|
|
| Introduction of Molecular Docking-Based Drug Screening |
Molecular docking is a computational technique that predicts optimal binding
modes and affinities between ligands (e.g., drug candidates or peptides) and
protein receptors. It identifies stable ligand-receptor configurations through
simulated interactions, allowing researchers to prioritize promising
compounds-thereby conserving time, resources, and effort in early drug
discovery. |
| Why AI (Artificial Intelligence) is Revolutionizing in Drug Screening |
Traditional molecular docking relies on rigid algorithms, often limited by
computational expense and accuracy. In the early stages of drug discovery,
AI-enhanced molecular docking changes the game: |
• Faster screening
of
millions of
compounds in silico
• Higher accuracy in
predicting binding
affinities
• Lower costs by
reducing wet-lab
trial-and-error |
|
|
| Figure 1. AI Application in Drug Screening[2]. |
| Target-Based AI Screening |
AI integrates deep learning (e.g., neural networks, random forests) with docking
simulations to: |
• Predict
protein-ligand interactions with higher fidelity
• Uncover novel binding
mechanisms
• Accelerate hit-to-lead
optimization |
|
|
Figure 2. An overall flowchart for predicting protein-ligand interactions based
on DL models (concept image from MCE). |
| Ligand-based AI screening |
Researchers can leverage ligand-based AI screening to search existing compound
libraries for molecules with desired chemical and/or biological properties. They
can also use known active compounds as a training set to analyze their
characteristics with AI tools, generating similar novel molecules. AI generative
models can explore a broader chemical space to identify new compounds and design
candidates with specific drug-like characteristics, ultimately enhancing the
efficiency and success rate of drug development. |
|
|
Figure 3. Graph neural networks predict the chemical properties of more than 109
molecules in silico (concept image from MCE). |
 |
Popular Diversity Library for Virtual Screening |
|
| Product Name |
Features |
MCE Screening Compound Library1
|
A collection of over 2 million screening compounds from
6+ manufacturers available at competitive prices,
suitable for virtual screening and AI-driven screening
applications. |
MCE Screening Compound Library2
|
A collection of over 9 million screening compounds from
18+ manufacturers. The data has been cleaned, suitable
for virtual screening and AI screening. |
Lead-like Diversity Library Plus
|
Contains 80,000+ compounds with novelty, drug-likeness,
diversity are available for repeated supply, making the
library a powerful tool for new drug development. |
5K Scaffold Library
|
Contains 5,000+ compounds, using a unique Bemis/Murcko
scaffold to ensure maximum structural diversity. |
Drug Fragment Library
|
Contains 1,200+ drug fragments are derived from over
3,000 FDA approved drug molecules, and fragments from
one drug can appear in other drugs, so these fragments
are somewhat correlated with good PK/PD properties. |
Natural Product-like Compound Library
|
Contains 5,000 compounds selected based on either
natural product-derived scaffolds or Tanimoto similarity
(>0.6) to natural products. The natural-likeness scoring
of these compounds is >-2. |
Lead-like Covalent Screening Library
|
Contains 1,000+ compounds with commonly used covalent
warheads, like acrylamide, aldehyde, sulfonyl fluoride,
are capable of reacting with specific amino acid
residues, including cysteine, lysine, serine, and
histidine. |
|
|
|
|
|
|
|
|
|
Products are for research use only and are not intended for human use. We do
not sell to patients.
|
|
|
