In order to predict interactions (either co-occurrence or mutual exclusion) among bacteria and phages, CoNet is used to evaluate the co-occurrence and co-exclusion relationships between the abundance of bacterial genera and phage strains across samples.

 Co-occurrence relationships between phage and bacterial abundances reflect several likely interactions: (i) the increase of bacterial host abundance directly allows growth of phage populations, (ii) the increased phage populations remove a major competitor of the co-occurring bacterial population, and (iii) the increased phage populations directly infect and benefit the co-occurring bacterium (e.g. by facilitating gene transfer).

 Mutual exclusion relationships might reflect: (i) lytic phages directly reducing bacterial populations or (ii) bacteria competing with the excluded phage’s host, lowering phage populations.

Oligonucleotide ‘words’, also known as k-mers are short subsequences of a specified length that may be extracted from the genome sequence. Over time, phages ameliorate their genomic oligonucleotide usage profile to that of the host they infect. So phages and their host genomes often share similar oligonucleotide frequency (ONF) patterns, which can be used to predict the host of a given phage by finding the host with the greatest ONF similarity.

 VirHostMatcher approach is used to predict the bacterial host of phages by examining various oligonucleotide frequency (ONF) based distance/dissimilarity. The potential host of a query phage returned is the host to which it has the strongest similarity.

 WIsH approach is used to predict the prokaryotic host of metagenomic phage contigs by estimating k-mer frequencies.

Using the CRISPR system, bacteria place a short fragment of an infecting phage genome sequence, as a spacer into a CRISPR array. So the CRISPR spacers are computationally identifiable sequence signature of previous phage-host infections. The identification of phage–host links by CRISPR spacer matching is likely to be most suitable for detecting recent phage–host interactions, such as within a metagenomic sample where both bacteria and virus components have been sequenced.

It is known that many phages insert their genomes into that of their hosts; the integrated phages are known as prophages. So through identifying putative prophage sequences from bacterial genomes, the phage–host links could be inferred.

Some studies indicate that phages genomic similarity correlates well with related bacterial hosts. So through calculating the similarity between the query phage genome and the genomes of reference phages with known hosts, the links of the query phage and the known host could also be inferred.

 HostPhinder program bases its predictions on co-occurring k-mers between the query phage genome and the genomes of reference phages with known hosts.

 “GeneNet” algorithm bases its prediction on summarizing gene sharing across phages and identifying genes whose presence or absence has the strongest correspondence to phage host range.

Bacteriophages as accessory genetic elements play a crucial role in the dissemination of genes and the promotion of genetic diversity within bacterial populations. Bacteriophages can transfer genes encoding virulence factors (VF) such as toxins, adhesins and agressins, to promote the virulence of the host bacteria. Also, Antibiotic resistance genes (ARGs) in bacterial chromosomes or plasmids can be mobilized by phages during the infection cycle, to increase in antibiotic resistance.

 ShortBRED was used to search predicted ORFs against VFDB (Virulence Factors of Pathogenic Bacteria) database.

 Resistance Gene Identifier (RGI), v3.1.1 with database from July 2016, was used to search predicted ORFs against the Comprehensive Antibiotic Resistance Database (CARD)

Protein interaction experiments were used for the detection of pairs of genes whose proteins are known to interact. Interactions between bacteriophage proteins and bacterial proteins are important for efficient infection of the host cell. We assigned bacterial and phage genes to protein families by aligning them to the Pfam database using the Diamond protein alignment algorithm. We then identified which pairs of proteins were predicted to interact using the Pfam interaction information within the Intact database.