Current Computer - Aided Drug Design - Volume 4, Issue 1, 2008

The cost of bringing a new drug to market has been estimated to be close to a billion US dollars. With the recent failure of promising candidates late in the clinical trial stage and the need to add caution to the use of certain approved drugs, it has become necessary to review and critique the techniques currently used in drug discovery, including the area of computational molecular design. The theme of this guest issue is “Evolving Paradigms in Drug Design and Discovery”. Beyond the traditional approaches to drug design, newer techniques such as novel decision-making algorithms that help identify new structural candidates and identify chemical diversity metrics having biological information encoded are fast becoming an integral part of the mainstream drug discovery programs. The need for efficient data mining methods, not only at the early research stages, but also during clinical trials is paramount as the decision making processes in bringing a successful drug into market are dynamic and encompass the results of the various constituent experiments. This can prove difficult within an enterprise environment. The incorporation and use of customized automated workflows is a tool that can address such issues successfully. There has been a tendency to assume or take for granted the in-silico workhorses of Computer-Aided Drug Design - the computers that evaluate the formulas and provide the numerical results of complicated algorithms to be the successful approach. High Performance Computing has evolved over the past years in terms of processor speeds, networking and clustering configurations and the efficiency of the operating systems. It is important to review the impact of these advances in the area of molecular design. An understanding of efflux transporter mechanisms is fast becoming an area of active interest in drug design and discovery. A review on the computational modeling of the P-glycoprotein (Pgp) transporter using pharmocophoric and quantitative structureactivity relationship (QSAR) techniques within the context of optimizing the central nervous system penetration has been included. The evolving trend of introducing computational pharmacokinetic and pharmacodynamic techniques early in the drug discovery process necessitates that the available methodologies are reviewed. Commercially available software packages and applications in the area of drug discovery have been discussed in this issue. The application of techniques in the area of oncology-based drug design and discovery such as positron emission tomography (PET) and single photon emission computed tomography (SPECT) imaging studies in the area of tumor biology has been reviewed. Such techniques, when incorporated into a drug discovery paradigm, can reduce the time taken to discover potential liabilities in the metabolism pathways of drug candidates. In summary, it is hoped that this issue will illustrate the many aspects of various multi-disciplinary inputs that are increasingly becoming mainstream technologies in bringing a successful drug into the commercial arena. It is with this focus that this guest issue of Current Computer-Aided Drug Design reviews the areas of change in computer hardware, workflow logistics, novel methods and algorithms in drug design, together with computational pharmacokinetics and the contributions of imaging techniques in the evolving drug design, discovery and development processes.

Computer-aided drug design (CADD) has become a mainstream component of the drug discovery and development process. High Performance Computing (HPC) provides the power that allows CADD researchers to explore more designs in less time, and some of the greatest improvements in CADD result directly from advances in HPC. This paper examines some of the more significant trends in HPC that influence computer-aided drug design (CADD).

Workflow solutions driven by data pipelining are increasingly becoming popular for accessing, aggregating and analyzing disparate data to make informed and intelligent decisions. Uses of workflow technologies which facilitate business intelligence (BI) improve productivity, decision making and research efficiency. In order to provide BI in a scientific or clinical based organization, it is imperative that the application or workflow technology must be compatible with multiple data types and formats, be able to analyze the data and make it available throughout the organization. We term this as Scientific Business Intelligence (SBI) and discuss how modeling, simulations and informatics software, integrated with open and standards-based scientific operating platform (SOP), can deliver scientifically-relevant BI solutions. We illustrate SBI with several examples encompassing all levels of users within an organization.

Despite great advances in the efficiency of analytical and synthetic chemistry, time and available starting material still limit the number of unique compounds that can be practically synthesized and evaluated as prospective therapeutics. Chemical diversity analysis (the capacity to identify finite diverse subsets that reliably represent greater manifolds of drug-like chemicals) thus remains an important resource in drug discovery. Despite an unproven track record, chemical diversity has also been used to posit, from preliminary screen hits, new compounds with similar or better activity. Identifying diversity metrics that demonstrably encode bioactivity trends is thus of substantial potential value for intelligent assembly of targeted screens. This paper reports novel algorithms designed to simultaneously reflect chemical similarity or diversity trends and apparent bioactivity in compound collections. An extensive set of descriptors are evaluated within large NCI screening data sets according to bioactivity differentiation capacities, quantified as the ability to co-localize known active species into bioactive-rich K-means clusters. One method tested for descriptor selection orders features according to relative variance across a set of training compounds, and samples increasingly finer subset meshes for descriptors whose exclusion from the model induces drastic drops in relative bioactive colocalization. This yields metrics with reasonable bioactive enrichment (greater than 50% of all bioactive compounds collected into clusters or cells with significantly enriched active/inactive rates) for each of the four data sets examined herein. A second method replaces variance by an active/inactive divergence score, achieving comparable enrichment via a much more efficient search process. Combinations of the above metrics are tested in 2D rectilinear diversity models, achieving similarly successful colocalization statistics, with metrics derived from the active/inactive divergence score typically outperforming those selected from the variance criterion and computed from the DiverseSolutions software.

This paper focuses on the recent development of rule-based methods and their applications to the drug discovery process. For a given target, the path for designing new drugs with a lower attrition rate is based on an effective mining of the huge amount of experimental in vitro and in vivo data which has been collected. These data often come in various formats, from many different areas such as chemistry, biology, pharmacology, toxicity and extraction of the critical information is not an easy task. To guide the multi-objective optimization, we have developed a decision-support system (KEM®), based on the Galois lattices theory and constraint satisfaction programming (CSP). After a brief overview of machine learning applications, we will describe the methodology used in KEM for data mining and prediction. Two examples of applications in the drug discovery area will be discussed.

Spiraling drug developmental costs and lengthy time-to-market introduction are two critical challenges facing the pharmaceutical industry. The clinical trials success rate for oncology drugs is reported to be 5% as compared to other therapeutic categories (11%) with most failures often encountered late in the clinical development process. PET and SPECT nuclear imaging technologies could play an important role in facilitating the drug development process improving the speed, efficiency and cost of drug development. This review will focus on recent studies of PET and SPECT radioligands in oncology and their application in the investigation of tumor biology. The use of clinically-validated radioligands as imaging-based biomarkers in oncology could significantly impact new cancer therapeutic development.

Computer-aided pharmacokinetic, pharmacodynamic, and pharmacokinetic/pharmacodynamic methods are commonly applied to quantify the disposition and the pharmacological effects of the drug, to explore exposure-response relationships, and to predict safety and efficacy outcomes. Use of modeling and simulation throughout the drug development continuum can support more efficient preclinical and clinical study design and interpretation. Mechanism-based approaches where sound biological understanding exists provide meaningful quantitative comparisons between candidates and are sought to support science-based decisions. Simulations from these models allow for scientists to investigate a variety of trial designs where assumptions are clearly stated. The objectives of this review article are to describe commercially available PK/PD software packages and present examples of their application in drug discovery and development. With industry and regulatory support, use of exposure response information may optimize the path to delivery of new medicines to patients. This review is focused on the most common computer software applications in discovery through early development (i.e., GastroPlus, Simcyp Population-based ADME simulator, SAAM II, and WinNonlin), in development (i.e., NONMEM, ADAPT II, MATLAB, WinBUGS, Trial Simulator, and Drug Model Explorer), and across the continuum for data management (i.e., SAS, S-PLUS, and R).

Development of a drug involves several aspects, one of which is an adequate DMPK profile that is related to its absorption, distribution, metabolism and excretion. The distribution of the drug to its site of action is partly regulated by several biological membrane barriers. One such barrier is created by the brain capillaries of the endothelial cells, also known as the Blood-Brain-Barrier (BBB). Depending on the therapeutic action, one may need higher permeation of the drug through BBB if the site of action is in the CNS, or minimize the entry through the BBB if this biological target is located in the periphery. The physicochemical properties of the drug usually regulate its permeability through the BBB and constitute passive permeability. However, “non-passive permeation” may also exist and is affected by other transporter mechanisms present in the BBB, and may involve both efflux as well as influx systems. Amongst these, the PGlycoprotein (Pgp) has been the most extensively characterized efflux transporter. The “passive BBB” has been well studied and characterized by various theoretical groups, but the “non-passive BBB” (often caused by Pgp, for example) has gained more attention from computational methodologies in recent years. This review will provide a brief summary of the computational strategies that have addressed Pgp efflux inhibition, especially in the context of optimizing CNS penetration during rational drug design. The advances in the computational methods that have modeled the Pgp recognition while addressing non-passive permeation will be a chief focus, but coverage is also given to recent and impactful Pgp modeling approaches. These include computational approaches that analyze data from assays targeting Pgp in particular or multidrug resistance reversal assays where Pgp is a chief implicating factor.