Molecular Property Diagnostic Suite (MPDS):

A Component of Open Source Chemoinformatics Initiative

 

Council of Scientific and Industrial Research (CSIR), Govt. of India, has undertaken the Open Source initiative to establish a novel open source platform for both computational and experimental technologies to carry out Drug Discovery for infectious/ neglected diseases. An important challenge of this is to ensure that the drug molecule remains affordable to the people of the developing world. The Molecular Property Diagnostic Suite (MPDS) is a component of Opensource chemoinformatics initiative; conceptualized to assess and estimate the multifarious aspects of drug-likeliness of any given molecule, in order to diagnose their potential application as drug. MPDS compound library is updated till 31st December, 2016.

 

Category	Modules	Description
Data Library	Module 1: Literature	Contains Mtb proteins and its genetic information; FDA approved drug information and polypharmacological information.
	Module 2: Target Library	Contains crystal structures and homology models for Mtb proteins.
	Module 3: Compound Library	Contains a single window interface for searching a compound in MPDS compound database.
Data Processing	Module 4: File Format Conversion	Conversions of files from one chemical format to another chemical format, 2D to 3D file conversion using Open Babel.
	Module 5: Descriptor Calculation	Calculation of descriptors and fingerprints using PaDEL and CDK tools.
Data Analysis	Module 6: QSAR	Generation of QSAR models using the data mining tools, McQSAR and SVMlight.
	Module 7: Docking	Ligand Optimization; Conformer Generation and Protein-Ligand docking.
	Module 8: Screening	Prioritization of compounds for drug-like features using DruLiTo tool; Biopharmaceutical Classification System (BCS); Identification of toxicophoric groups in a compound.
	Module 9: Visualization	Visualizing protein-ligand interactions using Jmol and Ligplot.

 

Work Plan

• Database Creation: Storage and linking of molecules and their fragments to in-house databases, in convertible file formats. Upgradation of existing databases (like A2IDB, PLID and CAD of CMM-IICT group). Development of an interface using GALAXY platform to link the databases and softwares onto the common OSDD chemoinformatics portal.
(Groups Involved: IICT, IMTECH, JNU)
• Descriptor calculations: Development of new in-house descriptors for drug metabolism, toxicity, druglikeliness, reactivity etc. Writing codes to interface the available public domain packages with the in-house descriptors. Calculation of descriptors for the in-house database of molecules and fragments. Development of empirical models and QSA(P/T/E)R relationships for metabolism, toxicity and solvation.
(Groups Involved: IICT, JALEEL, CLRI, NCL)
• Chemoinformatics: Generation of fingerprints to represent the structure, interaction and property of a molecule. Use of pattern recognition, machine learning tools and cluster analysis for similarity analysis of a molecule/fragment with those in public domain databases like PubChem (> 37 million entries), ZINC (> 13 million compounds) and FDA approved drugs (4000).
(Groups Involved: IICT, IMTECH NCL)
• Filters:Development of target specific and general filters based on molecular properties to estimate the druggability of the candidate molecule. Use of these filters to design a quick and precise virtual screening protocol.
(Groups Involved: IICT,NCL,BBAU,NIPER)
• Predictive Computing: Employment of heavy duty computing techniques such as QM, QM/MM and MD simulations to address the problems of fundamental interest in Chemistry and Biology.
(Groups Involved: IICT,NIPER,BBAU,CLRI)

Deliverables

• Galaxy platform: Developing an interface through ′GALAXY′ to link the databases, software and tools onto the common ′OSDD chemoinformatics portal′.
• Descriptor calculations: Tools to calculate a large variety of molecular descriptors such as constitutional, topological, thermodynamic, electrostatic, shape-based, QM-based and DFT-based etc. will be developed. Apart from these conventional ones, descriptors to quantify the protein-ligand interactions such as free energy of binding, calculated with MMPBSA/GBSA methods, will also be calculated. Tools for exploring the probable modifications in a molecule (e.g., adding or removing a certain fragment at a certain site) to give rise to better activity, specificity and reduced toxicity will be developed. Development of QSAR and QSTR models based on the above calculated descriptors.
• Structure representation and storage: Representation of molecular structures in SMILES, 2D and 3D formats alongside the representation of chemical fingerprints present in the structures. Subsequent storage of them in a database along with their known experimental properties, biological activities and spectral properties.
• Similarity/Diversity based approaches: Development of codes and tools to assess the similarity and diversity among molecules. Subsequent categorization of them in quantitative terms into appropriate clusters, according to its property. Here, one of the principle objectives is data reduction without losing diversity. We plan to employ molecular fingerprints which will allow one to quantify similarity and diversity. Based on this, large scale analysis of data is possible.
• Fragment based approaches: Generation of fragment libraries specific for the active sites of various potential drug targets such as kinases, GPCRs etc. Development of tools for exploring the probable modifications in molecules those are already there in clinical trials. For example, adding or removing a certain fragment at a certain site in the molecule, to ensure better activity, specificity and reduced toxicity.
• Target specific filters: Identify target specific filters for screening large libraries of small molecules, drug-like molecules and natural products for identifying potential lead compounds for certain target proteins using docking, MD simulation and QM/MM based parameters.
• Drug metabolism and toxicity: Bond dissociation energy can be used as a descriptor to quantify the metabolism of a certain molecules by the Cytochrome P450s. Cytochrome P450s are the enzymes which take major role for metabolism of drug molecules by various chemical reactions such as epoxidation, sulphoxidation etc. All the breakable bonds and reaction centers in molecules can be identified and calculation of the corresponding bond dissociation energies will give an idea of the possible mechanisms of metabolism of the molecule and also the possible intermediated formed in each mechanism. This knowledge may give an insight to the toxicity of a particular molecule.
• Predictive computing: Application of QM/MM and Free Energy Perturbation (FEP) methods to predict the biological reactions mediated by various enzymes and metallo-enzymes. Prediction of structures of various drug targets using MD simulations, Monte Carlo sampling etc.
• In silico screening: Screening of all molecules made by OSDD, CSIR or any other team targeted towards M. tuberculosis.