Table 2 Performance comparisons for different methods in the absence of hand-crafted biological features (take Spearman rank correlation coefficient and mean squared error as evaluation index)
From: AttCRISPR: a spacetime interpretable model for prediction of sgRNA on-target activity
Method | WT-SpCas9 | eSpCas9(1.1) | SpCas9-HF1 | |||
|---|---|---|---|---|---|---|
|  | Spearman | MSE (× 10–3) | Spearman | MSE (× 10–3) | Spearman | MSE (× 10–3) |
XGBoost* | 0.845 | 11.7 | 0.831 | 11.5 | 0.818 | 13.5 |
MLP* | 0.842 | 11.7 | 0.846 | 10.5 | 0.844 | 11.2 |
CNN* | 0.846 | 11.3 | 0.831 | 11.3 | 0.834 | 12.0 |
RNN* | 0.856 | 10.4 | 0.849 | 10.2 | 0.851 | 10.6 |
TAC | 0.857 | 10.3 | 0.844 | 10.5 | 0.851 | 10.7 |
SpAC | 0.862 | 10.1 | 0.854 | 9.93 | 0.857 | 10.2 |
EnAC | 0.868 | 9.51 | 0.859 | 9.64 | 0.862 | 9.81 |
CRISPRpred# | 0.838 | – | 0.830 | – | 0.821 | – |