Welsh J. Welsh, PhD
Associate Director
Cheminformatics Analysis
Cheminformatics serves as a key component of the Institute’s Cancer Pharmacology and Precision Oncology programs. Cheminformatics refers to the use of computational tools and techniques for a range of problems in the field of chemistry, chemical and structural biology, and molecular discovery including drug discovery. It augments and coordinates with ongoing efforts at Rutgers Cancer Institute in Bioinformatics and Systems Biology.
The primary objectives are:
- To collaborate with Rutgers Cancer Institute investigators in the transformation of genomics and clinical data into tangible therapeutic strategies based on analysis of the target receptor’s structure and function.
- To accelerate drug discovery and informatics-driven translational biomedical research in cancer.
- To raise awareness and visibility of cheminformatics & drug discovery at Rutgers Cancer Institute.
Cheminformatics and Precision Oncology
Coupled with Bioinformatics and Systems Biology, Cheminformatics provides a powerful suite of computational tools for extending precision medicine to the level of protein structure and function and beyond to individual drug molecules. In the context of precision oncology, it offers the prospect to translate the genomic information of an evolving cancer emerging from bioinformatics and systems biology analyses into a three-dimensional (3D) representation of the aberrant protein’s structure down to the atomic level of resolution. Visualization of the protein’s 3D molecular structure enables the scientist to pinpoint the precise location of those mutations that drive tumor progression as well as resistance to standard-of-care drug therapies. Rapid in silico screening of the protein’s 3D structure against large pre-assembled databases of approved drugs and drug candidates may suggest new strategies for targeted therapy. This body of information is relayed back to the attending oncologist as guidance for optimal therapeutic management of the patient.
Cheminformatics and Cancer Pharmacology
Beyond existing therapies, cheminformatics can be extended into the realm of anticancer drug discovery. For example, the 3D molecular structure of mutated proteins that drive tumor progression and metastasis may serve as starting points for the rational (computer-aided) design of novel anticancer therapeutics. In many cases, the X-ray crystallographic structure of the target protein can be retrieved from public repositories such as the Protein Data Bank (http://www.rcsb.org/pdb/). In other cases where the experimental 3D structure is unavailable, computational methods known as homology or comparative modeling can be implemented to construct a hypothetical 3D structure of the target protein based solely on the protein’s primary sequence. Concurrently, in silico (virtual) screening tools can be employed to search extensive libraries of chemical compounds for drug hits and leads. These cheminformatics strategies have many uses, including the determination of the target protein’s structure, function, and mechanism of action as well as the discovery of novel anticancer therapeutics.
Services
Cheminformatics offers support, collaboration and training to Rutgers Cancer Institute investigators and to the entire Rutgers University community.
Structure-based drug design
Starting from the experimental (X-ray or NMR) structures or theoretical structural models of proteins, small drug-like molecules stored in extensive libraries of chemical compounds are computationally docked to the binding site. A cascade of filtering, ranking, scoring processes is carried out to select the top compounds for experimental assays. Iterative enrichment cycles of computational screening and biological characterization lead to top hits that serve as the starting points for further drug development.
Ligand-based drug design
Starting from one or more bioactive compounds, virtual screening of large chemical databases containing millions of compounds is conducted to identify a wider array of structural analogues that are synthesized or purchased for biological evaluation to assess their viability as potential drug candidates. This process of hit-to-lead optimization is accelerated by the development of computational pharmacophore models and quantitative structure-activity relationship (QSAR) models.
Peptide, protein, and antibody modeling
Sequence-to-structure prediction of peptides, proteins, and antibodies is achieved using powerful methods in homologue identification, sequence alignment, and 3D structural refinement. Large-scale atomistic molecular modeling and molecular dynamics simulations are performed to study protein:protein, antigen:antibody, and drug:protein interactions in the context of biologically realistic conditions (temperature, pH, solvent, ions, etc.).
Molecular database mining
Molecular databases contain millions of compound structures and their properties. The process of data mining and pattern recognition analyzes extensive data sets to obtain useful information leading to understanding of hidden relationships within the data sets. Different statistical methods (e.g. regression and clustering analyses) are used to analyze the data and to extract useful information.
Tools and techniques
Software packages used for docking include Autodock, DOCK, Gold and Vina. Software used for molecular dynamics simulations includes Amber, Gromacs and NAMD. Quantum mechanical programs like Gaussian and Spartan are used to study physical properties of small molecule drugs, small peptides and unusual nucleic acids. The Modeller program and Rosetta package are used for homology modeling and protein and peptide modeling. Molecular Operating Environment (MOE), a fully integrated drug discovery software program, is used for ligand-based and structure-based drug design.
For more information or to request services, please contact:
Youyi Peng, PhD
pengyo@cinj.rutgers.edu