AI/ML and high compute technologies, enhanced by
Enhanced by Nvidia Inception
Mapping of interactions to potency- On-target screening
- Pharmacophore/QSAR/Docking models
- Screening of large virtual databases
- Identification of hits and compound design
- Hit quality assessment through preliminary filters
- Iterative design to optimize hits into leads
- Fragment screening and de-novo design
Structure
Prediction
Prediction
Deep
Learning
Learning
Homology
Modeling
Modeling
Pocket
Druggability
Druggability
Patent
Digitization
Digitization
MPO
LLM
Clustering
Interactogram
Hydrogram
Molecular
Docking
Docking
Covalent
Docking
Docking
Protein Protein
Docking
Docking
Generative
Chemistry
Chemistry
Virtual
Screening
Screening
Ultra Large
Library Generation
Library Generation
Molecular
Dynamics
Dynamics
RNA Docking
DFT
QSAR
Machine
Learning
Learning
SAR
MMP
PROTAC
Modeling
Modeling
Conformational
Search
Search
Receptor Flexible
Docking
Docking
Chemo-
informatics
informatics
Free Energy
Estimation
Estimation
R-group
Analyses
Analyses
Peptide
Design
Design
FBDD
DEL/ASMS
Screen Planning
Screen Planning
Pharmacophore
MDTools
Explainable AI
- thinkMolecular takes an AI/ML augmented molecular simulations approach to enable drug design. The advent of high performance computing through GPUs has brought transformational technologies like the foundational and deep learning models to drug design. Enhanced by the Nvidia Inception Program, we have been quick to embrace these latest technologies and deploy them in a fit-for-purpose manner in our drug discovery projects.
- Protein structure and ligand MoA prediction through deep learning and foundational models are routinely accomplished thorough our access to the Nvidia inception program.
- The predicted structural models are subjected to long time scale MD simulations which are again GPU intensive
- Some of the highlights of our MD runs
- Largest box so far: 500000 ~ atoms
- Longest simulation: 1.2 µs
- Number of simulations: >1000
- Best ns/day: 400
- Diversity of systems: Enzymes, membrane bound proteins, RNAs, RNA-protein complexes, DNA-protein complexes, Small molecules, metalloproteins, PROTAC ternary complexes, Molecular glue complexes, Radioligand complexes, Proteins with modified residues
- Outputs: Interaction hotspots, Interactograms (IFP), Torsion profiles, RoG, RMSD, RMSF, PE, binding pocket dynamics and more
Structure
prediction
prediction
Ligand MoA
Molecular
dynamics
dynamics
Analyses –
thinkMolecular
MD tools
thinkMolecular
MD tools
Design insights