Table 2.
Performance of various deep neural networks models for IC50 prediction using different therapeutic and disease-state representations
Model Name | Therapeutic representation | Disease state representation | Spearman correlation |
---|---|---|---|
DrugCell5 | Morgan fingerprint | genomic | 0.80 |
Paccmann6 | SMILE language tokenization | transcriptomic | 0.88 |
DeepDSC55 | Morgan fingerprint | transcriptomic | 0.84 |
CDRScan56 | molecular fingerprint | genomic | 0.84 |
tCNNs57 | one-hot encoding | genomic | 0.84 |
GraphDRP34 | molecular graph | genomic | 0.85 |
DeepCDR35 | molecular graph | genomic and transcriptomic | 0.82 |
AGMI36 | molecular graph | genomic and transcriptomic | 0.92 |
Spearman correlations were reported in the papers referenced or in.36