Developing a computational integrative framework, which utilizes biological processes and functional datasets (such as protein-protein interactions between disease-causing pathogens and hosts), along with pharmaceutical datasets, can help elucidate critical processes and molecular and cellular networks that are usually difficult to explore experimentally. This approach aids in extracting drug targets and identifying drugs for repositioning or repurposing against an infection. Our method involves establishing activity relationships between targets and small molecules with the help of:

Network based analysis Machine learning and data mining predictive and optimization models Reverse docking Ligand chemical similarity and pharmacophore mapping

We use available structural information and experimental data to design customized strategies to identify HIT molecules that can bind to the target protein, DNA, and RNA using virtual screening and de novo design, including:

Structure based drug design Structure similarity based screening Pharmacophore based screening Scaffold/core hopping Fragment based design LSTM/RNN/GRAPH based molecule generation De novo design Molecular dynamics simulations Alchemical free energy estimations in explicit solvent

In the expansion of identified hit molecules, medicinal chemists can test chemically synthesized analogs to determine the compounds' structure-activity relationships (SAR). SAR information is then used to improve potency and identify the core structure of molecules, selectivity, and physicochemical properties for subsequent lead optimization. SAR analysis can also be performed computationally to identify improved potency molecules and selectivity towards the target. Our approach involves customizable strategies depending on the molecules for testing through experiments. A few of the strategies are as follows:

Identify similar chemical analogs De novo design and interaction analysis using R-group and positional analogs Merging the several Hit molecules to generate optimized molecules Molecular docking and key interaction pattern analysis Generate the closely related synthesizable analogs using RNN/Graph based generated models Building relationships between free energies and analogs of the Hit compound using alchemical free energy simulations ADME/Tox predictions using predictive and QSAR models Simultaneous optimization of the physicochemical properties of compounds

To improve the drug-likeness of the lead compound, the key parameters of absorption, distribution, metabolism, excretion, and toxicity (ADMET) need to be optimized for efficacy and safety in vivo, which pave the way for good chances of success in clinical trials. Medicinal chemists perform lead optimization of the compounds using advanced organic synthesis methods or biotechnological methods. The rational design of modifications and scaffold hopping approaches can be evaluated with the help of computational chemistry tools before going to synthesis and experiments.

The collection of purchasable small molecules, easily synthesizable virtual molecules, drug combinations with specific physicochemical properties, scaffolds, structural similarities, etc., from public databases, virtual libraries, and combinatorial databases is performed with the help of chemoinformatic tools. Target-focused libraries are generated using AI/ML generative models. The data management of target-focused libraries includes Zinc, ChemBel, DrugComDB, Enamine, ChemDiv, ChemBridge, DrugBank, and more.

We will help our clients build user-friendly drug discovery platforms for life science companies by integrating end-to-end tools using our subject matter experts, data engineers, technical business analysts, and product owners.