A Web Based Interface for Automatic Molecular Docking and Related Tools
Oral Defence Date:
Monday, December 3, 2018 - 14:30
Profs. Ilmi Yoon and Arno Puder
Molecular docking is widely applied to use a protein structure (a large set of molecules) for ligand (a single molecule) discovery. It is widely used in pharma industry and academia to find new small molecules to modulate the biological activity of proteins, and as such is the one of the earliest steps in drug discovery. tldr (Tools for Ligand Discovery Research) is a web-based system that enables investigators with minimal computing expertise to perform automated molecular docking. tldr, an improvement to DOCK Blaster, a first attempt at web-based docking published in 2009, consists of a series of features (Job Types) that allows users to perform an array of tasks that assist with molecular docking, including data preparation and analysis. Web applications such as DUD-E (A Database of Useful Decoys: Enhanced) and SEA (Similarity Ensemble Approach), both of which were developed at the University of California, San Francisco (UCSF), are implemented as job types, hence making tldr a super-set of ligand discovery and analysis. The application employs a fail-fast approach and provides access to intermediate results of submitted jobs. The innovation lies in the fact that tldr supports Dynamic Job Type Creation, allowing admin users to add support for new job types on the fly, with just a few clicks of the mouse. It also comes with an admin panel to track real time metrics of the system such as number of actively running jobs, CPU/Disk Usage, User Registration and Job Management to name a few. The tool also provides rich visualization features by integrating with UCSF ChimeraX and ZINC. The backend system exposes a programmable REST API, allowing it to be integrated into third-party applications. The end product is a robust, user-friendly and flexible platform that enables agile method development in computational ligand discovery and analysis.
tldr, molecular docking, ligand discovery, pharmaceutical, web application