Peer Review #1 of "Multi-schema computational prediction of the comprehensive SARS-CoV-2 vs. human interactome (v0.1)"

Kevin Dick, Anand Chopra, Kyle K. Biggar, James R Green, A Amos-Binks, C Patulea, S Pitre, A Schoenrock, Y Gui, A Golshani, F Dehne, D Baltimore, A Calderone, L Licata, G Cesareni, C Chang, -k, S.-c Sue, T.-h Yu, C.-m Hsieh, C.-k Tsai, Y.-c Chiang, S.-j Lee, H.-h Hsiao, W.-j Wu, W.-l Chang, C.-h Lin, T Huang, T Chen, C Guestrin, J Cui, F Li, Z.-l Shi, B Samanfar, B Barnes, E Cober, B Mimee, S Molnar, M Donoghue, F Hsieh, E Baronas, K Godbout, M Gosselin, N Stagliano, M Donovan, B Woolf, K Robison, R Jeyaseelan, R Breitbart, S Acton, Z Fan, Y Zhuo, X Tan, Z Zhou, J Yuan, B Qiang, J Yan, X Peng, G Gao, I Glowacka, S Bertram, P Herzog, S Pfefferle, I Steffen, M Muench, G Simmons, H Hofmann, T Kuri, F Weber, J Eichler, C Drosten, S Phlmann, J Guarner, M Hoffmann, H Kleine-Weber, S Schroeder, N Krger, T Herrler, S Erichsen, T Schiergens, G Herrler, N.-h Wu, A Nitsche, M Mller, K Pyrc, L Van Der Hoek, M Geier, B BerkhoutOpenalex Unknown 2021 · DOI · OpenAlex

Abstract

Background. Understanding the disease pathogenesis of the novel coronavirus, denoted SARS-CoV-2, is critical to the development of anti-SARS-CoV-2 therapeutics. The global propagation of the viral disease, denoted COVID-19 ("coronavirus disease 2019"), has unified the scientific community in searching for possible inhibitory small molecules or polypeptides. A holistic understanding of the SARS-CoV-2 vs. human inter-species interactome promises to identify putative protein-protein interactions (PPI) that may be considered targets for the development of inhibitory therapeutics.

Introduction

To develop effective treatments for COVID-19, researchers must understand the precise molecular "handshake" between viral and human proteins. These protein-protein interactions (PPIs) allow the virus to hijack cellular machinery. While experimental methods like yeast two-hybrid screening are standard for identifying these interactions, they are often slow, costly, and prone to false positives.

In the early stages of the pandemic, the SARS-CoV-2 proteome was poorly understood. This study addresses the knowledge gap by employing a multi-schema computational pipeline to predict the comprehensive interactome. Beyond simple binary predictions, the research identifies specific binding sites (peptides), providing a blueprint for designing drugs that physically block viral entry and replication.

Methods and Frameworks

The researchers utilized a consensus-based approach, integrating two independent sequence-based PPI predictors:

  1. PIPE4 (Protein-protein Interaction Prediction Engine): Analyzes similarity between short segments of protein sequences.
  2. SPRINT (Scoring PRotein INTeractions): Uses a "spaced seed" approach for rapid sequence similarity computation.

The search was organized into three distinct schemas to ensure robustness:

  • All Schema: Leverages all known virus-human PPIs as a training proxy.
  • Proximal Schema: Focuses on PPIs from evolutionarily related viruses, such as other Coronaviruses or Zika.
  • RP-PPI (Reciprocal Perspective): A framework that evaluates interactions from both the viral and human protein perspectives to refine results.
The 'All' Schema workflow for predicting SARS-CoV-2 vs. human protein interactions.

To pinpoint physical binding sites, the team applied the PIPE-Sites algorithm, generating 2D "landscapes" of amino acid sequences likely responsible for interaction.

Key Results

The multi-schema approach yielded a hierarchy of confidence levels:

  • All Schema: 279 high-confidence interactions.
  • Proximal Schema: 129 interactions.
  • RP-PPI Schema: 539 interactions.
  • Consensus Set: Seven "highest-confidence" PPIs were identified at the intersection of all three schemas.

A critical validation step involved the Spike-ACE2 interaction. When this known interaction was removed from the training data, both PIPE4 and SPRINT still ranked it as their top prediction, demonstrating the model's ability to discover biologically relevant interactions autonomously.

Discussion and Insights

The transition from identifying "who" interacts to "where" they interact is a significant advancement. By producing 2D maps of interaction "hot spots," the study enables the design of inhibitory polypeptides—short protein chains that mimic these sites to disrupt viral binding.

Predicted interactions with HLA (Human Leukocyte Antigen) proteins and hnRNPs (Heterogeneous Nuclear Ribonuclear Proteins) are particularly noteworthy, as these are central to immune response and viral replication.

Limitations

This work represents theoretical modeling. While the computational evidence is robust, every predicted interaction requires experimental "wet-lab" validation. Additionally, peptide-based therapeutics often face challenges regarding antiviral potency compared to small-molecule drugs. Given the rapid evolution of the Spike protein, future work must integrate these structural insights with genomic surveillance to ensure therapeutics remain effective against emerging variants.

Data Availability

The full set of predictions and PIPE-Sites landscapes have been released to the public DataVerse repository to facilitate further experimental validation by the global scientific community.

Summary updated

Abstract

Background. Understanding the disease pathogenesis of the novel coronavirus, denoted SARS-CoV-2, is critical to the development of anti-SARS-CoV-2 therapeutics. The global propagation of the viral disease, denoted COVID-19 (``coronavirus disease 2019"), has unified the scientific community in searching for possible inhibitory small molecules or polypeptides. A holistic understanding of the SARS-CoV-2 vs. human inter-species interactome promises to identify putative protein-protein interactions (PPI) that may be considered targets for the development of inhibitory therapeutics.
BibTeX
@inproceedings{dick2021peer,
  author = {Kevin Dick and Anand Chopra and Kyle K. Biggar and James R Green and A Amos-Binks and C Patulea and S Pitre and A Schoenrock and Y Gui and J Green and A Golshani and F Dehne and D Baltimore and A Calderone and L Licata and G Cesareni and C Chang and -K and S.-C Sue and T.-H Yu and C.-M Hsieh and C.-K Tsai and Y.-C Chiang and S.-J Lee and H.-H Hsiao and W.-J Wu and W.-L Chang and C.-H Lin and T Huang and T Chen and C Guestrin and J Cui and F Li and Z.-L Shi and K Dick and J Green and K Dick and B Samanfar and B Barnes and E Cober and B Mimee and S Molnar and K Biggar and A Golshani and F Dehne and J Green and M Donoghue and F Hsieh and E Baronas and K Godbout and M Gosselin and N Stagliano and M Donovan and B Woolf and K Robison and R Jeyaseelan and R Breitbart and S Acton and Z Fan and Y Zhuo and X Tan and Z Zhou and J Yuan and B Qiang and J Yan and X Peng and G Gao and I Glowacka and S Bertram and P Herzog and S Pfefferle and I Steffen and M Muench and G Simmons and H Hofmann and T Kuri and F Weber and J Eichler and C Drosten and S Phlmann and J Guarner and M Hoffmann and H Kleine-Weber and S Schroeder and N Krger and T Herrler and S Erichsen and T Schiergens and G Herrler and N.-H Wu and A Nitsche and M Mller and C Drosten and S Phlmann and H Hofmann and K Pyrc and L Van Der Hoek and M Geier and B Berkhout and S Phlmann},
  title = {{Peer Review \#1 of "Multi-schema computational prediction of the comprehensive SARS-CoV-2 vs. human interactome (v0.1)"}},
  booktitle = {Openalex Unknown},
  year = {2021},
  doi = {10.7287/peerj.11117v0.1/reviews/1},
}
Figure