Machine learning techniques for sequence-based prediction of viral–host interactions between SARS-CoV-2 and human proteins

Dey, Lopamudra; Chakraborty, Sanjay; Mukhopadhyay, Anirban

doi:10.1016/j.bj.2020.08.003

Cited by 89 publications

(64 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Pairs of the human and virus proteins that do not appear in the positive PPI dataset are randomly sampled as negative data. However, the random sampling method may incorrectly assign many positive samples to negative ones [5, 20]. To address this problem, the dissimilarity negative sampling method was developed [5], which used a sequence similarity-based method to explore the protein pairs that are unlikely to interact.…”

Section: Methodsmentioning

confidence: 99%

“…DeepViral also encoded host phenotype associations from PathoPhenoDB [18] and protein functions from the Gene Ontology (GO) database (The Gene Ontology Consortium, 2017) to predict PPIs. In addition, several constructed ML models were designed for certain individual virus species, limiting their generalizability to other human host-virus systems [19][20][21].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

LSTM-PHV: Prediction of human-virus protein-protein interactions by LSTM with word2vec

Tsukiyama

Hasan

Fujii

et al. 2021

Preprint

View full text Add to dashboard Cite

Viral infection involves a large number of protein-protein interactions (PPIs) between human and virus. The PPIs range from the initial binding of viral coat proteins to host membrane receptors to the hijacking of host transcription machinery. However, few interspecies PPIs have been identified, because experimental methods including mass spectrometry are time-consuming and expensive, and molecular dynamic simulation is limited only to the proteins whose 3D structures are solved. Sequence-based machine learning methods are expected to overcome these problems. We have first developed the LSTM model with word2vec to predict PPIs between human and virus, named LSTM-PHV, by using amino acid sequences alone. The LSTM-PHV effectively learnt the training data with a highly imbalanced ratio of positive to negative samples and achieved an AUC of 0.976 with an accuracy of 98.4% using 5-fold cross-validation. By using independent test dataset, we compared the LSTM-PHV with existing state-of-the-art PPI predictors including DeepViral. In predicting PPIs between human and unknown or new virus, the LSTM-PHV presented higher performance than the existing predictors when they were trained by multiple host protein-including datasets. LSTM-PHV learnt multiple host protein sequence contexts more efficiently than the DeepViral. Interestingly, learning of only sequence contexts as words presented remarkably high performances. Use of uniform manifold approximation and projection demonstrated that the LSTM-PHV clearly distinguished the positive PPI samples from the negative ones. We presented the LSTM-PHV online web server that is freely available at http://kurata35.bio.kyutech.ac.jp/.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

LSTM-PHV: Prediction of human-virus protein-protein interactions by LSTM with word2vec

Tsukiyama

Hasan

Fujii

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…In this approach, the intra-race training of [25] using a combination of human PPI and Bacillus Anthracis data of different species was done for inter-race forecasting, resulting in a binary classifier that predicted traces of Bacillus Anthracis in human with moderate accuracy of 89.0%. Recently, Dey et al [97] employed various ML models, including SVM, to predict interactions between SARS-CoV2 and human protein pairs, wherein the SVM method performed adequately for RBF and polynomial kernels, with the accuracy of 69.67% and 68.03%, respectively. However, [97] proposed an ensemble technique that outperformed the other models with an accuracy of 72.33%.…”

Section: A Research Directionmentioning

confidence: 99%

“…Recently, Dey et al [97] employed various ML models, including SVM, to predict interactions between SARS-CoV2 and human protein pairs, wherein the SVM method performed adequately for RBF and polynomial kernels, with the accuracy of 69.67% and 68.03%, respectively. However, [97] proposed an ensemble technique that outperformed the other models with an accuracy of 72.33%. Lastly, we present a summary of the contributions and limitations of the publications reviewed, listed in Table 17.…”

Section: A Research Directionmentioning

confidence: 99%

Determining Protein–Protein Interaction Using Support Vector Machine: A Review

Chakraborty¹,

Mitra²,

et al. 2021

IEEE Access

View full text Add to dashboard Cite

Protein-Protein Interaction (PPI) is a network of protein interconnections which regulates most of the biological methods. A sound state of biota largely depends on synchronized interactions between protein molecules, and any aberrant interactions between protein molecules may lead to complications, including cervical leukemia, tuberculosis, and other neural disorders. In PPI investigation, a plethora of computational methods have been developed over the years to analyze and predict PPI conclusively; however, a majority of these techniques proved to be strenuous and expensive. Therefore, the need for faster, accurate, and critical analysis of PPI warrants the adoption of Machine Learning (ML) methods such as Support Vector Machine (SVM), Artificial Neural Network (ANN), and Random Forest Model (RFM). These classifiers are useful in PPI unfolding in terms of amino acid sequence data. The SVM classifier, in particular, is serviceable in solving a majority of complex classification problems producing robust results in a reasonable time frame. This publication summarizes some state-of-art SVM based PPI investigations and challenges incurred in the application of the SVM method.INDEX TERMS Artificial neural network, machine learning, protein-protein interaction, support vector machine.• Detection of protein complexes.• Identification of domain interactions.

show abstract

“…Molecular-dynamics simulations have massively extended the utility of experimental structural data, providing novel insights via the rapid analysis of in silico mutagenesis for many of the key viral proteins Rynkiewicz et al, 2021;Sheik Amamuddy et al, 2020) and the ability to model glycosylation of the spike protein Woo et al, 2020). Computational approaches have also helped to probe potential host-pathogen protein interactions (HPIs), contributing network-based and machine-learning-based assessments of putative interactions leveraging multiple HPI databases (Dey et al, 2020;Messina et al, 2020). One ambitious project, Folding@Home, has even employed crowdsourcing to access exascale computing resources for multiple SARS-CoV-2 simulation projects (Achdout et al, 2020;Zimmerman & Bowman, 2021).…”

Section: Computational Approaches For Sas-cov-2 Proteinsmentioning

confidence: 99%

Structural biology in the time of COVID-19: perspectives on methods and milestones

Lynch

Snell

Bowman

2021

Int Union Crystallogr J

View full text Add to dashboard Cite

The global COVID-19 pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has wreaked unprecedented havoc on global society, in terms of a huge loss of life and burden of morbidity, economic upheaval and social disruption. Yet the sheer magnitude and uniqueness of this event has also spawned a massive mobilization of effort in the scientific community to investigate the virus, to develop therapeutics and vaccines, and to understand the public health impacts. Structural biology has been at the center of these efforts, and so it is advantageous to take an opportunity to reflect on the status of structural science vis-à-vis its role in the fight against COVID-19, to register the unprecedented response and to contemplate the role of structural biology in addressing future outbreak threats. As the one-year anniversary of the World Health Organization declaration that COVID-19 is a pandemic has just passed, over 1000 structures of SARS-CoV-2 biomolecules have been deposited in the Worldwide Protein Data Bank (PDB). It is rare to obtain a snapshot of such intense effort in the structural biology arena and is of special interest as the 50th anniversary of the PDB is celebrated in 2021. It is additionally timely as it overlaps with a period that has been termed the `resolution revolution' in cryoelectron microscopy (CryoEM). CryoEM has recently become capable of producing biomolecular structures at similar resolutions to those traditionally associated with macromolecular X-ray crystallography. Examining SARS-CoV-2 protein structures that have been deposited in the PDB since the virus was first identified allows a unique window into the power of structural biology and a snapshot of the advantages of the different techniques available, as well as insight into the complementarity of the structural methods.

show abstract

Machine learning techniques for sequence-based prediction of viral–host interactions between SARS-CoV-2 and human proteins

Cited by 89 publications

References 43 publications

LSTM-PHV: Prediction of human-virus protein-protein interactions by LSTM with word2vec

LSTM-PHV: Prediction of human-virus protein-protein interactions by LSTM with word2vec

Determining Protein–Protein Interaction Using Support Vector Machine: A Review

Structural biology in the time of COVID-19: perspectives on methods and milestones

Contact Info

Product

Resources

About