WIsH: who is the host? Predicting prokaryotic hosts from metagenomic phage contigs

Galiez, Clovis; Siebert, Matthias; Enault, François; Vincent, Jonathan; Söding, Johannes

doi:10.1093/bioinformatics/btx383

Cited by 228 publications

(296 citation statements)

References 13 publications

Supporting

Mentioning

294

Contrasting

Order By: Relevance

“…Until now, there are several works on the identification of host of viruses, such as HostPhinder (6), WIsH (7), which all predict hosts for bacteriophages. However they were not designed for non-phage virus host-prediction, especially the virus causing Zoonotic diseases.…”

Section: Main Textmentioning

confidence: 99%

See 1 more Smart Citation

Host and infectivity prediction of Wuhan 2019 novel coronavirus using deep learning algorithm

Guo

Wang

et al. 2020

Preprint

105

View full text Add to dashboard Cite

The recent outbreak of pneumonia in Wuhan, China caused by the 2019 Novel Coronavirus (2019-nCoV) emphasizes the importance of detecting novel viruses and predicting their risks of infecting people. In this report, we introduced the VHP (Virus Host Prediction) to predict the potential hosts of viruses using deep learning algorithm. Our prediction suggests that 2019-nCoV has close infectivity with other human coronaviruses, especially the severe acute respiratory syndrome coronavirus (SARS-CoV), Bat SARS-like Coronaviruses and the Middle East respiratory syndrome coronavirus (MERS-CoV). Based on our prediction, compared to the Coronaviruses infecting other vertebrates, bat coronaviruses are assigned with more similar infectivity patterns with 2019-nCoVs. Furthermore, by comparing the infectivity patterns of all viruses hosted on vertebrates, we found mink viruses show a closer infectivity pattern to 2019-nCov. These consequences of infectivity pattern analysis illustrate that bat and mink may be two candidate reservoirs of 2019-nCov.These results warn us to beware of 2019-nCoV and guide us to further explore the properties and reservoir of it. One Sentence Summary: It is of great value to identify whether a newly discovered virus has the risk of infecting human. Guo et al. proposed a virus host prediction method based on deep learning to detect what kind of host a virus can infect with DNA sequence as input.

show abstract

Section: Main Textmentioning

confidence: 99%

“…Considering difference in input sequence lengths, two BiPathCNNs (BiPathCNN-A and BiPathCNN-B) were built for predicting hosts of viral sequences from 100 bp to 400 bp and 400 bp to 800 bp respectively. The dataset(7) for training and testing includes genomes of all DNA viruses, coding sequences of all RNA viruses and their host information in…”

mentioning

confidence: 99%

Host and infectivity prediction of Wuhan 2019 novel coronavirus using deep learning algorithm

Guo

Wang

et al. 2020

Preprint

105

View full text Add to dashboard Cite

show abstract

“…In the meantime, the development of methods to add information to this genome-centric database is currently at full steam. These include approaches for large scale virus taxonomy classification (Bolduc et al, 2017b;Meier-Kolthoff and Göker, 2017;Nishimura et al, 2017;Aiewsakun and Simmonds, 2018), as well as host linkage for uncultivated viruses either computationally (Edwards et al, 2016;Galiez et al, 2017;Ahlgren et al, 2016) or experimentally (Tadmor et al, 2011;Martínez-García et al, 2014;Deng et al, 2014;Roux et al, 2014;Labonté et al, 2015;Spencer et al, 2016). Beyond virus-specific tools, we anticipate significant improvements in genome annotation capabilities stemming from (i) "multi 'omics approaches" combining transcriptomics, proteomics and metabolomics studies of individual environments (Franzosa et al, 2015) and (ii) improved functional prediction tools leveraging protein structure constraints and large-scale comparative genomics (Alva et al, 2016;Ovchinnikov et al, 2017).…”

Section: Wishful Thinking or Realistic Path Forward?mentioning

confidence: 99%

A viral reckoning: viruses emerge as essential manipulators of global ecosystems

Roux

Brum

2018

Environ Microbiol Rep

View full text Add to dashboard Cite

“…Several investigators have utilized the fact that viruses are similar to their hosts compared with other unrelated host species in terms of their genomic signatures (or k-mer usage) 11,[24][25][26] . They predicted the host of a virus as the one closest to the viral genome based on some dissimilarity measures using k-mers.…”

Section: Introductionmentioning

confidence: 99%

“…The recently developed dissimilarity measure d * 2 that subtracts expected k-mer frequency from the observed frequency achieves the highest accuracy among all current genomic signature-based measures, including the commonly used Euclidean and Manhattan distances 25 . Similarly, Galiez et al 26 predicted the host of a virus to be the one under whose Markov chain model the viral sequence has the highest likelihood, and the method has good accuracy for short viral fragments. The genomic signature-based measures are often referred as alignment-free sequence comparison measures.…”

Section: Introductionmentioning

confidence: 99%

A network-based integrated framework for predicting virus-host interactions

Wang

Ren

Tang

et al. 2018

Preprint

View full text Add to dashboard Cite

High-throughput metagenomic sequencing technologies have profoundly increased the ability to identify viral genomic sequences without isolation. With the discovery of new viruses, one of the most fundamental challenges is to predict their hosts. A pair of virus and host is referred as interacting if the virus infects the host. In this study, we develop a network-based integrated framework for predicting virus-host interactions based on various networks for virus-virus similarity, virus-host similarity, and known virus-host interactions. We integrate these networks systematically using a Markov random field model, in which the similarity is defined based on both alignment-based similarity measures including CRISPR matches and sequence homology, and alignment-free similarity measures including s * 2 and WIsH, to predict virus-host interactions. We evaluate our method on a benchmark dataset of 1,075 viruses and their hosts. Host prediction accuracy was 61% and 88% at the genus and phylum levels, respectively, representing a large improvement in host prediction accuracy over previous approaches. The model also markedly improves prediction accuracy for the hosts of metagenomically assembled short viral contigs. Requiring a minimum prediction score for making predictions further improves accuracy without much cost of the recall rate. We use the method to predict the hosts of new viruses discovered from single-cell amplified genomes and crAssphage, a ubiquitous virus assembled from human gut metagenomes. The predicted hosts are mostly consistent with the putative hosts, and other plausible hosts including Akkermansia muciniphila are suggested. Since Akkermansia muciniphila was shown to be associated with the effectiveness of cancer immunotherapy treatment and metabolic disorders, our study suggests that crAssphage may play important roles in these diseases. With the flexibility of incorporating other additional features, the model has great potential to be updated along with the development of the biological understanding of virus-host interactions.

show abstract

WIsH: who is the host? Predicting prokaryotic hosts from metagenomic phage contigs

Cited by 228 publications

References 13 publications

Host and infectivity prediction of Wuhan 2019 novel coronavirus using deep learning algorithm

Host and infectivity prediction of Wuhan 2019 novel coronavirus using deep learning algorithm

A viral reckoning: viruses emerge as essential manipulators of global ecosystems

A network-based integrated framework for predicting virus-host interactions

Contact Info

Product

Resources

About