Home
  Instrumentation / Laboratory
  High Throughput Screening
  Microarray Screening
  Protease Proteomics
  Zebra Fish Screening
  Synthetic Chemistry
  Metallo-Organic Pharmacophores
  Cheminformatics and Computational Chemistry
  Chemical Genomics/Organism HTS
spacer spacer
Director
spacer
NIH Molecular Libraries and Imaging
spacer spacer
NIH Small Molecule Repository
spacer
Publications
spacer spacer
Pubchem
spacer spacer
News Archive
spacer spacer
Top 200 Brand Name Drugs
(compiled by the Njarðarson Group at the University of Arizona)
spacer spacer
 
spacer spacer
spacer

Email pcmd@seas.upenn.edu

spacer

 

spacer spacer

Cheminformatics and Computational Chemistry

Computational Strategies for Optimizing Biologically Active Leads in the PCMD

Lead chemical series identified through automated screening of the MLSMR (Molecular Libraries Small Molecule Repository, NIH) can be further optimized using a variety of computational strategies. These include: 1.) substructure and similarity searching, 2.) virtual library enumeration around the active lead series followed by protein/ligand docking and scoring, 3.) quantitative structure activity relationship (QSAR) model derivation, and 4.) chemical reactivity estimations. These methodologies are currently used within the PCMD by a multi-disciplinary team of scientists in Medicinal Chemistry, Biochemistry/Bioengineering, and Computational Chemistry to understand the activities of chemical probes at the molecular level for various therapeutic targets.

Software Tools

Protein/ligand docking: several packages available from Schrodinger, Inc., including Glide for binding site docking (standard precision and extra precision modules), LigPrep for Glide small molecule database creation, and MMPBSA and MMGBSA to augment the SP and XP scoring schemes.

Chemical reactivity estimates: PC Spartan (Wavefunction, Inc.), used for semi-empirical geometry optimizations prior to docking and pharmacophore searching, and also for calculations of electrostatic potentials for estimating electrophilicity at specific molecule centers (for ex., as applied to covalently bound protease inhibitors).

QSAR model derivation: MOE software, available from Chemical Computing Group, CCG, used for descriptor calculation and model building. Linear models utilize PCR and PLS regression techniques. ADMETPredictor and ADMETModeler (available from SimulationsPlus, Inc.) used for descriptor calculations and non-linear regression model building.

3D Pharmacophore Searching: Unity software, Tripos, Inc., and Phase software, Schrodinger, Inc. Databases available: LeadQuest, Zinc, and MLSMR (NIH). Flexible 3D pharmacophore searching can be carried out on the ‘ligand only’, or, if protein structural information is available, on the complementary ligand/protein interaction sites. When creating 3D pharmacophores, generalized chemical features such as H-bond donor and acceptor sites, as well as aromatic and hydrophobic functionalities, can be assigned to atomic positions in the lead molecular series. Explicit chemical functions can also be retained as pharmacophore features. Partial matching algorithms allow for various combinations of pharmacophore features to be selected from an initial large pool of many pharmacophore features. This is especially useful when there is limited SAR information available for the lead series. 


Go to UPenn's Homepage