Fig. 5: DeepTernary reaches acceptable accuracy for MG(D)-induced ternary complex structure prediction.
From: SE(3)-equivariant ternary complex prediction towards target protein degradation
a Since no MG(D)-induced complex prediction method exists, we compared DeepTernary with the traditional protein–protein interaction (PPI) prediction method EquiDock. DeepTernary significantly outperforms EquiDock by precisely modeling ternary interactions. b Comparison with EquiDock and the recently realeased AlphaFold3 (AF3) on PDB ID 7BQU. The predicted aligned error (PAE) matrix from AF3 is shown in the top right corner. AF3 predicts the structure with plDDT values between 50 and 90 but shows low confidence for docking, with PAE values exceeding 20 between p1 and p2. c Visualization of the predicted ternary structure for PDB ID 4JDD (Group 2), displayed using both cartoon and surface illustrations. d Two predicted results from Group 1. e Performance comparison across different interaction modes: domain-domain (Group 1) and sequence motif-domain (Group 2). (All box plots in this figure extend from the first quartile (Q1) to the third quartile (Q3) of the data, with a line at the median. The whiskers extend from the box to the farthest data point lying within 1.5x the inter-quartile range (IQR) from the box. Flier points are those past the end of the whiskers. Statistical analyses were performed using the two-sided independent t-test. The sample number (n) is annotated under each plot. Source data are provided as a Source Data file.
