Skip to main content

Table 2 Performance comparison with other methods on binary complex dataset

From: PPI-Graphomer: enhanced protein-protein affinity prediction using pretrained and graph transformer models

Method

Test set 1

Test set 2

PCC\(\uparrow\)

PCC\(\uparrow\)

\({\hbox {Rosetta-InterfaceAnalyze}}^1\)

0.482

0.054

\({\hbox {Foldx-AnalyseComplex}}^2\)

0.076

\(-\) 0.159

\({\hbox {ColabFold-iPAE}}^3\)

0.588

0.156

ours

0.708

0.633

  1. \(^{1}\)The dG_separated score generated by RosettaFold’s InterfaceAnalyzer to indicate binding affinity. Prior to this, the PDB files were processed using relax and jd_score2
  2. \(^{2}\)The Interaction Energy score generated by FoldX’s AnalyseComplex module to indicate binding affinity
  3. \(^{3}\)The iPAE yielded by ColabFold to indicate binding affinity