A de novo small molecule rationally designed by a graph neural network (GNN) and reinforcement learning (RL) pipeline trained on 50,284 compounds, targeting conserved catalytic motifs across 5 viral families.
N#Cc1cnc2[nH]c(-c3ccc(N4CCN(CC5CC5)CC4)cc3)nc2c1NC(=O)C1CC1N-(4-{[(2-amino-6-{cyano}pyrimidin-4-yl)amino]methyl}phenyl)-imidazo[1,2-a]pyridine-carboxamide
Cys-X(2)-His catalytic dyad
| Viral Family | Target Site | Docking (kcal/mol) | Key Residue |
|---|---|---|---|
| SARS-CoV-2 | Conserved active site | -11.25 | Asp30 |
| Influenza | Conserved active site | -11.01 | Asp30 |
| HIV | Conserved active site | -10.21 | Lys103 |
| Dengue | Conserved active site | -10.73 | Ser600 |
| HCV | Conserved active site | -9.84 | Asp30 |
| Viral Family | Residue | Interaction | Distance (Å) |
|---|---|---|---|
| SARS-CoV-2 | Glu166 | Hydrophobic | 1.92 |
| SARS-CoV-2 | His57 | Hydrophobic | 2.64 |
| SARS-CoV-2 | Gln189 | π-stacking | 3.88 |
| Influenza | Cys145 | π-stacking | 3.83 |
| Influenza | Asn142 | Salt bridge | 3.26 |
| Influenza | Ser600 | Salt bridge | 3.25 |
| HIV | Asp30 | Salt bridge | 2.32 |
| HIV | His41 | Salt bridge | 2.25 |
| HIV | Asn142 | Hydrophobic | 1.95 |
| Dengue | Asn142 | Salt bridge | 3.97 |
| Dengue | Cys145 | Hydrophobic | 3 |
| HCV | Lys103 | π-stacking | 2.75 |
| HCV | Ser600 | π-stacking | 3.05 |
Moderate risk. Monitor with periodic lab panels during clinical use.
Moderate risk. Monitor with periodic lab panels during clinical use.
Within safe therapeutic window. No dose-limiting toxicity expected.
Moderate risk. Monitor with periodic lab panels during clinical use.
Moderate risk. Monitor with periodic lab panels during clinical use.
50,284 compounds converted to molecular graphs. Node features: atom type, hybridization, formal charge, aromaticity. Edge features: bond order, stereo.
Graph Attention Network (GAT) with 6 layers, multi-head attention, trained to predict multi-target binding affinity with cross-entropy loss.
Proximal Policy Optimization (PPO) agent in a molecular generation environment. Reward: weighted sum of multi-target docking + QED + SA − toxicity.
Top candidates validated via AutoDock Vina, 100 ns MD in explicit solvent (TIP3P), MM-GBSA free energy decomposition, and QSAR toxicity ensemble.