Title: Identifying synthetic lethality based on a knowledge graph and protein interaction network

Authors: Canling Huang; Fuheng Xiao; Qian Deng; Jing Zhang; Yan Wang; Ali Chen; Wei Xiao; Zhanchao Li

Addresses: School of Chemistry and Chemical Engineering, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Center for Drug Research and Development, Guangdong Provincial Key Laboratory of Advanced Drug Delivery System, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Key Laboratory of Glucolipid Metabolic Disorder, Ministry of Education, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Department of Nephrology, Integrated Hospital of Traditional Chinese Medicine, Southern Medical University, Guangzhou, 510315, China ' School of Chemistry and Chemical Engineering, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Center for Drug Research and Development, Guangdong Provincial Key Laboratory of Advanced Drug Delivery System, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Key Laboratory of Glucolipid Metabolic Disorder, Ministry of Education, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Department of Nephrology, Integrated Hospital of Traditional Chinese Medicine, Southern Medical University, Guangzhou, 510315, China ' School of Chemistry and Chemical Engineering, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Center for Drug Research and Development, Guangdong Provincial Key Laboratory of Advanced Drug Delivery System, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Key Laboratory of Glucolipid Metabolic Disorder, Ministry of Education, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Department of Nephrology, Integrated Hospital of Traditional Chinese Medicine, Southern Medical University, Guangzhou, 510315, China ' School of Chemistry and Chemical Engineering, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Center for Drug Research and Development, Guangdong Provincial Key Laboratory of Advanced Drug Delivery System, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Key Laboratory of Glucolipid Metabolic Disorder, Ministry of Education, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Department of Nephrology, Integrated Hospital of Traditional Chinese Medicine, Southern Medical University, Guangzhou, 510315, China ' School of Chemistry and Chemical Engineering, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Center for Drug Research and Development, Guangdong Provincial Key Laboratory of Advanced Drug Delivery System, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Key Laboratory of Glucolipid Metabolic Disorder, Ministry of Education, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Department of Nephrology, Integrated Hospital of Traditional Chinese Medicine, Southern Medical University, Guangzhou, 510315, China ' School of Chemistry and Chemical Engineering, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Center for Drug Research and Development, Guangdong Provincial Key Laboratory of Advanced Drug Delivery System, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Key Laboratory of Glucolipid Metabolic Disorder, Ministry of Education, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Department of Nephrology, Integrated Hospital of Traditional Chinese Medicine, Southern Medical University, Guangzhou, 510315, China ' School of Chemistry and Chemical Engineering, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Center for Drug Research and Development, Guangdong Provincial Key Laboratory of Advanced Drug Delivery System, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Key Laboratory of Glucolipid Metabolic Disorder, Ministry of Education, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Department of Nephrology, Integrated Hospital of Traditional Chinese Medicine, Southern Medical University, Guangzhou, 510315, China ' School of Chemistry and Chemical Engineering, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Center for Drug Research and Development, Guangdong Provincial Key Laboratory of Advanced Drug Delivery System, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Key Laboratory of Glucolipid Metabolic Disorder, Ministry of Education, Guangdong Pharmaceutical University, Guangzhou, 510006, China; Department of Nephrology, Integrated Hospital of Traditional Chinese Medicine, Southern Medical University, Guangzhou, 510315, China

Abstract: Synthetic lethality (SL) is an effective way of treating cancer by using special genetic interactions to inhibit the partner of the mutated gene, causing cancer cells to die while leaving normal cells intact. Although the traditional experimental method has high accuracy in identifying SL pairs, it usually suffers from low efficiency. Therefore, a novel computational method was proposed to identify potential SL pairs. The developed method used knowledge graph convolution networks to learn higher-order structural and semantic information for embedding representation. Meanwhile, topological features of the protein interaction network were extracted by deep random walks to characterise SL pairs. Finally, extreme gradient boosting was utilised to construct a model for recognising the potential SL pairs. Based on the 5-fold cross-validation test, areas under the receiver operating characteristic curve reached 0.9874. Among the top ten predicted SL pairs, three were verified by the database and literature.

Keywords: synthetic lethality pair; cancer; gene; knowledge graph; protein interaction network.

DOI: 10.1504/IJDMB.2026.154715

International Journal of Data Mining and Bioinformatics, 2026 Vol.30 No.3/4, pp.379 - 404

Received: 23 Aug 2024
Accepted: 11 Aug 2025
Published online: 10 Jul 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article