Table 3.
Comparison of Spearman correlation coefficients between eukaryotic sgRNA activity and eukaryotic model predictions
| Independent test datasets | Size | DeepCas9 | TSAM_U6 | CNN_5layers |
|---|---|---|---|---|
| chari2015Train293T | 1234 | — | 0.3812 | 0.3607 |
| doench2014HsA375 | 1276 | 0.3237 | 0.3187 | 0.3369 |
| doench2016 | 2333 | 0.3527 | 0.3439 | 0.3945 |
| hart2016-GbmAvg | 4272 | 0.3795 | 0.4242 | 0.4404 |
| hart2016-Hct1162lib1Avg | 4239 | 0.3679 | 0.4161 | 0.4288 |
| hart2016-Hct1162lib2Avg | 3617 | 0.3196 | 0.3598 | 0.3829 |
| hart2016-HelaLib1Avg | 4256 | 0.3403 | 0.3879 | 0.4033 |
| hart2016-HelaLib2Avg | 3845 | 0.3617 | 0.3942 | 0.4390 |
| hart2016-Rpe1Avg | 4214 | 0.2519 | 0.3094 | 0.3044 |
| wang2015hg19 | 2921 | 0.2030 | 0.1882 | 0.2291 |
| xu2015TrainMEsc | 981 | 0.3668 | 0.4088 | 0.4111 |
The DeepCas9 training set contains all chari2015Train293T samples