

QSAR modeling is used to create quantitative mathematical models between chemical structures and biological activities. QSAR models based on AI/ML algorithms have the potential the constraints by early methods and the capability to accelerate the drug discovery process by several folds.


University of Illinois at Urbana-Champaign

- peptide synthesis and small molecule design
- identification of drug dosage 
- prediction of bioactive agents
- protein–protein interactions
- identification of protein folding and misfolding
- structure and ligand-based VS
- QSAR modeling
- prediction of physiochemical 
- prediction of drug toxicity

Application of AI in Drug Discovery

Gupta, R., Srivastava, D., Sahu, M. et al. Artificial intelligence to deep learning: machine intelligence approach for drug discovery. Mol Divers (2021). https://doi.org/10.1007/s11030-021-10217-3

Artificial intelligence to deep learning: machine intelligence approach for drug discovery 

Peptides are small chains of 2-50 amino acids.They are being explored for therapeutic purposes as they have the ability to cross cellular barriers and reach the target site. AI tools are being used for the creation of new peptides.


Small molecules are molecules with very low molecular weight. SImilar to peptides, they are also being explored for therapeutic purposes using AI tools.  

Peptide synthesis and Small Molecule design using AI

 Identifying the proper safe dosage of a drug is crucial. This has been a difficult task to achieve by making sure the drug achieves the desired efficay with minimal side-effects. But with the advent of AI, researchers are making use of ML and DL algorithms to determine the appropriate drug dosage.

Identification of drug dosage using AI






Determing bioactive ligands is an important step for selecting a potent drug for a specific target. AI is being used by researchers to determine bioactive compounds which can be used for specific targets associated with a disease. 






Prediction of Bioactive Agents using AI


Analyzing protein-protein interations (PPI) is an importannt part of drug development. Bayesian Network (BN) has been used to predict PPIs using gene co-expression, gene ontology and other biological processes. Machine and statisticlal learning approaches (KNN, Naive Bayes, SVM, etc.) have been used to predict hindrance in PPIs. 
.

Prediction of protein-protein interactions through AI




Proteins are longs chains consisting of amino acids. For a protein to perform its task in a cell, it must "fold". Folding consists of transforming the molecule into a complex 3-D structure that will then interact with its target in the cell. Predicting how a protein will fold using AI can be crucial for revolutionizing drug design. 


Identification of protein folding and misfolding using AI




Virtual screening is used for identifying small chemical compounds that bind to a drug target. Ligand-based VS is different from Structure based as it does not depend on the 3-D structure of the target protein and hence, has a lower accuracy.

Machine learning has the capability of speeding up VS and producing more accuracy by generating less false positives.

Structure and Ligand-based Virtual Screening using AI

QSAR modeling using AI



The physiochemical properties of compounds must be heavily considered while designing new drug molecules as it may hinder the compound and drug target binding efficiency. To aid this, different AI-based tools have been developed to predict the physiochemical properties of chemical compounds.




Learn Before

Related