Table 2.
ML and DL-based applications in different areas of drug discovery and development
| Sl. No. | Software/tools | Method | Web address | Remarks | Reference |
|---|---|---|---|---|---|
| 1. | AutoGrow4 | Genetic algorithm | http://durrantlab.com/autogrow4 | It is used for the de novo drug designing and lead optimization purposes | Spiegel and Durrant68 |
| 2. | TrixX | ML | – | The structure-based molecule catalog applied for extensive virtual screening in sublinear time | Schellhammer and Rarey69 |
| 3. | LS-align | ML | http://zhanglab.ccmb.med.umich.edu/LS-align/ | The atomic-level, flexible ligand structural alignment algorithm used for high-throughput virtual screening | Hu et al.70 |
| 4. | StackCBPred | ML | https://bmll.cs.uno.edu/ | The prediction of protein-carbohydrate binding sites from the available sequence using the stacking-based | Gattani et al.71 |
| 5. | DrugFinder | ML | https://drugfinder.ca/ | In silico virtual screening service for search a drug or medical condition | Lagarde et al.72 |
| 6. | LigGrep | ML | http://durrantlab.com/liggrep/ | The web tool applied for filtering docked complex to improve virtual-screening hit rates | Ha et al.73 |
| 7. | LSA | Conventional similarity algorithms | – | The local-weighted structural alignment web tool used for virtual pharmaceutical screening | Li et al.74 |
| 8. | DEEPScreen | CNNs | https://github.com/cansyl/DEEPscreen | The web tool used for high-performance DTI prediction | Rifaioglu et al.75 |
| 9. | DLIGAND2 | Distance-scaled | https://github.com/sysu-yanglab/DLIGAND2 | This web toll used for analysis of improved knowledge-based energy purpose for protein–ligand interactions | Chen et al.76 |
| 10. | Dr.VAE | ML | https://github.com/rampasek/DrVAE | It models both the drug response in relations of viability and the cellular transcriptomic perturbations | Rampášek et al.77 |
| 11. | SMDIP | ML | – | It shows the pharmacokinetic and pharmacodynamic profiles of the drug molecules | Ibrahim et al.78 |
| 12. | MoleculeNet | ML | https://moleculenet.org/ | Used for accurate predictions about molecular properties of drug –and its comparison | Wu et al.79 |
| 13. | DTI-CNN | DL | – | The DTI prediction tool performed to outperform the prevailing state-of-the-art methods by the intelligent interface | Nag et al.67 |
| 14. | DeepDTA | DL | https://github.com/hkmztrk/DeepDTA | The non-structure-based method and usages SMILES as input data for drugs. The amino acids sequences are likewise encoded in SMILES | Ozturk et al.80 |
| 15. | WideDTA | DL | – | The web tool holds text-based sources of information as input, where the proteins are signified by smaller lengths of residues are not identified in full-length sequence | Nag et al.67 |
| 16. | PADME | DL | – | This tool predict method, which usages drug molecules-target landscapes and fingerprints (as the input) | Feng et al.81 |
| 17. | DeepAffinity | DL | – | Structural property sequence representation that annotates the sequence with structural information, which are shorter than the other representations, provides structural details efficiently and gives higher resolution of the sequences | Karimi et al.82 |
| 18. | DeepChem | DL | https://github.com/deepchem/deepchem | The DNNs applied to analyze medicines and predict drug-related features, including as bioactivities and physicochemical qualities | Altae-Tran et al.83 |
| 19. | DeepConv-DTI | DL | https://github.com/GIST-CSBL/DeepConv-DTI. | This tool predict the model capturing local residue patterns of proteins in identification of DTIs | Lee et al.84 |
| 20. | DeepCPI | DL | https://github.com/FangpingWan/DeepCPI | It accurately predict and identification of compound-protein interactions at a large scale | Wan et al.85 |
| 21. | DeepDTnet | DL | https://github.com/ChengF-Lab/deepDTnet | This DL-based tool offers a potent network-based methodology for identification of target to expedite drug repurposing and reduce the translational gap in drug development | Zeng et al.86 |
| 22. | DeepGRMF | DL | https://github.com/renshuangxia/DeepGRMF | It offers anintegrated graph models, NNs, and matrix-factorization methods to operate diverse information from drug chemical structures and predict cell response to drugs | Ren et al.87 |
| 23. | DeepLIFT | DL | https://github.com/kundajelab/deeplift | This tool predicts drug response in cancer cell lines and their mechanism of action and evaluated its performance using three cross-validation schemes | Sada Del Real and Rubio88 |
| 24. | DeepSide | DL | http://github.com/OnurUner/DeepSide | It exploits data concerning the drug targets, structural fingerprints, and drug side effects | Arshed et al.89 |