CBRC congratulates Lara Monagel on securing sixth place in the IBDAA Competition after going through several stages among 103,000 participants.
Lara Monagel is a 12th grade student at the Dar Jana school in Jeddah, Saudi Arabia. She worked with Prof. Xin Gao in the Structural and Functional Bioinformatics Group. She is a student of the Saudi Research Science Institute (SRSI) program and was working on building an efficient machine learning model that predicts drug-target interactions with a high accuracy. She is mentored by CBRC's Ph.D. candidate Maha A. Thafar.
SRSI is a rigorous summer research program designed for 11th grade students from high schools throughout Saudi Arabia. Students work under the guidance of world-renowned KAUST faculty, while participating in holistic activities that combine theory courses, hands-on research, and co- and extracurricular events to encourage greater teamwork and leadership.
Lara will represent the Kingdom in the 2022 Regeneron International Science and Engineering Fair (ISEF) that will take place in Atlanta, Georgia, between May 7th and 13th. CBRC is beyond proud of her incredible hard work and dedication.
The research Abstract:
The lack of treatments for many diseases increases the demand for new drugs discovery. For this reason, the drug repositioning (DR) strategy is widely used, which is the process of finding new therapeutic purposes for existing drugs. Drug-Target Interaction (DTI) prediction is a crucial step in DR. The computational methods to predict the DTIs exclude a significant number of experiments on false interactions. The utilization of computational methods is due to their reliability and high success rate. Accordingly, it speeds up the novel DTI discovery while reducing the expenses to a reasonable cost. In this study, a network-based method is developed to predict novel DTIs using machine learning and a graph embedding technique namely Node2vec, with high prediction performance. Our method constructs a heterogeneous network by integrating three graphs (drug-drug similarity, target-target similarity, and DTIs) that are given within the used dataset. Then it applies a graph embedding technique on this network to auto-generate features representation for each drug and target. Moreover, we implement a fusion function to obtain features for each drug-target pair. Finally, we evaluate our method's performance in drug-target link prediction using several benchmark datasets by utilizing multiple classifiers. Compared to the state-of-the-art methods, our method achieved the highest accuracy for each dataset, Nuclear receptor: 98%, G Protein-coupled receptor: 99%, Ion channel: 99%, and Enzyme: 99%. For future directions, we will utilize the best-obtained model to predict novel DTIs. These newly predicted DTIs will be verified using reliable databases and published research.
