William J. Welsh, Ph.D.
Norman H. Edelman Professor in Bioinformatics Department of Pharmacology
Robert Wood Johnson Medical School

Director, The UMDNJ Informatics Institute

Phone: 732-235-3234
Fax: 732-235-3475
Email: welshwj@umdnj.edu

Our laboratory specializes in the development and application of computational tools in bioinformatics, cheminformatics, and computer-aided drug discovery. Examples of active research projects are described below.

· Novel Opioid Receptor Active Agents

Our laboratory has discovered (worldwide patent pending) a new class of small-molecule compounds, structurally distinct from morphine-like opioids, that exhibit high (nanomolar) binding affinity and selectivity for the opioid receptors. These compounds can be chemically synthesized in high yield by a facile reaction that is amenable to large-scale parallel synthesis. Potential therapeutic applications include analgesics, immunomodulatory agents, and treatments for drug addiction.

· Tubulin Ligands as Anti-Cancer Therapeutic Agents

One of our new initiatives is the rational design of small-molecule anti-tubulin/anti-microtubule compounds as potential anti-cancer therapeutic agents. These compounds are structural analogues of the opioid receptor active agents mentioned above, thus synthesis and biological evaluation is underway.

· Na, K-ATPase Inhibitors for the Therapeutic Treatment of Cardiovascular Diseases

Our laboratory has developed (patent pending) a structural model for human Sodium Potassium (Na, K-) ATPase, the target for cardioglycosides such as digoxin and digitoxin which are used for the treatment of congestive heart failure and related conditions. Unfortunately, these two cardioglycosides have narrow therapeutic indices resulting in severe toxic side effects. Based on our structural model, together with interpretation of biological data, we have elucidated the putative mechanism of action of these cardioglycosides. Furthermore, we have constructed molecular model to guide the rational design of a new generation of effective yet safer therapeutic agents.

· Nuclear Hormone Receptors

The nuclear receptors, such as the estrogen receptor (ER) and androgen receptor (AR) are hormone-dependent transcription factors that control many reproductive, developmental, and metabolic functions in humans and wildlife. Our laboratory has developed computer-based molecular models to guide the discovery of novel therapeutic agents for the treatment of pathologies that are mediated through nuclear receptors such as breast cancer and prostate cancer.

· Kinase Inhibitors for the Treatment of Malaria

Our laboratory has constructed structural models for two novel kinases (PfPK5 and Pfmrk) that play an essential role in the life cycle of Plasmodium (P.) falciparum, the major causative parasitic agent responsible for over 95% of the 1-3 million worldwide deaths resulting annually from malarial infection. Using this structural information for PfPK5 and Pfmrk, we have developed a strategy for the rational design of potent and selective P. falciparum kinase inhibitors as therapeutic agents for the treatment of malaria.

· Shape Signatures Tool

Together with Dr. Randy Zauhar (USIP, Phila, PA), we have developed (patent pending) a novel computational tool known as “Shape Signatures” that is useful for many applications in drug discovery such as in silico screening that rapidly matches small drug–like molecules against each other or against receptor pockets based on similarity in shape and electrostatic properties.

· Computational Tool for Predicting Propensity for Amyloid Fibril Formation

Amyloid fibril formation is associated with many lethal diseases including Alzheimer’s disease, Parkinson’s disease, and Type II diabetes. Our laboratory has developed (patent pending) a computational tool that identifies those sequences within proteins or polypeptides which are especially susceptible to amyloid fibril formation. We call this susceptibility “hidden ?-strand propensity”, since it refers to the strong propensity for specific sequences to transform from their native helical or coil secondary structure to b-strands under certain physiologically relevant conditions. This predictive tool is useful for many applications, such as the discovery of new diagnostic and therapeutic agents useful for the prevention and treatment of amyloid diseases.

· “Docking & Scoring” Scheme for Virtual Drug Screening

We have recently developed (patent pending) a novel scheme for docking & scoring, a computational technique widely used in drug discovery for virtual screening of large small-molecule databases against known protein targets. Our technique, named OSKAR, represents an innovative strategy to achieving a balance between speed and accuracy.

· Computational Tools for Analysis and Interpretation of Microarray Data

DNA microarray technology has led to an explosion of gene expression data. However, virtually every experiment contains missing entries arising from blemishes on the microarray, and values of missing entries must be estimated before clustering can be applied. Our laboratory has recently co-developed (patent pending) a missing value estimation method based on Gaussian mixture modeling. Our estimation method has been shown empirically to be more accurate than existing methods.

· Optimal Design of Biomaterials

In conjunction with Dr. Joachim Kohn (Rutgers Univ. & NJ Center for Biomaterials), we are applying rational computer-guided molecular modeling tools toward the optimal design of polymers for biomedical applications such as implants for tissue regeneration.