Computational Chemistry

The drug discovery program generally starts with target selection, followed by hit identification, hit-to-lead transition, lead optimisation and clinical candidate selection. Computational approaches including virtual screening (VS), not only has economic advantages, but also a way to access millions of commercially available drug-like small molecules that are not in corporate compound inventories.

The identification of high quality lead compounds is the crucial first step in the small molecule drug discovery process. Leads can be identified through a variety of methods, including high throughput screening (HTS), Computational chemistry approaches including virtual screening (VS) and literature sources.

computational-chemistry

Structure Based Design


Structure Based Design

  • In silico Target identification and validation
  • Homology modeling
  • Binding pose & interaction fingerprinting analysis
  • Focused library design & enumeration
  • Receptor Based Screening

Ligand Based Design


Ligand Based Design

  • Feature and shape field based pharmacophore models
  • Similarity & diversity analysis
  • Scaffold hopping studies
  • Bioisostere replacement studies
  • Virtual Library Screening
  • QSAR Models

e-ADME


e-ADME

  • Aqueous solubility
  • Microsomal stability
  • Caco-2 & MDCK permeability
  • Blood-brain barrier (BBB)
  • Human intestinal absorption (HIA)
  • Plasma protein binding (PPB)
  • Oral bioavailability

Predictive Toxicology


Predictive Toxicology

  • Hepatotoxicity
  • Carcinogenicity Model
  • Mutagenicity Model
  • Skin Sensitisation

Structure Based Virtual Screening

  • Virtual screening of compound libraries has become an important technology in modern drug discovery strategies. If a suitable structure of the target with/without inhibitor/substrate is available, molecular docking can be used to discriminate between putative binders and non-binders in large databases of small-molecules and to substantially reduce number of compounds to be subjected to experimental screening.
  • Structure-based virtual screening requires well-resolved X-ray crystal structures of the molecular target, which are used to dock large database of compounds to identify compounds that potentially have high affinity for the target.
  • These crystal structures are either internally generated or are retrieved from the Protein Database (PDB), a public repository for all published crystal structures.
structure-based-virtual-screening

Binding Pose Analysis (Normal or Flexible Docking)

  • Docking of small molecules into the binding site of a receptor and estimating the binding affinity and analysing the binding poses of the complex is an important part of the structure-based drug design process.
  • To understand the structural principles that determine the strength of a protein/ligand complex either in normal docking or in flexible docking (Induced fit) and the ability to visualise binding geometries and interactions.
13_Binding Pose Analysis (Normal or Flexible Docking)

Focussed Library Design

  • Small-molecule drug discovery project ultimately lies in the choice of the scaffolds to be screened – chosen from millions of available compounds.
  • Currently, there is a trend towards the construction of receptor structure-based focused libraries. Recent advances in high-throughput computational docking, NMR and crystallography have facilitated the development of these libraries.
  • A structure-based target-specific library can save time and money by reducing the number of compounds to be experimentally tested.
  • It improves the drug discovery success rate by identifying more-potent and specific binder focused libraries that not only reduce waste by eliminating priori compounds that are unlikely to bind to the target (thus saving time and money), but could also eventually lead to an increase in the potency or specificity of binders.
  • GVK BIO has internally curated medchem database of 6.5 million compounds and also vendor library of 12.5 million compounds will be helpful in generating focused target libraries.
focussed-library-design

Pharmacophore Modelling

  • Substrate / inhibitor-based pharmacophore can be used for hit identification in absence / presence of protein-substrate / inhibitor co-crystal structure or when structure of a protein cannot be modelled.
  • Both qualitative and quantitative feature-based pharmacophores will be modelled including shape, electrostatic features and functional nature groups of bound conformation of a substrate or known inhibitor.
  • The validated pharmacophore models used for virtual screening of large database and top hits are selected for further docking studies. Compounds that obey lead / drug like properties will be suggested for synthesis.
14_Pharmacophore Modelling

Scaffold Hopping & Bioisostere Replacement

  • The bound conformation of high active compound (HIT) in the active site of protein is considered to identify suitable scaffold or bioisostere replacement.
  • Selected hit molecule scaffolds will be replaced with suitable scaffolds to enhance the binding of compound with protein.
  • Scaffold replacements will be performed to optimise steric and electrostatic complementarities between compound and target protein.
  • Each scaffold or fragment will be searched for suitable replacements against a bioisosteres / fragment databases using Muse, Brood and Spark database / software.
  • The selected novel compounds will be docked into protein active site and amino acid interactions will be analysed.
scaffold-hopping-bioisostere-replacement

Virtual Library Screening

  • Virtual screening is an important tool to design novel drug like compounds and in the process of Hit or lead optimisation, where crystal structure of protein target is not available.
  • There are a wide range of comparable and contrasting methodological protocols available in screening databases for the lead compounds.
  • The number of methods and software packages are available for ligand based virtual screening, However, the general understanding on the applicability and limitations of these methodologies.
  • To overcome the limitations, GVK BIO has developed a comprehensive approach for virtual screening by combining both shape & electrostatic (ROCS & EON) and Feature based (Catalyst) pharmacophore models.
  • The advantage of GVK BIO in terms of in house database availability of 6.5 million medchem compounds with biological activity and 12.5 million vendor databases.
virtual-library-screening

QSAR Modelling

  • A series of qualitative and quantitative pharmacophore models will be generated using a small set of known inhibitors (training set).
  • The derived QSAR models will be validated using a set of parameters including cost analysis, test set prediction, enrichment factor, and goodness of hit and also with test set compounds.
  • Simultaneously docking calculations of known inhibitors will be performed using different programs (Glide and Ligand Fit etc) to calculate their scores against a target along with Physicochemical properties of compounds will be calculated and considered in model generation.
  • Known inhibitors with experimentally determined inhibitory activities (IC50 values), structural diversity, and similarity in the target assay conditions will be collected from journals and patents and divided this data set into training (model generation) and test set (model validation).
15_QSAR Modelling
  • The goal of In silico ADME is to predict disposition behaviour of the compounds in the whole body by assembling all kinetic processes and to identify chemotypes and lead compounds that have good pharmacokinetic properties.
  • It is time consuming step in the lead optimisation stage of drug discovery.
  • Large database compounds with experimental pharmacokinetic data is available in-house at GVK BIO.
  • The available models at GVK BIO are:
    • LogP
    • pKa
    • Solubility
    • Microsomal stability
    • Caco2 Permeability
    • Plasma Protein Binding

ADME Parameter

# Clinical Compounds

# of Drugs

Absorption 396 1300
AUC 9193 16657
Bioavailability 3989 4504
Clearance 4851 9640
Cmax 11504 28650
Cmin 19 44
Vd 10960 12158
Elimination 587 1020
Excretion 4778 15196
T1 / 2 11268 17655
Metabolism 1007 1674
Plasma Binding 1052 4927
Tmax 5439 11101
LD50 4088 35387

  • Quantitative Structure Toxicity Relationship (QSTR) models will be developed using Toxico properties of compound like Carcinogenicity, Mutagenicity, LD50 etc.
  • Bayesian analysis technique and Regression analysis technique will be applied to generate and validate the models.
  • Several softwares are available at GVK BIO to predict the toxicity includes GOSTAR, Topkat, Toxtree and Vega.
  • Compounds with experimental Toxicity data available at GVK BIO.
  • The following toxicity end points can be predicted at GVK BIO for API or Intermediates or Discovery compounds:
    • Number of drug or drug like compounds with toxicity information – 24,554

Prediction for Compound 1 (S-62)

predictive-toxicology

© 2016 | GVK Biosciences Private Limited, All Rights Reserved | Terms & Conditions

Log in with your credentials

Forgot your details?