Data to Decisions: Creating a Culture of Model-Driven Drug Discovery.

Merck & Co., Inc., Kenilworth, NJ, USA, is undergoing a transformation in the way that it prosecutes R&D programs. Through the adoption of a "model-driven" culture, enhanced R&D productivity is anticipated, both in the form of decreased attrition at each stage of the process and by providing a rational framework for understanding and learning from the data generated along the way. This new approach focuses on the concept of a "Design Cycle" that makes use of all the data possible, internally and externally, to drive decision-making. These data can take the form of bioactivity, 3D structures, genomics, pathway, PK/PD, safety data, etc. Synthesis of high-quality data into models utilizing both well-established and cutting-edge methods has been shown to yield high confidence predictions to prioritize decision-making and efficiently reposition resources within R&D. The goal is to design an adaptive research operating plan that uses both modeled data and experiments, rather than just testing, to drive project decision-making. To support this emerging culture, an ambitious information management (IT) program has been initiated to implement a harmonized platform to facilitate the construction of cross-domain workflows to enable data-driven decision-making and the construction and validation of predictive models. These goals are achieved through depositing model-ready data, agile persona-driven access to data, a unified cross-domain predictive model lifecycle management platform, and support for flexible scientist-developed workflows that simplify data manipulation and consume model services. The end-to-end nature of the platform, in turn, not only supports but also drives the culture change by enabling scientists to apply predictive sciences throughout their work and over the lifetime of a project. This shift in mindset for both scientists and IT was driven by an early impactful demonstration of the potential benefits of the platform, in which expert-level early discovery predictive models were made available from familiar desktop tools, such as ChemDraw. This was built using a workflow-driven service-oriented architecture (SOA) on top of the rigorous registration of all underlying model entities.

Structural basis for the cooperative allosteric activation of the free fatty acid receptor GPR40.

Clinical studies indicate that partial agonists of the G-protein-coupled, free fatty acid receptor 1 GPR40 enhance glucose-dependent insulin secretion and represent a potential mechanism for the treatment of type 2 diabetes mellitus. Full allosteric agonists (AgoPAMs) of GPR40 bind to a site distinct from partial agonists and can provide additional efficacy. We report the 3.2-Å crystal structure of human GPR40 (hGPR40) in complex with both the partial agonist MK-8666 and an AgoPAM, which exposes a novel lipid-facing AgoPAM-binding pocket outside the transmembrane helical bundle. Comparison with an additional 2.2-Å structure of the hGPR40-MK-8666 binary complex reveals an induced-fit conformational coupling between the partial agonist and AgoPAM binding sites, involving rearrangements of the transmembrane helices 4 and 5 (TM4 and TM5) and transition of the intracellular loop 2 (ICL2) into a short helix. These conformational changes likely prime GPR40 to a more active-like state and explain the binding cooperativity between these ligands.

Molecular Simulations Identify Binding Poses and Approximate Affinities of Stapled α-Helical Peptides to MDM2 and MDMX.

Traditionally, computing the binding affinities of proteins to even relatively small and rigid ligands by free-energy methods has been challenging due to large computational costs and significant errors. Here, we apply a new molecular simulation acceleration method called MELD (Modeling by Employing Limited Data) to study the binding of stapled α-helical peptides to the MDM2 and MDMX proteins. We employ free-energy-based molecular dynamics simulations (MELD-MD) to identify binding poses and calculate binding affinities. Even though stapled peptides are larger and more complex than most protein ligands, the MELD-MD simulations can identify relevant binding poses and compute relative binding affinities. MELD-MD appears to be a promising method for computing the binding properties of peptide ligands with proteins.

The evolution of drug design at Merck Research Laboratories.

On October 5, 1981, Fortune magazine published a cover article entitled the "Next Industrial Revolution: Designing Drugs by Computer at Merck". With a 40+ year investment, we have been in the drug design business longer than most. During its history, the Merck drug design group has had several names, but it has always been in the "design" business, with the ultimate goal to provide an actionable hypothesis that could be tested experimentally. Often the result was a small molecule but it could just as easily be a peptide, biologic, predictive model, reaction, process, etc. To this end, the concept of design is now front and center in all aspects of discovery, safety assessment and early clinical development. At present, the Merck design group includes computational chemistry, protein structure determination, and cheminformatics. By bringing these groups together under one umbrella, we were able to align activities and capabilities across multiple research sites and departments. This alignment from 2010 to 2016 resulted in an 80% expansion in the size of the department, reflecting the increase in impact due to a significant emphasis across the organization to "design first" along the entire drug discovery path from lead identification (LID) to first in human (FIH) dosing. One of the major advantages of this alignment has been the ability to access all of the data and create an adaptive approach to the overall LID to FIH pathway for any modality, significantly increasing the quality of candidates and their probability of success. In this perspective, we will discuss how we crafted a new strategy, defined the appropriate phenotype for group members, developed the right skillsets, and identified metrics for success in order to drive continuous improvement. We will not focus on the tactical implementation, only giving specific examples as appropriate.

Data is the currency of R&D, and that currency is now generated and traded globally.

Changes in the understanding of biological science, translational research and corporate business models require a corresponding change in the approach to chemical and biological information management. The concept of operations being partitioned into discrete departments for drug discovery is beginning to be replaced by a translational approach to this process. Pharmaceutical business and organizational models are also constantly evolving. Traditional approaches to transactional systems, transferring data up to a departmental data warehouse, are no longer meeting the needs of pharmaceutical scientists and, thus, IT departments are not considered as relevant to the business. These changes and their impact on information systems, as well as some solutions to the challenges faced, are discussed in this editorial.

Practical approaches to efficient screening: information-rich screening protocol.

The past approach of high-throughput screening of everything in the corporate collection has been shown to be very expensive in terms of reagents cost, disposal cost, and compound collection depletion. It is well known that screening campaigns produce several hits, of which only 50% confirm on average. More efficient ways of screening can provide an informative structure-activity relationship (SAR), which in turn can be used to build mathematical models for further probing the activity space and directing chemical synthesis. The authors report new methods and insights to extract the maximum possible information from a screening experiment and find most of the possible hits in the corporate collection while screening as few compounds as possible.

Gaussian docking functions.

A shape-based Gaussian docking function is constructed which uses Gaussian functions to represent the shapes of individual atoms. A set of 20 trypsin ligand-protein complexes are drawn from the Protein Data Bank (PDB), the ligands are separated from the proteins, and then are docked back into the active sites using numerical optimization of this function. It is found that by employing this docking function, quasi-Newton optimization is capable of moving ligands great distances [on average 7 A root mean square distance (RMSD)] to locate the correctly docked structure. It is also found that a ligand drawn from one PDB file can be docked into a trypsin structure drawn from any of the trypsin PDB files. This implies that this scoring function is not limited to more accurate x-ray structures, as is the case for many of the conventional docking methods, but could be extended to homology models.