InterPep2: global peptide-protein docking using interaction surface templates

Abstract

Motivation: Interactions between proteins and peptides or peptide-like intrinsically disordered regions are involved in many important biological processes, such as gene expression and cell life-cycle regulation. Experimentally determining the structure of such interactions is time-consuming and difficult because of the inherent flexibility of the peptide ligand. Although several prediction-methods exist, most are limited in performance or availability.

Results: InterPep2 is a freely available method for predicting the structure of peptide-protein interactions. Improved performance is obtained by using templates from both peptide-protein and regular protein-protein interactions, and by a random forest trained to predict the DockQ-score for a given template using sequence and structural features. When tested on 252 bound peptide-protein complexes from structures deposited after the complexes used in the construction of the training and templates sets of InterPep2, InterPep2-Refined correctly positioned 67 peptides within 4.0 Å LRMSD among top10, similar to another state-of-the-art template-based method which positioned 54 peptides correctly. However, InterPep2 displays a superior ability to evaluate the quality of its own predictions. On a previously established set of 27 non-redundant unbound-to-bound peptide-protein complexes, InterPep2 performs on-par with leading methods. The extended InterPep2-Refined protocol managed to correctly model 15 of these complexes within 4.0 Å LRMSD among top10, without using templates from homologs. In addition, combining the template-based predictions from InterPep2 with ab initio predictions from PIPER-FlexPepDock resulted in 22% more near-native predictions compared to the best single method (22 versus 18).

Availability and implementation: The program is available from: http://wallnerlab.org/InterPep2.

Supplementary information: Supplementary data are available at Bioinformatics online.

Fig. 1.

Summary of the InterPep2 method

Fig. 2.

The number of the 251 complexes in the bound test set that was correctly modeled using different performance measures. From bottom to top (dark to lighter shades), number of correct models at top1, number of correct models for best of top10 and number of top1 models with the peptide at the correct site. The two columns for top1 and top10 denote LRMSD (left) and fnat (right) as a criteria for docking success, respectively

Fig. 3.

Precision–recall curve showing the capability of InterPep2 and GalaxyPepDock to correctly rank their own predictions, calculated using the top1 prediction for each target in the bound test set

Fig. 4.

Comparison of the ability of different methods to produce correctly docked peptide structures on the unbound set among top10 [data for HADDOCK, pepATTRACT and PIPER-FlexPepDock from (Alam et al., 2017)]

Fig. 5.

The number of the 252 complexes in the PDB16–19 set that was correctly modeled by the different methods using different performance measures. From bottom to top (dark to lighter shades), number of correct models at top1, number of correct models choosing the best of top10 for each target and number of top1 models with the peptide at the correct site. In the cases of correct at top1 and top10, the left columns denote using LRMSD as a criteria for docking success, whereas the right columns denote using fnat

Fig. 6.

An example of a successful InterPep2 prediction on the complex of DNMT3a ADD domain binding to H3 peptide [PDB: 4QBQ (Noh et al., 2015)]. (A) The top10 predictions made by InterPep2, peptides in green, as well as the native structure, peptide in pink. The receptor is blue and its surface a semitransparent gray. (B) A closer look at the top1 prediction from InterPep2, in green, together with the native peptide, in pink. The predicted peptide is positioned 2.1 Å RMSD from the native peptide conformation, counting backbone positions. The images were constructed through PyMOL (Schrödinger, 2015). (Color version of this figure is available at Bioinformatics online.)

Fig. 7.

Histogram for the predicted DockQ scores of the top10 predictions for every target, separated by if the prediction found the correct site, another kind of binding site or an incorrect site. All three distributions are significantly different, with P < 0.0001 using d by two-sided Kolmogorov–Smirnov test