Real Intelligence (RI) is needed to feed AI (Artificial Intelligence) approaches. Experience informed by experimental and quantum chemical (QM) data reveals critical components of ligand - protein binding interactions that are often overlooked. The most important of these is the role of water molecules in binding / potency, selectivity and kinetics. Recent experimental structures and computational studies reveal that water mediated interactions can be responsible for all polar interactions of a ligand with a protein binding site and the perturbation of the remaining waters in a binding site can be as important as the ones displaces. Most approaches to compare ligands or to dock ligands to produce models in QSAR/ML (Machine Learning) and Pharmacophore / Structure-based modeling ignore this critical dimension of ligand binding, and the resultant models can be wrong and misleading for design. Multiple GPCR ligand - protein structures will be presented to reveal this, that overlap in ways not predictable from considering just the ligand properties. The key interaction that is seen in all ligand – protein interactions is lipophilic hotspots, as calculated by GRID/FLAP(WaterFLAP); many of these hotspots are missed by other software programs that often identify just more aromatic hydrophobic regions,  There is usually  an “unhappy” water molecule in these regions, where a methyl group etc can give a significant binding energy increase.   This will be illustrated with multiple X-ray structures, with the recently disclosed Orexin receptor structures for diverse ligands showing that the common binding feature of all the diverse ligands is the lipophilic hotspots.<br/>  Protein flexibility and water networks are not generally included in docking / virtual screening studies, yet are key. Free energy perturbation (FEP) methods for the prediction of binding affinity are another example of RI that can now be used, and examples of successful application to multiple GPCR targets will be discussed. These targets are particularly difficult due the need to include a model of the membrane and often deep buried waters that change as the ligand changes (structurally or binding mode). Other approaches where RI applied to ML can give greatly improved results will also be discussed (eg for the calculation of logP).<br/>

AI needs RI: 40 years of pharma drug discovery reveals key design factors

Macromolecular Chemistry: The Second Century

[Author Benefits](https://scimeetings.acs.org/?utm_source=pubs_content_marketing&utm_medium=website&utm_campaign=0320_MCF_NPI_Launch_Spring_Homepage&ref=pubs_content_marketing)

[How to use SciMeetings](https://storage.googleapis.com/dl.morressier.com/acs/SciMeetings-ACSSpring2020-HowToGuide_M3020-Final.pdf)

[Frequently Asked Questions](https://storage.googleapis.com/dl.morressier.com/acs/SciMtng_ACSNMSp2020_FAQ_M3120_Final.pdf)

ACS Spring 2020 National Meeting & Expo

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.