PHANOTATE: a novel approach to gene identification in phage genomes.

McNair K, Zhou C, Dinsdale EA, Souza B, Edwards RA. · 2019

Read PHANOTATE: a novel approach to gene identification in phage genomes. by McNair K, Zhou C, Dinsdale EA, Souza B, Edwards RA. (2019) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc6853651.txt.

Introduction

McNair and co-authors present a gene-finding tool in Bioinformatics built for viruses that infect bacteria. These viruses, phages, pack their genomes tightly. Genes sit side by side, overlap, or even nest inside longer genes. Tools made for bacterial genomes miss much of that. No tool had been designed for phage genomes at all.

Their program, PHANOTATE, treats the genome as a network of possible paths. Candidate genes are the routes through it, and the tight packing becomes a clue rather than a nuisance. The authors compare its calls with those of adapted tools on test genomes. They argue phage studies need this purpose-built approach as phage research grows.

Catalog record

CREATORMcNair K, Zhou C, Dinsdale EA, Souza B, Edwards RA.
YEAR2019
COLLECTIONPAPERS
FORMATTXT
FILE SIZE31 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC6853651

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

PHANOTATE: a novel approach to gene identification in phage genomes By Katelyn McNair, Carol Zhou, Elizabeth A Dinsdale, Brian Souza, Robert A Edwards, John Hancock Published in: Bioinformatics DOI: 10.1093/bioinformatics/btz265 ABSTRACT Abstract Motivation Currently there are no tools specifically designed for annotating genes in phages. Several tools are available that have been adapted to run on phage genomes, but due to their underlying design, they are unable to capture the full complexity of phage genomes. Phages have adapted their genomes to be extremely compact, having adjacent genes that overlap and genes completely inside of other longer genes. This non-delineated genome structure makes it difficult for gene prediction using the currently available gene annotators. Here we present PHANOTATE, a novel method for gene calling specifically designed for phage genomes. Although the compact nature of genes in phages is a problem for current gene annotators, we exploit this property by treating a phage genome as a network of paths: where open reading frames are favorable, and overlaps and gaps are less favorable, but still possible. We represent this network of connections as a weighted graph, and use dynamic programing to find the optimal path. Results We compare PHANOTATE to other gene callers by annotating a set of 2133 complete phage genomes from GenBank, using PHANOTATE and the three most popular gene callers. We found that the four programs agree on 82% of the total predicted genes, with PHANOTATE predicting more genes than the other three. We searched for these extra genes in both GenBank’s non-redundant protein database and all of the metagenomes in the sequence read archive, and found that they are present at levels that suggest that these are functional protein-coding genes. Availability and implementation https://github.com/deprekate/PHANOTATE Supplementary information Supplementary data are available at Bioinformatics online. INTRODUCTION Phages, viruses that infect bacteria, provide unique challenges for bioinformatics. There is a limit to how much DNA can be packaged in a capsid, and therefore phage genomes are generally short, typically in the range 20–100 kb. By necessity, their genomes are compact: phage genes are shorter than their bacterial homologs are frequently co-transcribed, and adjacent open reading frames (ORFs) often overlap (Kang et al., 2017). In a few cases, phage genes are encoded within each other (Cahill et al., 2017; Summer et al., 2007). In contrast, bacterial genes generally are longer, separated by intergenic spacers and frequently switch strands (Kang et al., 2017). There are no bioinformatics tools specifically designed to identify genes in phage genomes, so algorithms designed to identify bacterial genes are typically used (McNair et al., 2018). For example, from 31 phage genomes published between October 14, 2016 and August 1, 2018, the genes in ten phage genomes were identified by GeneMark software (GeneMark/GeneMarkS/GeneMark.hmm), the genes in 10 phage genomes were identified by RAST, the genes in 7 phage genomes by Glimmer, 3 phage genomes each by Geneious, the NCBI ORF Finder, PHAST (which uses Glimmer as a gene caller; Arndt et al., 2016), PROKKA (which uses Prodigal as a default gene caller; Seemann, 2014), 2 phage genomes by Prodigal and 1 phage genome by MetaVir, RASTtk, SerialCloner or SnapGene (Supplementary Table S1; note that in many publications several different tools were used to identify genes in phage genomes). Each of these algorithms relies on information that is not available and calculations that are not possible with short genomes. For example, there are no conserved genes in phage genomes that can be used to build universal training sets (Rohwer and Edwards, 2002), fewer genes means the statistics used to identify start codons are less accurate (Wu et al., 2003), and because many phage genes or the proteins they encode have no homolog in the databases, similarity searches are unreliable (Roux et al., 2015). There are alternate gene calling approaches, such as using positional nucleotide frequency (Besemer and Borodovsky, 1999), or the multivariate entropy of amino acid usage used by Glimmer (Ouyang et al., 2004), but these are designed for complete bacterial genomes and have not been optimized for use with phage genomes. Here, we introduce a novel method for gene identification that is specifically designed for phage genomes. We make several presumptions based on studying hundreds of phages genomes. First, we noted that since phages have physical limits on their genome sizes they contain minimal non-coding DNA. Second, we showed that phage genes are usually on the same strand of the DNA, presumably because they are co-transcribed (Akhter, 2012; Kang et al., 2017). Based on these observations, we designed a completely novel approach to phage gene identification, tiling opening reading frames to minimize non-coding DNA bases and strand switching. We treat a phage genome as a network of paths in which ORFs are more favorable, and overlaps and gaps are less favorable. We solved this weighted graph problem using the Bellman-Ford algorithm (Bellman, 1958; Ford, 1956), and by optimizing the parameters for phages genomes we are able to enhance phage gene prediction algorithms. In the absence of supporting data to confirm our new predictions, we turned to high-volume sequence similarity searches to explore the predicted proteins. Regions of the genome that encode proteins are more likely to be conserved at the amino acid level than regions that encode regulatory regions, replication regions, sites of integration and other, DNA-based, information components of the phage genome (Badger and Olsen, 1999). These searches showed that the predicted phage genes might encode novel proteins that have been missed by existing gene callers designed to annotate bacterial genomes. 2 MATERIALS AND METHODS 2.1 The PHANOTATE algorithm The first step PHANOTATE takes in identifying the genes in a phage genome is creating a weighted graph from the ORFs in that genome. By default, we allow for three start codons (codonsstart = {ATG, GTG, TTG}), and three stop codons (codonsstop = {TAA, TAG, TGA}), and the default minimum length of an ORF is 90 nt. The directed weighted graph consists of nodes that represent start and stop codons, and edges that represent either an ORF if the edge connects a start codon to a subsequent stop codon in the same reading frame; a gap if the edge connects a stop codon to a subsequent start codon in any reading frame on the same strand, or if the edge connects a stop codon to a subsequent stop codon on the alternate strand; or an overlap if the edge connects a stop codon to a preceding start codon in any other reading frame on the same strand, or to a preceding stop codon on the alternate strand. Since phages rarely have >300 bp of untranslated DNA, and to reduce computational burden, we only connect ORFs within ±300 bp of each other. When there is a very large span without an ORF, we connect ORFs on each side of the region with a linear penalty. For each edge, we calculate a weight depending on the feature type: ORF, overlap, or gap. To calculate the weight of an ORF (worf), we use an adjusted likelihood of not finding a stop codon in an ORF of this length. We count the fraction of each base in each ORF, and use that to determine the overall probability encountering a stop codon over the entire

Read the complete text