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.

Shared Consensus Machine Learning Models for Predicting Blood Stage Malaria Inhibition.

The development of new antimalarial therapies is essential, and lowering the barrier of entry for the screening and discovery of new lead compound classes can spur drug development at organizations that may not have large compound screening libraries or resources to conduct high-throughput screens. Machine learning models have been long established to be more robust and have a larger domain of applicability with larger training sets. Screens over multiple data sets to find compounds with potential malaria blood stage inhibitory activity have been used to generate multiple Bayesian models. Here we describe a method by which Bayesian quantitative structure-activity relationship models, which contain information on thousands to millions of proprietary compounds, can be shared between collaborators at both for-profit and not-for-profit institutions. This model-sharing paradigm allows for the development of consensus models that have increased predictive power over any single model and yet does not reveal the identity of any compounds in the training sets.

Fragment library design: using cheminformatics and expert chemists to fill gaps in existing fragment libraries.

Fragment based screening (FBS) has emerged as a mainstream lead discovery strategy in academia, biotechnology start-ups, and large pharma. As a prerequisite of FBS, a structurally diverse library of fragments is desirable in order to identify chemical matter that will interact with the range of diverse target classes that are prosecuted in contemporary screening campaigns. In addition, it is also desirable to offer synthetically amenable starting points to increase the probability of a successful fragment evolution through medicinal chemistry. Herein we describe a method to identify biologically relevant chemical substructures that are missing from an existing fragment library (chemical gaps), and organize these chemical gaps hierarchically so that medicinal chemists can efficiently navigate the prioritized chemical space and subsequently select purchasable fragments for inclusion in an enhanced fragment library.

Four disruptive strategies for removing drug discovery bottlenecks.

Drug discovery is shifting focus from industry to outside partners and, in the process, creating new bottlenecks. Technologies like high throughput screening (HTS) have moved to a larger number of academic and institutional laboratories in the USA, with little coordination or consideration of the outputs and creating a translational gap. Although there have been collaborative public-private partnerships in Europe to share pharmaceutical data, the USA has seemingly lagged behind and this may hold it back. Sharing precompetitive data and models may accelerate discovery across the board, while finding the best collaborators, mining social media and mobile approaches to open drug discovery should be evaluated in our efforts to remove drug discovery bottlenecks. We describe four strategies to rectify the current unsustainable situation.

Chemical space: missing pieces in cheminformatics.

Cheminformatics is at a turning point, the pharmaceutical industry benefits from using the various methods developed over the last twenty years, but in our opinion we need to see greater development of novel approaches that non-experts can use. This will be achieved by more collaborations between software companies, academics and the evolving pharmaceutical industry. We suggest that cheminformatics should also be looking to other industries that use high performance computing technologies for inspiration. We describe the needs and opportunities which may benefit from the development of open cheminformatics technologies, mobile computing, the movement of software to the cloud and precompetitive initiatives.

Using open source computational tools for predicting human metabolic stability and additional absorption, distribution, metabolism, excretion, and toxicity properties.

Ligand-based computational models could be more readily shared between researchers and organizations if they were generated with open source molecular descriptors [e.g., chemistry development kit (CDK)] and modeling algorithms, because this would negate the requirement for proprietary commercial software. We initially evaluated open source descriptors and model building algorithms using a training set of approximately 50,000 molecules and a test set of approximately 25,000 molecules with human liver microsomal metabolic stability data. A C5.0 decision tree model demonstrated that CDK descriptors together with a set of Smiles Arbitrary Target Specification (SMARTS) keys had good statistics [κ = 0.43, sensitivity = 0.57, specificity = 0.91, and positive predicted value (PPV) = 0.64], equivalent to those of models built with commercial Molecular Operating Environment 2D (MOE2D) and the same set of SMARTS keys (κ = 0.43, sensitivity = 0.58, specificity = 0.91, and PPV = 0.63). Extending the dataset to ∼193,000 molecules and generating a continuous model using Cubist with a combination of CDK and SMARTS keys or MOE2D and SMARTS keys confirmed this observation. When the continuous predictions and actual values were binned to get a categorical score we observed a similar κ statistic (0.42). The same combination of descriptor set and modeling method was applied to passive permeability and P-glycoprotein efflux data with similar model testing statistics. In summary, open source tools demonstrated predictive results comparable to those of commercial software with attendant cost savings. We discuss the advantages and disadvantages of open source descriptors and the opportunity for their use as a tool for organizations to share data precompetitively, avoiding repetition and assisting drug discovery.

Strategies to support drug discovery through integration of systems and data.

Much progress has been made over the past several years to provide technologies for the integration of drug discovery software applications and the underlying data bits. Integration at the application layer has focused primarily on developing and delivering applications that support specific workflows within the drug discovery arena. A fine balance between creating behemoth applications and providing business value must be maintained. Heterogeneous data sources have typically been integrated at the data level in an effort to provide a more holistic view of the data packages supporting key decision points. This review will highlight past attempts, current status, and potential future directions for systems and data integration strategies in support of drug discovery efforts.

Prediction of drug-like molecular properties: modeling cytochrome p450 interactions.

Preventing drug-drug interactions and reducing drug-related mortalities dictate cleaner and costlier medicines. The cost to bring a new drug to market has increased dramatically over the last 10 years, with post-discovery activities (preclinical and clinical) costs representing the majority of the spend. With the ever-increasing scrutiny that new drug candidates undergo in the post-discovery assessment phases, there is increasing pressure on discovery to deliver higher-quality drug candidates. Given that compound attrition in the early clinical stages can often be attributed to metabolic liabilities, it has been of great interest lately to implement predictive measures of metabolic stability/ liability in the drug design stage of discovery. The solution to this issue is wrapped in understanding the basic of the cytochrome P450 (CYP) enzymes functions and structures. Recently, experimental information on the structure of a variety of cytochrome P450 enzymes, major contributors to phase I metabolism, has become readily available. This, coupled with the availability of experimental information on substrate specificities, has lead to the development of numerous computational models (macromolecular, pharmacophore, and structure-activity) for the rationalization and prediction of CYP liabilities. A comprehensive review of these models is presented in this chapter.

A comparative QSAR study using CoMFA, HQSAR, and FRED/SKEYS paradigms for estrogen receptor binding affinities of structurally diverse compounds.

The three-dimensional quantitative structure-activity relationship (QSAR) technique of comparative molecular field analysis (CoMFA) has demonstrated the ability to provide accurate predictions for diverse chemical compounds when trained with molecules of diverse chemical type. Although predictive, the derivation and utilization of models of this type are quite computationally and person power intensive. It is this intensity that pragmatically limits the widespread implementation of these models as predictive tools. In this study, two newer QSAR techniques were evaluated as possible alternatives to CoMFA based QSAR models for the purpose of rapidly identifying estrogen receptor ligands from diverse collections of molecules. The first of these is Hologram QSAR, or HQSAR. HQSAR utilizes Tripos molecular fingerprints as descriptors in conjunction with partial least squares (PLS) regression and cross-validation routines. The HQSAR technique demonstrated the ability to rapidly develop QSAR models independent of the intense user input (i.e. geometry optimization, conformational analysis, and molecular superposition were not required). Second, a newly developed QSAR paradigm that utilizes Molecular Design Limited (MDL) substructure keys (SKEYS) as descriptors in combination with an evolutionary algorithm, Fast Random Elimination of Descriptors (FRED), was evaluated. By utilizing the FRED/SKEYS algorithm, a simple substructure-based QSAR model was derived that was comparable in statistical robustness and predictive ability to both CoMFA and HQSAR derived models. A comparison of the utility of these three approaches as computational tools for the rapid identification of estrogen receptor ligands as potential endocrine disruptors as assessed by model predictive ability will be described.