Search

For any given bacteriophage genome or phage-derived sequences in metagenomic data sets, we are unable to assign a function to 50-90% of genes, or more. Structural protein-encoding genes constitute a large fraction of the average phage genome and are among the most divergent and difficult-to-identify genes using homology-based methods. To understand the functions encoded by phages, their contributions to their environments, and to help gauge their utility as potential phage therapy agents, we have developed a new approach to classify phage ORFs into ten major classes of structural proteins or into an "other" category. The resulting tool is named PhANNs (Phage Artificial Neural Networks). We built a database of 538,213 manually curated phage protein sequences that we split into eleven subsets (10 for cross-validation, one for testing) using a novel clustering method that ensures there are no homologous proteins between sets yet maintains the maximum sequence diversity for training. An Artificial Neural Network ensemble trained on features extracted from those sets reached a test F1-score of 0.875 and test accuracy of 86.2%. PhANNs can rapidly classify proteins into one of the ten structural classes or, if not predicted to fall in one of the ten classes, as "other," providing a new approach for functional annotation of phage proteins. PhANNs is open source and can be run from our web server or installed locally.

Abstract Background Metagenomics is the study of the microbial genomes isolated from communities found on our bodies or in our environment. By correctly determining the relation between human health and the human associated microbial communities, novel mechanisms of health and disease can be found, thus enabling the development of novel diagnostics and therapeutics. Due to the diversity of the microbial communities, strategies developed for aligning human genomes cannot be utilized, and genomes of the microbial species in the community must be assembled de novo. However, in order to obtain the best metagenomic assemblies, it is important to choose the proper assembler. Due to the rapidly evolving nature of metagenomics, new assemblers are constantly created, and the field has not yet agreed on a standardized process. Furthermore, the truth sets used to compare these methods are either too simple (computationally derived diverse communities) or complex (microbial communities of unknown composition), yielding results that are hard to interpret. In this analysis, we interrogate the strengths and weaknesses of five popular assemblers through the use of defined biological samples of known genomic composition and abundance. We assessed the performance of each assembler on their ability to reassemble genomes, call taxonomic abundances, and recreate open reading frames (ORFs). Results We tested five metagenomic assemblers: Omega, metaSPAdes, IDBA-UD, metaVelvet and MEGAHIT on known and synthetic metagenomic data sets. MetaSPAdes excelled in diverse sets, IDBA-UD performed well all around, metaVelvet had high accuracy in high abundance organisms, and MEGAHIT was able to accurately differentiate similar organisms within a community. At the ORF level, metaSPAdes and MEGAHIT had the least number of missing ORFs within diverse and similar communities respectively. Conclusions Depending on the metagenomics question asked, the correct assembler for the task at hand will differ. It is important to choose the appropriate assembler, and thus clearly define the biological problem of an experiment, as different assemblers will give different answers to the same question.

Phages play critical roles in the survival and pathogenicity of their hosts, via lysogenic conversion factors, and in nutrient redistribution, via cell lysis. Analyses of phage- and viral-encoded genes in environmental samples provide insights into the physiological impact of viruses on microbial communities and human health. However, phage ORFs are extremely diverse of which over 70% of them are dissimilar to any genes with annotated functions in GenBank. Better identification of viruses would also aid in better detection and diagnosis of disease, in vaccine development, and generally in better understanding the physiological potential of any environment. In contrast to enzymes, viral structural protein function can be much more challenging to detect from sequence data because of low sequence conservation, few known conserved catalytic sites or sequence domains, and relatively limited experimental data. We have designed a method of predicting phage structural protein sequences that uses Artificial Neural Networks (ANNs). First, we trained ANNs to classify viral structural proteins using amino acid frequency; these correctly classify a large fraction of test cases with a high degree of specificity and sensitivity. Subsequently, we added estimates of protein isoelectric points as a feature to ANNs that classify specialized families of proteins, namely major capsid and tail proteins. As expected, these more specialized ANNs are more accurate than the structural ANNs. To experimentally validate the ANN predictions, several ORFs with no significant similarities to known sequences that are ANN-predicted structural proteins were examined by transmission electron microscopy. Some of these self-assembled into structures strongly resembling virion structures. Thus, our ANNs are new tools for identifying phage and potential prophage structural proteins that are difficult or impossible to detect by other bioinformatic analysis. The networks will be valuable when sequence is available but in vitro propagation of the phage may not be practical or possible.