Table 2 The identification results of the proposed method based on different modes with Skip-TSSS.
From: A deep learning framework combined with word embedding to identify DNA replication origins
Dataset | Training mode | Acc | Sp | Sn | MCC | AUC |
|---|---|---|---|---|---|---|
S. cerevisiae (\(S_{1}\)) | Default mode | 0.969 | 0.962 | 0.976 | 0.939 | 0.990 |
Embedding training mode | 0.975 | 0.966 | 0.983 | 0.940 | 0.975 | |
Two channel mode | 0.972 | 0.962 | 0.981 | 0.944 | 0.972 | |
S. pombe (\(S_{2}\)) | Default mode | 0.731 | 0.741 | 0.720 | 0.461 | 0.754 |
Embedding training mode | 0.751 | 0.742 | 0.760 | 0.502 | 0.785 | |
Two channel mode | 0.765 | 0.780 | 0.750 | 0.530 | 0.800 | |
K. lactis (\(S_{3}\)) | Default mode | 0.867 | 0.894 | 0.840 | 0.735 | 0.903 |
Embedding training mode | 0.857 | 0.872 | 0.842 | 0.714 | 0.891 | |
Two channel mode | 0.811 | 0.824 | 0.797 | 0.622 | 0.852 | |
P. pastoris (\(S_{4}\)) | Default mode | 0.933 | 0.946 | 0.921 | 0.866 | 0.968 |
Embedding training mode | 0.956 | 0.964 | 0.948 | 0.912 | 0.972 | |
Two channel mode | 0.949 | 0.964 | 0.933 | 0.898 | 0.972 |
