A genome-wide positioning systems network algorithm for in silico drug repurposing
Feixiong Cheng, Weiqiang Lu, Chuang Liu, Jiansong Fang, Yuan Hou, Diane E. Handy, Ruisheng Wang, Yuzheng Zhao, Yi Yang, Jin Huang, David E. Hill, Marc Vidal, Charis Eng & Joseph Loscalzo
https://doi.org/10.1038/s41467-019-10744-6
After the completion of the human genome project in 2003, there has been unexpected enthusiasm for how genetics and genomics would inform drug discovery and development1. Although it is in its infancy, the use of genomics in the drug discovery and development pipeline has generated some successes2,3. For example, proprotein convertase subtilisin/kexin type 9 (PCSK9), first discovered by human genetics studies in 2003, has generated great interest in genomics-informed drug discovery in cardiovascular disease4. However, the overall clinical efficacy of genome-derived approved drugs has remained limited owing to the heterogeneity of complex diseases.
Drug development in the genomics era has become a highly integrated systems pipeline in which complementary multi-omics and computational methods are used3. Recent technological and computational advances in genomics and systems biology have now made it possible to identify new druggable targets and therapeutic agents by uniquely targeting cancer type-specific mechanisms (e.g., perturbed pathways in the disease module) that cause or contribute to human disease5,6,7,8. For example, network-based approaches have offered possibilities for drug repurposing8, target identification9, and combination therapy10 by quantifying the proximity of drug targets and disease proteins in the human protein–protein interactome. It remains unclear as to how generalizable these network-based disease module identification methodologies are in exploiting the wealth of massive multi-omics data from genomics studies and offering novel insights into cancer type-specific mechanisms. Were they to be successful, we would ultimately be able to target precisely human disease pathways, promoting the development of precision medicine.
In this study, we present a novel network-based disease module identification and in silico drug repurposing methodology, denoted the Genome-wide Positioning Systems network (GPSnet) algorithm. Specifically, we demonstrate the feasibility of individualized disease module identification by integrating large-scale DNA sequencing and transcriptome (RNA-seq) profiling across approximately 5000 human tumor genomes to the human protein–protein interactome. We find that gene expression of disease modules identified by GPSnet can predict drug responses in cancer cell lines with high accuracy. Importantly, we show that GPSnet can be used for in silico drug repurposing by uniquely targeting the specific disease module (i.e., cancer type-specific module) through combining network proximity measure and gene-set enrichment analysis (GSEA) approaches. Furthermore and importantly, we validate these network-based predictions experimentally. From a translational perspective, if broadly applied, GPSnet offers a powerful network-based tool for target identification and drug repurposing. We believe this approach can minimize the translational gap between genomics studies and drug development, currently a significant bottleneck in precision medicine.