Skip to main content
. 2019 Aug 2;21(4):1437–1447. doi: 10.1093/bib/bbz081

Table 4.

The total numbers of proteins and the EFs of three de novo (SVM, KNN and PNN) identified by two similarity-based (BLAST and HMMER) and the deep learning (CNN) methods proposed in this study based on the training and testing data sets (for all 20 studied GO families) of the lowest similarity. Each GO family was indicated by its GO ID, and its corresponding GO term was provided in Table 1

GO ID No. of proteins identified by each method EF
CNN SVM KNN PNN BLAST PHMM CNN SVM KNN PNN BLAST PHMM
GO:0009975 56 235 657 1416 1594 1347 53.61 0 2.28 3.18 8.48 11.14
GO:0097472 202 434 619 561 1598 923 44.59 14.65 8.56 13.22 8.29 14.92
GO:0004721 352 712 1824 1402 2403 1844 32.73 12.36 5.17 7.17 6.28 8.41
GO:0003712 637 1403 2648 3114 4408 3541 10.79 3.38 2.19 2.40 2.80 3.65
GO:0051347 725 1689 3873 1421 5858 4521 9.15 3.37 1.22 4.17 2.43 3.22
GO:0043086 1087 2291 3822 528 5911 4888 7.75 2.63 1.92 9.37 2.14 2.67
GO:0016746 924 2173 3878 401 2076 1604 12.61 3.51 2.51 17.44 7.76 10.44
GO:0019787 1954 2339 3822 3921 3328 2499 7.72 4.79 2.67 3.06 5.02 6.62
GO:0016757 881 1265 4146 2352 1770 1451 11.20 6.05 2.56 3.88 8.36 10.20
GO:0003735 520 750 965 1564 1207 1029 33.03 21.22 17.8 10.98 14.58 17.10
GO:0051336 1436 4334 6885 6693 7187 5887 3.76 1.10 0.81 1.26 1.81 2.28
GO:0043085 2152 4710 8846 8082 7984 6470 4.27 1.78 1.23 1.53 1.83 2.29
GO:0044093 3458 5172 9417 2664 8654 7048 2.65 1.61 1.10 2.79 1.69 2.07
GO:0008233 2850 3842 8106 5906 5409 4327 3.52 1.85 1.18 1.50 2.24 2.77
GO:0003700 3877 6586 6861 6604 6537 5167 4.07 1.87 1.66 1.84 2.58 3.27
GO:0016788 3975 4867 9351 7019 5338 4414 2.45 1.75 1.18 1.52 2.72 3.33
GO:0004672 2904 5918 8287 4895 6621 4633 3.87 1.40 1.35 2.05 2.21 2.96
GO:0016817 2207 5443 10 649 8903 6531 5053 5.15 1.89 1.24 1.48 2.43 3.13
GO:0038023 4230 5919 7314 6684 5616 3573 3.06 1.93 1.79 2.03 2.60 3.48
GO:1902494 4133 6900 10 017 8399 6466 4991 2.45 1.31 1.04 1.35 2.16 2.84