Our software platform (SWP) for Scientific Applications is a new platform developed to host applications (APPs) to perform the computation for scientific applications.
We also develop APPs that are hosted on our SWP for providing solutions to the industries working in materials, chemicals, energy, and pharmaceutical domains. These APPs perform different tasks e.g., new chemical (drug) entity screening for potentials target (protein, DNA, RNA), screening of molecules for specific use (additive, paints, adhesive surfactant), predict bulk propertied (of polymers, small molecules, solvents, ionic liquids), design surfactants for oil recovery, etc. The SWP provides all the necessary backend support to these APPs. The SWP consists of three major backend supports, 1. Data-Connector 2. Modules 3. Visualization tools.
The Data-Connector supports the user to upload, download data, manage files, and connect to public cloud such as Google Cloud. It also manages all the local servers, at the premise, clusters, and automatically runs (and load balance) a large number of calculations. Data-Connector is a major component of the SWP as this is developed to aid users to scale up a number of calculations without any manual intervention and without any in-house computational resources.
The modules in the SWP are the backend types of machinery for the APPs. The SWP currently consists of QM, MD, MC, file conversion, analysis modules which can be called (integrated) in any APP. Mostly these modules are open-source well tested and scalable in high-performance computing (HPC) environments and public cloud.
The SWP could host visualization tools for user interactions with the data and analysis of outputs. This layer of SWP currently populated with molecular visualizers and plotters.
Multi Target Multi Ligand Enhanced Sampling Screening (MTMLESS) module has been developed to handle multiple target and ligand systems at the same time. For example, if we have m targets and n ligands, the total number of systems one could screen are (m*n-1). In this case targets (m) could be proteins and nucleic acids and n could be a few thousands of chemical compounds. The number of targets could vary from 1 to n, and n depends on the selection of the targets (which could be inhibited by ligand/s) from the pathways which highly relevant to the disease. These computations are automated in this module and the user only needs to provide the targets and ligands. The automation protocol is highly parallel, so dependent on the computations resource one could scale the total number of ligands and targets. The MTML module uses molecular docking and enhanced sampling MD simulations combined computational approach to screen the ligands for the targeted disease.
Molecule simulator is designed to build and predict material properties for industry relevant molecular systems. A customizable material can be designed by choosing an industrially organic molecule and a relevant solvent. Molecular dynamics simulation is used to predict its physical and structural properties. The simulator is versatile where an organic molecule can be either polymers, lipids, surfactants, rubber, or ionic liquids.
Cheminformatics module is an Artificial Intelligence (AI) powered drug molecule predictor for target proteins, the functional biomolecules that are inhibited by biologically active drug molecule. Generation of a target specific drug molecule dataset and understanding their probability of inhibiting that target is a very important element in modern bio-medical research. This module will provide the user an automated generation of target specific drug dataset either from open-source curated databases of bioactive molecules having drug like properties or by using machine-learning and deep-learning models.
Multi Target Multi Ligand Enhanced Sampling Screening (MTMLESS) module has been developed to handle multiple Multi Target Multi Ligand Enhanced Sampling Screening (MTMLESS) module has been developed to handle multiple
Current state of the art high performance computers (CPU based servers) are not cost effective for such projects as the computational demand is very high. In recent times we have innovated and redesigned our codes to run in much cheaper gaming personal computers/workstations. With Graphical Processing Units (GPU) we received 3X more performance at 3X lower cost than any comparable enterprise level CPU based server or any cloud service provider. So we are developing and optimising all our codes for GPU ready and assembling and using our own innovative high performance GPU workstation. We are offering our software platform to all interested partners with cloud enabled features. So, companies and institutions without access to any computing resources could use our platform in public cloud.