Predicting protein-peptide interaction sites using distant protein complexes as structural templates

Johansson-Åkhe, Isak; Mirabello, Claudio; Wallner, Björn

doi:10.1101/398768

Cited by 13 publications

(22 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…So far a number of computational methods have been developed for predicting the binding sites on the protein surface in PepPI predictions [32, 14, 33]. These methods learn from 3D structure information of peptide-protein complexes and can pinpoint binding sites on protein surface with relatively good accuracy.…”

Section: Resultsmentioning

confidence: 99%

“…Although the peptide drugs have increasingly attracted immense attention and the number of approved peptide therapeutics has been on the incline over the recent decades, only a few works have been proposed to exploit machine learning or deep learning methods to model peptide-protein interactions. Furthermore, for deciphering the underlying mechanisms of peptide-protein interactions, the existing approaches mainly focus on identifying peptide-binding residues on protein surface, such as the sequence-based method Pep-Bind [13] and the structure-based method InterPep [14]. PepBind [13] is a sequence-based method for peptide-binding residue prediction, which assumes that a protein would have fixed binding residues even interacting with different peptides.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

CAMP: a Convolutional Attention-based Neural Network for Multifaceted Peptide-protein Interaction Prediction

Lei

Liu

et al. 2020

Preprint

View full text Add to dashboard Cite

Peptide-protein interactions (PepPIs) are involved in various fundamental cellular functions and their identification is crucial for designing efficacious peptide therapeutics. To facilitate the peptide drug discovery process, a number of computational methods have been developed to predict peptide-protein interactions. However, most of the existing prediction approaches heavily depend on high-resolution structure data. Although several deep-learning-based frameworks have been proposed to predict compound-protein interactions or protein-protein interactions, few of them are particularly designed to specifically predict peptide-protein interactions. In this paper, We present a sequence-based Convolutional Attention-based neural network for Multifaceted prediction of Peptide-protein interactions, called CAMP, including predicting binary peptide-protein interactions and corresponding binding residues in the peptides. We also construct a benchmark dataset containing high-quality peptide-protein interaction pairs with the corresponding peptide binding residues for model training and evaluation. CAMP incorporates convolution neural network architectures and attention mechanism to fully exploit informative sequence-based features, including secondary structures, physicochemical properties, intrinsic disorder features and position-specific scoring matrix of the protein. Systematical evaluation of our benchmark dataset demonstrates that CAMP outperforms the state-of-the-art baseline methods on binary peptide-protein interaction prediction. In addition, CAMP can successfully identify the binding residues involved non-covalent interactions for peptides. These results indicate that CAMP can serve as a useful tool in peptide-protein interaction prediction and peptide binding site identification, which can thus greatly facilitate the peptide drug discovery process. The source code of CAMP can be found in https://github.com/twopin/CAMP.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

CAMP: a Convolutional Attention-based Neural Network for Multifaceted Peptide-protein Interaction Prediction

Lei

Liu

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…For each protein, n was chosen to maximize the score. As done in a previous study (Johansson-Åkhe et al, 2019), the weight assigned to each component of the score was chosen to maximize the correlation between the MCC of the prediction for each protein in the validation dataset, and its score (Fig S4B,C). This was motivated by the fact that the confidence of the model should correlate with its correctness.…”

Section: Peptide-agnostic Prediction Allows the Identification Of Putmentioning

confidence: 99%

“…Different machine learning approaches have been applied to the preliminary problem of predicting the binding sites of peptides with a varying amount of success (Johansson-Åkhe et al, 2019;Zhao et al, 2018;Taherzadeh et al, 2016Taherzadeh et al, , 2018Wardah et al, 2020;Iqbal & Hoque, 2018). Deep learning approaches have resulted in large improvements in many area, including in the domains of protein and structural biology (Senior et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

PepNN: a deep attention model for the identification of peptide binding sites

Abdin

Wen

Kim

2021

Preprint

View full text Add to dashboard Cite

Protein-peptide interactions play a fundamental role in facilitating many cellular processes, but remain underexplored experimentally and difficult to model computationally. Here, we introduce PepNN-Struct and PepNN-Seq, structure and sequence-based approaches for the prediction of peptide binding sites on a protein given the sequence of a peptide ligand. The models make use of a novel reciprocal attention module that is able to better reflect biochemical realities of peptides undergoing conformational changes upon binding. To compensate for the scarcity of peptide-protein complex structural information, we make use of available protein-protein complex and protein sequence information through a series of transfer learning steps. PepNN-Struct achieves state-of-the-art performance on the task of identifying peptide binding sites, with a ROC AUC of 0.893 and an MCC of 0.483 on an independent test set. Beyond prediction of binding sites on proteins with a known peptide ligand, we also show that the developed models make reasonable agnostic predictions, allowing for the identification of novel peptide binding proteins.

show abstract

“…Interaction sites can also be predicted using the PEP‐SiteFinder server which generates 3D de novo conformations of peptides based on their sequence and performs a fast blind rigid docking of these conformations on the complete protein surface to map the most favorable binding sites. A more homology‐based strategy is also proposed in the InterPep pipeline which uses distant protein complex structures as structural templates for the identification of residues likely involved in binding flexible peptides. InterPep begins with a search for remote homologs in a template library constructed from the PDB and including about 400,000 template pairs of interacting protein chains (including both globular and nonglobular partners).…”

Section: Modeling the Structure Of Protein Assembliesmentioning

confidence: 99%

Structural prediction of protein interactions and docking using conservation and coevolution

Andréani

Quignot

Guérois

2020

WIREs Comput Mol Sci

View full text Add to dashboard Cite

Knowledge of the detailed structure of macromolecular interactions is key to a better understanding and modulation of essential cellular functions and pathological situations. Great efforts are invested in the development of improved computational prediction methods, including binding site prediction and protein-protein docking. These tools should benefit from the inclusion of evolutionary information, since the pressure to maintain functional interactions leads to conservation signals on protein surfaces at interacting sites and coevolution between contacting positions. However, unveiling such constraints and finding the best way to integrate them into computational pipelines remains a challenging area of research. Here, we first introduce evolutionary properties of interface structures, focusing on recent work exploring evolutionary mechanisms at play in protein interfaces and characterizing the complexity of evolutionary signals, such as interface deep mutational scans. Then, we review binding site predictors and interface structure modeling methods that integrate conservation and coevolution as important ingredients to improve predictive capacity, ending with a section dedicated to the prediction of binding modes between a globular protein domain and a short motif located within an intrinsically disordered or flexible region. Throughout the review, we discuss practical applications and recent promising developments, in particular regarding the use of coevolutionary information for interface structural prediction and the integration of these coevolution signals with machine learning and deep learning algorithms.

show abstract

Predicting protein-peptide interaction sites using distant protein complexes as structural templates

Cited by 13 publications

References 49 publications

CAMP: a Convolutional Attention-based Neural Network for Multifaceted Peptide-protein Interaction Prediction

CAMP: a Convolutional Attention-based Neural Network for Multifaceted Peptide-protein Interaction Prediction

PepNN: a deep attention model for the identification of peptide binding sites

Structural prediction of protein interactions and docking using conservation and coevolution

Contact Info

Product

Resources

About