Skip to main content

Table 11 The results of comparison between MKSVM (with MKL-CKA) model and other existing methods on PDB186 data set (independent test)

From: A sequence-based multiple kernel model for identifying DNA-binding proteins

Methods ACC (%) MCC SN (%) Spec (%)
IDNA-Prot|dis 72.0 0.445 79.5 64.5
IDNA-Prot 67.2 0.344 67.7 66.7
DNA-Prot 61.8 0.240 69.9 53.8
DNAbinder 60.8 0.216 57.0 64.5
DBPPred 76.9 0.538 79.6 74.2
iDNAPro-PseAAC 71.5 0.442 82.8 60.2
Kmer1+ACC 71.0 0.431 82.8 59.1
Local-DPP 79.0 0.625 92.5 65.6
DPP-PseAAC 77.4 0.550 83.0 70.9
Adilina’s work 82.3 0.670 95.0 69.9
MKSVM-HKA 81.2 0.648 94.6 67.7
MSFBinder 79.6 0.616 93.6 65.6
FKRR-MVSF 81.7 0.676 98.9 64.5
RF* 79.0 0.593 89.3 68.8
FNN* 75.3 0.520 87.1 63.4
Our method* (MKSVM with MKL-CKA) 83.7 0.691 93.6 74.2
  1. The bold font indicates the largest value in the column
  2. * The model is built via the 6 types of our features