Artificial neural network (ANN) based modelling for D1 like and D2 like dopamine receptor affinity and selectivity

Area of Science:

• Pharmacology
• Medicinal Chemistry
• Computational Chemistry

Background:

• Dopamine receptors (five subtypes: D1-like and D2-like) are crucial in neurological disorders such as Parkinson's disease and schizophrenia.
• Developing subtype-specific drugs is difficult due to highly similar receptor binding sites, leading to potential side effects.

Purpose of the Study:

• To identify key molecular characteristics that enable selective binding to dopamine D1-like and D2-like receptors.
• To utilize quantitative structure-activity relationships (QSARs) for predicting and understanding receptor-ligand interactions.

Main Methods:

• Developed QSAR models using datasets of 29 (D1-like) and 69 (D2-like) molecules with known dissociation constants (Ki).
• Employed hybrid neural network modeling with 62 theoretical molecular descriptors (categorical and continuous) as input features.
• Incorporated categorical molecular descriptors to enhance model performance.

Main Results:

• QSAR models successfully identified distinct molecular features for D1-like and D2-like receptor selectivity.
• Secondary amines and nitrogen-containing groups were crucial for D1-like receptor binding.
• Molecular size, volume, and the presence of tertiary/quaternary carbons were significant for D2-like receptor binding.

Conclusions:

• This research provides valuable insights into the structural requirements for selective dopamine receptor ligands.
• The findings can guide the rational design of novel therapeutics with improved efficacy and reduced side effects for neurological disorders.
• QSAR modeling proves effective in dissecting complex structure-activity relationships in drug discovery.