Table 1 Average of five-fold cross validation of deep learning models in Human and Mouse Hi-C datasets given various TAD: non-TAD ratios
From: Pattern recognition of topologically associating domains using deep learning
Models | 1:1 | 1:1.5 | 1:2 | ||||||
|---|---|---|---|---|---|---|---|---|---|
Acc. | F1 | AUC | Acc. | F1 | AUC | Acc. | F1 | AUC | |
_Human | |||||||||
SE-FCN | 0.892 | 0.894 | 0.889 | 0.914 | 0.895 | 0.914 | 0.915 | 0.872 | 0.905 |
SE-ResNet | 0.877 | 0.869 | 0.877 | 0.897 | 0.867 | 0.888 | 0.898 | 0.835 | 0.873 |
_Mouse | |||||||||
SE-FCN | 0.836 | 0.825 | 0.836 | 0.857 | 0.809 | 0.846 | 0.879 | 0.817 | 0.866 |
SE-ResNet | 0.868 | 0.853 | 0.856 | 0.883 | 0.856 | 0.883 | 0.880 | 0.812 | 0.858 |