Advances in Activity/Property Prediction from Chemical Structures.

Area of Science:

• Computational chemistry
• Artificial intelligence in drug discovery
• Pharmacology

Background:

• AI modeling of molecular property databases has advanced drug design opportunities.
• Quantitative structure-activity relationships (QSARs) offer improved predictions for biological activity and toxicology.
• In-silico models and combined databases enhance drug discovery and analysis methods.

Purpose of the Study:

• To highlight the impact of AI advancements on drug design and development.
• To emphasize the role of QSARs in predicting molecular properties and toxicological assessments.
• To discuss the integration of in-silico models for generating and evaluating potential drug candidates.

Main Methods:

• Utilizing AI for modeling molecular property databases.
• Applying Quantitative Structure-Activity Relationships (QSARs) for predictive analysis.
• Combining disparate structure-activity databases and in-silico models.

Main Results:

• Improved accuracy in predicting biological activity and toxicological profiles of compounds.
• Generation of viable compounds for potential in vitro synthesis and development.
• Early discontinuation of compounds with determined toxicology, saving time and resources.

Conclusions:

• AI and QSARs significantly accelerate drug discovery and development processes.
• Expert review remains crucial for in-silico predictions, though automation is advancing.
• The field is progressing towards fully automated drug discovery and evaluation pipelines.