Phenotypic screening is used to identify substances that alter the phenotype of a cell or an organism. It has been a historical basis for the discovery of new drugs and in biological research and only after compounds were tested positive, biological targets are determined. While most drugs with novel mechanisms of action come from phenotypic screening, it is also of interest in the area of drug repositioning.

Assembling the best libraries and selecting the best computer-aided drug design (CADD) techniques are of paramount importance to maximize the probabilities to select the best candidate following a phenotypic screening strategy.
Ligand based techniques and Artificial Intelligence are the best approach to guide a phenotypic based drug discovery process from a rational point of view.

Henriette Willems, Senior Research Associate Computational Chemist at ALBORADA DDI at the University of Cambridge, shared her strategies for assembling an annotated library for phenotypic screening. Using specific metrics that helped define selective compounds, she was able to analyze public database to generate output data useful to rank the most active compounds in a meaningful way.

Paul Finn, CEO at Oxford Drug Design & Professorial and Research Fellow at the University of Buckingham, gave his insight on ligand-based tools and strategies for the analysis of phenotypic screening results. Depending on the composition of the screening library, clustering can help identify interesting chemotypes. Clustering and pharmacophore approaches can help SAR analysis whereas 3D ligand-based methods assess similarity of molecular shape, electrostatics and other properties. Therefore, traditional ligand-based computational approaches can be useful in analyzing phenotypic screening results.