Modern drug discovery relies on in-silico computational simulations such as molecular docking. Molecular docking models biochemical interactions to predict where and how two molecules would bind. The results of large-scale molecular docking simulations can provide valuable insight into the relationship between two molecules. This is useful to a biomedical scientist before conducting in-vitro or in-vivo wet-lab experiments. Although this ˝eld has seen great advancements, feedback from biomedical scientists shows that there is a need for storage and further analysis of molecular docking results. To meet this need, biomedical scientists need to have access to computing, data, and network resources, and require speci˝c knowledge or skills they might lack.
Therefore, a conceptual framework speci˝cally tailored to enable biomedical scientists to reuse molecular docking results, and a methodology which uses regular input from scientists, has been proposed. The framework is composed of 5 types of elements and 13 interfaces. The methodology is light and relies on frequent communication between biomedical sciences and computer science experts, speci˝ed by particular roles. It shows how developers can bene˝t from using the framework which allows them to determine whether a scenario ˝ts the framework, whether an already implemented element can be reused, or whether a newly proposed tool can be used as an element.
Three scenarios that show the versatility of this new framework and the methodology based on it, have been identi˝ed and implemented. A methodical planning and design approach was used and it was shown that the implementations are at least as usable as existing solutions. To eliminate the need for access to expensive computing infrastructure, state-of-the-art cloud computing techniques are used.
The implementations enable faster identi˝cation of new molecules for use in docking, direct querying of existing databases, and simpler learning of good molecular docking practice without the need to manually run multiple tools. Thus, the framework and methodol-ogy enable more user-friendly implementations, and less error-prone use of computational methods in drug discovery. Their use could lead to more e˙ective discovery of new drugs.