zitieren
APA
M, P. P., & A, S. A. M. (2026). MPRDR: A Multi-Path Relational Drug Repurposing Framework Grounded in Graph-Theoretic Principles. International Journal of Drug Delivery Technology, 16(17s). https://doi.org/10.25258/ijddt.16.17s.48
Vancouver
M PP, A SAM. MPRDR: A Multi-Path Relational Drug Repurposing Framework Grounded in Graph-Theoretic Principles. International Journal of Drug Delivery Technology. 2026;16(17s). doi:10.25258/ijddt.16.17s.48
BibTeX
@article{M2026,
author = {M, Pharsana Parveen and A, Stanis Arul Mary},
title = {MPRDR: A Multi-Path Relational Drug Repurposing Framework Grounded in Graph-Theoretic Principles},
journal = {International Journal of Drug Delivery Technology},
year = {2026},
volume = {16},
number = {17s},
doi = {10.25258/ijddt.16.17s.48}
}
EUVIMED