Machine-Learning Aided Drug Design - An Extensive Assessment of the e-LEA3D Web Server

o o Guiragos et al. conducted and published a thorough examination of digital ligand-protein modeling and binding tools to assist in drug discovery. In this, first a research project to generate and evaluate machine learning and empirical scoring functions was undertaken. After, these scoring functions were applied to ligands generated by the e-LEA3D web server which could potentially result in new drug discovery. Of the several dozen ligands generated and explored, ten were found to meet the pharmacokinetics, drug likeness, and binding affinity thresholds to pursue further assessment of their potential (to be plausibly useful medicinal drugs).

o This research project was perhaps the most holistic and intensive undertaking of my undergraduate career as it featured a large collaborative effort from the Schools of IAS, CS, and Biology. It was a project which, when chosen, undeniably interesting but completely out of my known fields of expertise. Before bioinformatics, my knowledge of biology was strictly from pre-university. During research I was able to get caught up and clearly applied freshly learned machine-learning and scripting techniques to automate our research goals. It is a project that clearly demonstrates a clear delineation and segregation of roles within a research group and the usefulness of that approach in generating useful experiment data. Further, it stressed my ability to learn a subject quickly (and it partially shows in the final paper). Overall, I feel that this project more than any other meets all five learning objectives of the school of IAS in its collaborative nature.