Optimizing Electrostatic Similarity for Virtual Screening: A New Methodology

Abstract

<div>Ligand Based Virtual Screening (LBVS) methods are widely used in drug discovery as filters for subsequent in-vitro and in-vivo characterization. This means, increasing accuracy of LBVS approaches may have a huge impact on increasing chances of success. Since the databases processed in drug discovery campaigns are enormously large, this pre-selection process requires the use of fast and precise methodologies. The similarity between compounds can be measured using different descriptors such as shape, pharmacophore or electrostatic similarity. The latter is the goal of this work, i.e., we want to improve the process of obtaining the compounds most similar to a query in terms of electrostatic similarity. To do so, the current and widely proposed methodology in the literature is based on the use of ROCS to assess the similarity of compounds in terms of shape and then evaluate a small subset of them with ZAP for prioritization regarding electrostatic similarity. This paper proposes an alternative methodology that consists of directly optimizing electrostatic similarity and works with the entire database of compounds without using shape cut-offs. For this purpose, a new and improved version of the OptiPharm software has been developed. OptiPharm implements a parameterizable metaheuristic algorithm able to solve any optimization problems directly related to the involved molecular conformations. We show that our new method completely outperforms the classical proposal widely used in the literature. Accordingly, we are able to conclude that many of the compounds proposed with our novel approach could not be discovered with the classical one. As a result, this methodology opens up new horizons in Drug Discovery.</div>

A Comparative Study of Stochastic Optimizers for Fitting Neuron Models. Application to the Cerebellar Granule Cell

“…This approach is aligned with the proposal in Marić et al (2014) since the local search is applied to a random subset of the population rather than to all of them, which increases diversity. As introduced, the local optimizer used is SASS, by Solis and Wets (1981), which is also the local method included in the memetic algorithms applied in Cruz et al (2018), Puertas-Martín et al (2020. Thus, every individual is a potential starting point for an independent execution of SASS.…”