=Paper=
{{Paper
|id=Vol-1327/3
|storemode=property
|title=Drug Target Prediction for Colorectal Cancer by Combining Ontology and Network Approaches
|pdfUrl=https://ceur-ws.org/Vol-1327/icbo2014_paper_15.pdf
|volume=Vol-1327
|dblpUrl=https://dblp.org/rec/conf/icbo/TaoSZCX14
}}
==Drug Target Prediction for Colorectal Cancer by Combining Ontology and Network Approaches==
https://ceur-ws.org/Vol-1327/icbo2014_paper_15.pdf
ICBO 2014 Proceedings
Drug target prediction for colorectal cancer by
combining ontology and network approaches
Cui Tao1*, Jingchun Sun1*, W.Jim Zheng1, Junjie Chen2, Hua Xu1#
1
Center for Computational Biomedicine, School of Biomedical informatics, University of Texas
Health Science Center at Houston and 2Department of Experimental Radiation Oncology, The
University of Texas MD Anderson Cancer Center, Houston, Houston, TX 77030, USA
Drug discovery is a time-consuming and expensive process, associated genes were collected from the Cancer Gene
especially for complex diseases. In the last decade, target- Census, the Online Mendelian Inheritance in Man (OMIM),
based methods for drug discovery have become more common and the Genetic Association database (GAD). Among the 113
and effective comparing to traditional observation-based drug genes, 15 were selected as the promising drug targets based on
discovery. Recently, computational approaches for target their neighborhood of CRC disease genes in the context of one
prediction and drug repurposing have become more common human PPI network. For example, three of them encode
and effective compared to traditional observation-based drug known CRC drug targets (EGFR, TOP1, and VEGFA). EGFR
discovery. However, since the data about underlying is targeted by the drugs cetuximab and panitumumab, TOP1 is
molecular mechanisms of drugs distribute among different targeted by the drug irinotecan, and VEGFA is targeted by the
knowledge domains and different databases, it is very aflibercept and bevacizumab. Additionally, CCND1 (cyclin
challenging to design effective strategies to discover novel D1) is targeted by the drug arsenic trioxide, which is used to
drug targets and propose successful drug repurposing. To treat leukemia; and PTGS2 (prostaglandin-endoperoxide
alleviate this problem, we propose a computational framework synthase 2) is targeted by multiple drugs such as lenalidomide,
to integrate complex relationship among different types of which is used for treating lymphoma, pomalidomide and
data and infer the potential drug targets by using the semantic thalidomide, both of which are used for treating multiple
web technology, and to improve performance through network myeloma and other plasma cell neoplasms. Among the
neighborhood effect modeling. In this study, we utilize the remaining genes, CCND1 is a well-recognized oncongene that
colorectal cancer (CRC) as a proof-of-concept use case to is amplified and/or overexpressed in a substantial proportion
evaluate the approach. of human cancers including colon cancer, prostate cancer and
breast cancer. Therefore, it might be a promising anti-cancer
We first constructed a CRC ontology including drugs, therapeutic target. The gene PTGS2 encodes prostaglandin
diseases, genes, pathways, SNPs, and their relations from the G/H synthase-2, which catalyses the first two steps in the
PharmGKB. The PharmGKB is a pharmacogenomics metabolism of arachadonic acid. It is overexpressed in many
knowledge resource, which collects, curates, and disseminates types of cancer such as colon, stomach, breast, and lung.
knowledge about the impact of human genetic variation on Additionally, it has three variations with pharmacogenomic
drug responses through the following activities. On top of this significance (rs20417, rs5275, and rs689466). Therefore, its
PharmGKB ontology, we further specified drug target inhibition with drugs such as aspirin, celecoxib, and ibuprofen
prediction ontology for CRC. A new CRC Drug class has been can be used in the prevention and treatment of cancer.
created to serve as the basis of our drug target inference. We
further specified OWL DL (Description Logic) rules to infer Therefore, in this study, we developed a unique computational
possible CRC drug target genes. Starting from eight FDA- framework to integrate the ontology technology and network
approved CRC drugs and the CRC ontology, we inferred 113 neighborhood modeling for drug target prediction. The results
potential CRC drug targets. To prioritize the most promising demonstrate that this framework indeed identifies the novel
targets among the ontology-driven CRC potential drug targets, targets. Besides, we see many opportunities to improve upon
we utilized their relationships with CRC-associated genes in the basic design of this integration, including integrating more
the context of the human protein-protein interaction (PPI) relationship from multiple data sources during the ontology
network. Starting from these 113 potential targets, we ranked construction, application of score strategies in the ontology
them based on the fraction of CRC-associated genes in their reference, and the combination of more network properties
neighborhood at the first, second, and third shortest-path into the gene ranking.
distances and then integrated the three sets of ranking scores
using a robust rank aggregation (RRA) method. These CRC-
67