GLP1R library

GLP1R library 13938 compounds. Discovery of GLP1R agonists.

GLP-1R (Glucagon-like peptide-1 receptor) is a crucial target for therapeutic agent development because its activation plays a significant role in regulating glucose levels and appetite. GLP-1R agonists are used for treating type 2 diabetes and obesity, making the discovery of new, effective compounds particularly important.

The following technologies were employed to identify new potential GLP-1R agonists:

 1. Tanimoto + PhysChem Filters Approach :

• Physicochemical Filters: Criteria included the number of hydrogen bond donors and acceptors, the log partition coefficient (LogP), molecular weight, the number of rotatable bonds, and total polar surface area (TPSA).

• ChemDiv Library: Initially contained 1,666,984 compounds, which were filtered down to 85,874 compounds after applying the physicochemical filters.

• Tanimoto Similarity: Used to search for compounds with a specified similarity, employing threshold values of 0.3 and 0.4 for 256 and 512 bits. This reduced the number of compounds to 4,315.

 2. Molecular Docking:

• Docking Score: Identifying potential agonists based on their binding affinity to GLP-1R using molecular docking. Compounds with the best docking scores (up to -11) were selected.

 3. Deep Learning:

•  Deep Learning Model: Trained using data from various patents and public databases to classify active and inactive compounds.

• Model Architecture: Included three Graph Attention Layers (GAT), a Global Mean Pooling layer, and two fully connected layers with a sigmoid activation function for final classification.

• Results: The model showed high metrics for accuracy, ROC AUC, and PR AUC, indicating its effectiveness in predicting compound activity.