In the battle of the host against lentiviral pathogenesis, the immune response is crucial. However, several questions remain unanswered about the interaction with different viruses and their influence on disease progression. The simian immunodeficiency virus (SIV) infecting nonhuman primates (NHP) is widely used as a model for the study of the human immunodeficiency virus (HIV) both because they are evolutionarily linked and because they share physiological and anatomical similarities that are largely explored to understand the disease progression. The HIHISIV database was developed to support researchers to integrate and evaluate the large number of transcriptional data associated with the presence/absence of the pathogen (SIV or HIV) and the host response (NHP and human). The datasets are composed of microarray and RNA-Seq gene expression data that were selected, curated, analyzed, enriched, and stored in a relational database. Six query templates comprise the main data analysis functions and the resulting information can be downloaded. The HIHISIV database, available at https://hihisiv.github.io , provides accurate resources for browsing and visualizing results and for more robust analyses of pre-existing data in transcriptome repositories.

The COVID-19 pandemic has been characterised by sequential variant-specific waves shaped by viral, individual human and population factors. SARS-CoV-2 variants are defined by their unique combinations of mutations and there has been a clear adaptation to more efficient human infection since the emergence of this new human coronavirus in late 2019. Here, we use machine learning models to identify shared signatures, i.e., common underlying mutational processes and link these to the subset of mutations that define the variants of concern (VOCs). First, we examined the global SARS-CoV-2 genomes and associated metadata to determine how viral properties and public health measures have influenced the magnitude of waves, as measured by the number of infection cases, in different geographic locations using regression models. This analysis showed that, as expected, both public health measures and virus properties were associated with the waves of regional SARS-CoV-2 reported infection numbers and this impact varies geographically. We attribute this to intrinsic differences such as vaccine coverage, testing and sequencing capacity and the effectiveness of government stringency. To assess underlying evolutionary change, we used non-negative matrix factorisation and observed three distinct mutational signatures, unique in their substitution patterns and exposures from the SARS-CoV-2 genomes. Signatures 1, 2 and 3 were biased to C→T, T→C/A→G and G→T point mutations. We hypothesise assignments of these mutational signatures to the host antiviral molecules APOBEC, ADAR and ROS respectively. We observe a shift amidst the pandemic in relative mutational signature activity from predominantly Signature 1 changes to an increasingly high proportion of changes consistent with Signature 2. This could represent changes in how the virus and the host immune response interact and indicates how SARS-CoV-2 may continue to generate variation in the future. Linkage of the detected mutational signatures to the VOC-defining amino acids substitutions indicates the majority of SARS-CoV-2's evolutionary capacity is likely to be associated with the action of host antiviral molecules rather than virus replication errors.

Predicting virus-host associations is essential to determine the specific host species that viruses interact with, and discover if new viruses infect humans and animals. Currently, the host of the majority of viruses is unknown, particularly in microbiomes. To address this challenge, we introduce EvoMIL, a deep learning method that predicts the host species for viruses from viral sequences only. It also identifies important viral proteins that significantly contribute to host prediction. The method combines a pre-trained large protein language model (ESM) and attention-based multiple instance learning to allow protein-orientated predictions. Our results show that protein embeddings capture stronger predictive signals than sequence composition features, including amino acids, physiochemical properties, and DNA k-mers. In multi-host prediction tasks, EvoMIL achieves median F1 score improvements of 10.8%, 16.2%, and 4.9% in prokaryotic hosts, and 1.7%, 6.6% and 11.5% in eukaryotic hosts. EvoMIL binary classifiers achieve impressive AUC over 0.95 for all prokaryotic hosts and range from roughly 0.8 to 0.9 for eukaryotic hosts. Furthermore, EvoMIL identifies important proteins in the prediction task, capturing key functions involved in virus-host specificity.

Human immunodeficiency virus type 1 (HIV-1) continues to be a major cause of disease and premature death. As with all viruses, HIV-1 exploits a host cell to replicate. Improving our understanding of the molecular interactions between virus and human host proteins is crucial for a mechanistic understanding of virus biology, infection and host antiviral activities. This knowledge will potentially permit the identification of host molecules for targeting by drugs with antiviral properties. Here, we propose a data-driven approach for the analysis and prediction of the HIV-1 interacting proteins (VIPs) with a focus on the directionality of the interaction: host-dependency versus antiviral factors. Using support vector machine learning models and features encompassing genetic, proteomic and network properties, our results reveal some significant differences between the VIPs and non-HIV-1 interacting human proteins (non-VIPs). As assessed by comparison with the HIV-1 infection pathway data in the Reactome database (sensitivity > 90%, threshold = 0.5), we demonstrate these models have good generalization properties. We find that the 'direction' of the HIV-1-host molecular interactions is also predictable due to different characteristics of 'forward'/pro-viral versus 'backward'/pro-host proteins. Additionally, we infer the previously unknown direction of the interactions between HIV-1 and 1351 human host proteins. A web server for performing predictions is available at http://hivpre.cvr.gla.ac.uk/.

SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2) is a novel virus of the family Coronaviridae. The virus causes the infectious disease COVID-19. The biology of coronaviruses has been studied for many years. However, bioinformatics tools designed explicitly for SARS-CoV-2 have only recently been developed as a rapid reaction to the need for fast detection, understanding and treatment of COVID-19. To control the ongoing COVID-19 pandemic, it is of utmost importance to get insight into the evolution and pathogenesis of the virus. In this review, we cover bioinformatics workflows and tools for the routine detection of SARS-CoV-2 infection, the reliable analysis of sequencing data, the tracking of the COVID-19 pandemic and evaluation of containment measures, the study of coronavirus evolution, the discovery of potential drug targets and development of therapeutic strategies. For each tool, we briefly describe its use case and how it advances research specifically for SARS-CoV-2. All tools are free to use and available online, either through web applications or public code repositories. Contact:[email protected].

The rise in metagenomics has led to an exponential growth in virus discovery. However, the majority of these new virus sequences have no assigned host. Current machine learning approaches to predicting virus host interactions have a tendency to focus on nucleotide features, ignoring other representations of genomic information. Here we investigate the predictive potential of features generated from four different 'levels' of viral genome representation: nucleotide, amino acid, amino acid properties and protein domains. This more fully exploits the biological information present in the virus genomes. Over a hundred and eighty binary datasets for infecting versus non-infecting viruses at all taxonomic ranks of both eukaryote and prokaryote hosts were compiled. The viral genomes were converted into the four different levels of genome representation and twenty feature sets were generated by extracting k-mer compositions and predicted protein domains. We trained and tested Support Vector Machine, SVM, classifiers to compare the predictive capacity of each of these feature sets for each dataset. Our results show that all levels of genome representation are consistently predictive of host taxonomy and that prediction k-mer composition improves with increasing k-mer length for all k-mer based features. Using a phylogenetically aware holdout method, we demonstrate that the predictive feature sets contain signals reflecting both the evolutionary relationship between the viruses infecting related hosts, and host-mimicry. Our results demonstrate that incorporating a range of complementary features, generated purely from virus genome sequences, leads to improved accuracy for a range of virus host prediction tasks enabling computational assignment of host taxonomic information.

BACKGROUND: The consolidation of pathway databases, such as KEGG, Reactome and ConsensusPathDB, has generated widespread biological interest, however the issue of pathway redundancy impedes the use of these consolidated datasets. Attempts to reduce this redundancy have focused on visualizing pathway overlap or merging pathways, but the resulting pathways may be of heterogeneous sizes and cover multiple biological functions. Efforts have also been made to deal with redundancy in pathway data by consolidating enriched pathways into a number of clusters or concepts. We present an alternative approach, which generates pathway subsets capable of covering all of genes presented within either pathway databases or enrichment results, generating substantial reductions in redundancy. RESULTS: We propose a method that uses set cover to reduce pathway redundancy, without merging pathways. The proposed approach considers three objectives: removal of pathway redundancy, controlling pathway size and coverage of the gene set. By applying set cover to the ConsensusPathDB dataset we were able to produce a reduced set of pathways, representing 100% of the genes in the original data set with 74% less redundancy, or 95% of the genes with 88% less redundancy. We also developed an algorithm to simplify enrichment data and applied it to a set of enriched osteoarthritis pathways, revealing that within the top ten pathways, five were redundant subsets of more enriched pathways. Applying set cover to the enrichment results removed these redundant pathways allowing more informative pathways to take their place. CONCLUSION: Our method provides an alternative approach for handling pathway redundancy, while ensuring that the pathways are of homogeneous size and gene coverage is maximised. Pathways are not altered from their original form, allowing biological knowledge regarding the data set to be directly applicable. We demonstrate the ability of the algorithms to prioritise redundancy reduction, pathway size control or gene set coverage. The application of set cover to pathway enrichment results produces an optimised summary of the pathways that best represent the differentially regulated gene set.

Shotgun sequencing has facilitated the analysis of complex microbial communities. Recently we have shown how local binary patterns (LBP) from image processing can be used to analyse the sequenced samples. LBP codes represent the data in a sparse high dimensional space. To improve the performance of our pipeline, marginalised stacked autoencoders are used here to learn frequent LBP codes and map the high dimensional space to a lower dimension dense space. We demonstrate its performance using both low and high complexity simulated metagenomic data and compare the performance of our method with several existing techniques including principal component analysis (PCA) in the dimension reduction step and fc-mer frequency in feature extraction step.

Sequencing technologies allow for an in-depth analysis of biological species but the size of the generated datasets introduce a number of analytical challenges. Recently, we demonstrated the application of numerical sequence representations and data transformations for the alignment of short reads to a reference genome. Here, we expand out approach for de novo assembly of short reads. Our results demonstrate that highly compressed data can encapsulate the signal sufficiently to accurately assemble reads to big contigs or complete genomes.

BACKGROUND: Many computational tools that detect recombination in viruses are not adapted for the ongoing genomic revolution. A computational tool is needed, that will rapidly scan hundreds/thousands of genomes or sequence fragments and detect candidate recombination events that may later be further analyzed with more sensitive and specialized methods. RESULTS: T-RECs, a Windows based graphical tool, employs pairwise alignment of sliding windows and can perform (i) genotyping, (ii) clustering of new genomes, (iii) detect recent recombination events among different evolutionary lineages, (iv) manual inspection of detected recombination events by similarity plots and (v) annotation of genomic regions. CONCLUSIONS: T-RECs is very effective, as demonstrated by an analysis of 555 Norovirus complete genomes and 2500 sequence fragments, where a recombination hotspot was identified at the ORF1-ORF2 junction.

MOTIVATION: Human immunodeficiency virus type 1 (HIV-1) hijacks host cellular processes to replicate within its host. Through interactions with host proteins, it perturbs and interrupts signaling pathways that alter key cellular functions. Although networks of viral-host interactions have been relatively well characterized, the dynamics of the perturbation process is poorly understood. Dynamic models of infection have the potential to provide insights into the HIV-1 host interaction. RESULTS: We employed a logical signal flow network to model the dynamic interactions between HIV-1 proteins and key human signal transduction pathways necessary for activation of CD4+ T lymphocytes. We integrated viral-host interaction and host signal transduction data into a dynamic logical model comprised of 137 nodes (16 HIV-1 and 121 human proteins) and 336 interactions collected from the HIV-1 Human Interaction Database. The model reproduced expected patterns of T-cell activation, co-stimulation and co-inhibition. After simulations, we identified 26 host cell factors, including MAPK1&3, Ikkb-Ikky-Ikka and PKA, which contribute to the net activation or inhibition of viral proteins. Through in silico knockouts, the model identified a further nine host cell factors, including members of the PI3K signalling pathway that are essential to viral replication. Simulation results intersected with the findings of three siRNA gene knockout studies and identified potential drug targets. Our results demonstrate how viral infection causes the cell to lose control of its signalling system. Logical Boolean modelling therefore provides a useful approach for analysing the dynamics of host-viral interactions with potential applications for drug discovery. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: As a natural consequence of being a computer-based discipline, bioinformatics has a strong focus on database and software development, but the volume and variety of resources are growing at unprecedented rates. An audit of database and software usage patterns could help provide an overview of developments in bioinformatics and community common practice, and comparing the links between resources through time could demonstrate both the persistence of existing software and the emergence of new tools. RESULTS: We study the connections between bioinformatics resources and construct networks of database and software usage patterns, based on resource co-occurrence, that correspond to snapshots of common practice in the bioinformatics community. We apply our approach to pairings of phylogenetics software reported in the literature and argue that these could provide a stepping stone into the identification of scientific best practice. AVAILABILITY AND IMPLEMENTATION: The extracted resource data, the scripts used for network generation and the resulting networks are available at http://bionerds.sourceforge.net/networks/.

BACKGROUND: Biology-focused databases and software define bioinformatics and their use is central to computational biology. In such a complex and dynamic field, it is of interest to understand what resources are available, which are used, how much they are used, and for what they are used. While scholarly literature surveys can provide some insights, large-scale computer-based approaches to identify mentions of bioinformatics databases and software from primary literature would automate systematic cataloguing, facilitate the monitoring of usage, and provide the foundations for the recovery of computational methods for analysing biological data, with the long-term aim of identifying best/common practice in different areas of biology. RESULTS: We have developed bioNerDS, a named entity recogniser for the recovery of bioinformatics databases and software from primary literature. We identify such entities with an F-measure ranging from 63% to 91% at the mention level and 63-78% at the document level, depending on corpus. Not attaining a higher F-measure is mostly due to high ambiguity in resource naming, which is compounded by the on-going introduction of new resources. To demonstrate the software, we applied bioNerDS to full-text articles from BMC Bioinformatics and Genome Biology. General mention patterns reflect the remit of these journals, highlighting BMC Bioinformatics's emphasis on new tools and Genome Biology's greater emphasis on data analysis. The data also illustrates some shifts in resource usage: for example, the past decade has seen R and the Gene Ontology join BLAST and GenBank as the main components in bioinformatics processing. ABSTRACT: Conclusions We demonstrate the feasibility of automatically identifying resource names on a large-scale from the scientific literature and show that the generated data can be used for exploration of bioinformatics database and software usage. For example, our results help to investigate the rate of change in resource usage and corroborate the suspicion that a vast majority of resources are created, but rarely (if ever) used thereafter. bioNerDS is available at http://bionerds.sourceforge.net/.

BACKGROUND: Next generation sequencing provides detailed insight into the variation present within viral populations, introducing the possibility of treatment strategies that are both reactive and predictive. Current software tools, however, need to be scaled up to accommodate for high-depth viral data sets, which are often temporally or spatially linked. In addition, due to the development of novel sequencing platforms and chemistries, each with implicit strengths and weaknesses, it will be helpful for researchers to be able to routinely compare and combine data sets from different platforms/chemistries. In particular, error associated with a specific sequencing process must be quantified so that true biological variation may be identified. RESULTS: Segminator II was developed to allow for the efficient comparison of data sets derived from different sources. We demonstrate its usage by comparing large data sets from 12 influenza H1N1 samples sequenced on both the 454 Life Sciences and Illumina platforms, permitting quantification of platform error. For mismatches median error rates at 0.10 and 0.12%, respectively, suggested that both platforms performed similarly. For insertions and deletions median error rates within the 454 data (at 0.3 and 0.2%, respectively) were significantly higher than those within the Illumina data (0.004 and 0.006%, respectively). In agreement with previous observations these higher rates were strongly associated with homopolymeric stretches on the 454 platform. Outside of such regions both platforms had similar indel error profiles. Additionally, we apply our software to the identification of low frequency variants. CONCLUSION: We have demonstrated, using Segminator II, that it is possible to distinguish platform specific error from biological variation using data derived from two different platforms. We have used this approach to quantify the amount of error present within the 454 and Illumina platforms in relation to genomic location as well as location on the read. Given that next generation data is increasingly important in the analysis of drug-resistance and vaccine trials, this software will be useful to the pathogen research community. A zip file containing the source code and jar file is freely available for download from http://www.bioinf.manchester.ac.uk/segminator/.

Large-scale parallel pyrosequencing produces unprecedented quantities of sequence data. However, when generated from viral populations current mapping software is inadequate for dealing with the high levels of variation present, resulting in the potential for biased data loss. In order to apply the 454 Life Sciences' pyrosequencing system to the study of viral populations, we have developed software for the processing of highly variable sequence data. Here we demonstrate our software by analyzing two temporally sampled HIV-1 intra-patient datasets from a clinical study of maraviroc. This drug binds the CCR5 coreceptor, thus preventing HIV-1 infection of the cell. The objective is to determine viral tropism (CCR5 versus CXCR4 usage) and track the evolution of minority CXCR4-using variants that may limit the response to a maraviroc-containing treatment regimen. Five time points (two prior to treatment) were available from each patient. We first quantify the effects of divergence on initial read k-mer mapping and demonstrate the importance of utilizing population-specific template sequences in relation to the analysis of next-generation sequence data. Then, in conjunction with coreceptor prediction algorithms that infer HIV tropism, our software was used to quantify the viral population structure pre- and post-treatment. In both cases, low frequency CXCR4-using variants (2.5-15%) were detected prior to treatment. Following phylogenetic inference, these variants were observed to exist as distinct lineages that were maintained through time. Our analysis, thus confirms the role of pre-existing CXCR4-using virus in the emergence of maraviroc-insensitive HIV. The software will have utility for the study of intra-host viral diversity and evolution of other fast evolving viruses, and is available from http://www.bioinf.manchester.ac.uk/segminator/.

Human immunodeficiency virus type 1 (HIV-1) exploits a diverse array of host cell functions in order to replicate. This is mediated through a network of virus-host interactions. A variety of recent studies have catalogued this information. In particular the HIV-1, Human Protein Interaction Database (HHPID) has provided a unique depth of protein interaction detail. However, as a map of HIV-1 infection, the HHPID is problematic, as it contains curation error and redundancy; in addition, it is based on a heterogeneous set of experimental methods. Based on identifying shared patterns of HIV-host interaction, we have developed a novel methodology to delimit the core set of host-cellular functions and their associated perturbation from the HHPID. Initially, using biclustering, we identify 279 significant sets of host proteins that undergo the same types of interaction. The functional cohesiveness of these protein sets was validated using a human protein-protein interaction network, gene ontology annotation and sequence similarity. Next, using a distance measure, we group host protein sets and identify 37 distinct higher-level subsystems. We further demonstrate the biological significance of these subsystems by cross-referencing with global siRNA screens that have been used to detect host factors necessary for HIV-1 replication, and investigate the seemingly small intersect between these data sets. Our results highlight significant host-cell subsystems that are perturbed during the course of HIV-1 infection. Moreover, we characterise the patterns of interaction that contribute to these perturbations. Thus, our work disentangles the complex set of HIV-1-host protein interactions in the HHPID, reconciles these with siRNA screens and provides an accessible and interpretable map of infection.

BACKGROUND: A common method for presenting and studying biological interaction networks is visualization. Software tools can enhance our ability to explore network visualizations and improve our understanding of biological systems, particularly when these tools offer analysis capabilities. However, most published network visualizations are static representations that do not support user interaction. RESULTS: JNets was designed as a network visualization tool that incorporates annotation to explore the underlying features of interaction networks. The software is available as an application and a configurable applet that can provide a flexible and dynamic online interface to many types of network data. As a case study, we use JNets to investigate approved drug targets present within the HIV-1 Human protein interaction network. Our software highlights the intricate influence that HIV-1 has on the host immune response. CONCLUSION: JNets is a software tool that allows interaction networks to be visualized and studied remotely, from within a standard web page. Therefore, using this free software, network data can be presented in an enhanced, interactive format. More information about JNets is available at http://www.manchester.ac.uk/bioinformatics/jnets.

BACKGROUND: The methodologies we use both enable and help define our research. However, as experimental complexity has increased the choice of appropriate methodologies has become an increasingly difficult task. This makes it difficult to keep track of available bioinformatics software, let alone the most suitable protocols in a specific research area. To remedy this we present an approach for capturing methodology from literature in order to identify and, thus, define best practice within a field. RESULTS: Our approach is to implement data extraction techniques on the full-text of scientific articles to obtain the set of experimental protocols used by an entire scientific discipline, molecular phylogenetics. Our methodology for identifying methodologies could in principle be applied to any scientific discipline, whether or not computer-based. We find a number of issues related to the nature of best practice, as opposed to community practice. We find that there is much heterogeneity in the use of molecular phylogenetic methods and software, some of which is related to poor specification of protocols. We also find that phylogenetic practice exhibits field-specific tendencies that have increased through time, despite the generic nature of the available software. We used the practice of highly published and widely collaborative researchers ("expert" researchers) to analyse the influence of authority on community practice. We find expert authors exhibit patterns of practice common to their field and therefore act as useful field-specific practice indicators. CONCLUSION: We have identified a structured community of phylogenetic researchers performing analyses that are customary in their own local community and significantly different from those in other areas. Best practice information can help to bridge such subtle differences by increasing communication of protocols to a wider audience. We propose that the practice of expert authors from the field of evolutionary biology is the closest to contemporary best practice in phylogenetic experimental design. Capturing best practice is, however, a complex task and should also acknowledge the differences between fields such as the specific context of the analysis.

Recombinant HIV-1 genomes contribute significantly to the diversity of variants within the HIV/AIDS pandemic. It is assumed that some of these mosaic genomes may have novel properties that have led to their prevalence, particularly in the case of the circulating recombinant forms (CRFs). In regions of the HIV-1 genome where recombination has a tendency to convey a selective advantage to the virus, we predict that the distribution of breakpoints--the identifiable boundaries that delimit the mosaic structure--will deviate from the underlying null distribution. To test this hypothesis, we generate a probabilistic model of HIV-1 copy-choice recombination and compare the predicted breakpoint distribution to the distribution from the HIV/AIDS pandemic. Across much of the HIV-1 genome, we find that the observed frequencies of inter-subtype recombination are predicted accurately by our model. This observation strongly indicates that in these regions a probabilistic model, dependent on local sequence identity, is sufficient to explain breakpoint locations. In regions where there is a significant over- (either side of the env gene) or under- (short regions within gag, pol, and most of env) representation of breakpoints, we infer natural selection to be influencing the recombination pattern. The paucity of recombination breakpoints within most of the envelope gene indicates that recombinants generated in this region are less likely to be successful. The breakpoints at a higher frequency than predicted by our model are approximately at either side of env, indicating increased selection for these recombinants as a consequence of this region, or at least part of it, having a tendency to be recombined as an entire unit. Our findings thus provide the first clear indication of the existence of a specific portion of the genome that deviates from a probabilistic null model for recombination. This suggests that, despite the wide diversity of recombinant forms seen in the viral population, only a minority of recombination events appear to be of significance to the evolution of HIV-1.

SUMMARY: CTree has been designed for the quantification of clusters within viral phylogenetic tree topologies. Clusters are stored as individual data structures from which statistical data, such as the Subtype Diversity Ratio (SDR), Subtype Diversity Variance (SDV) and pairwise distances can be extracted. This simplifies the quantification of tree topologies in relation to inter- and intra-cluster diversity. Here the novel features incorporated within CTree, including the implementation of a heuristic algorithm for identifying clusters, are outlined along with the more usual features found within general tree viewing software. AVAILABILITY: CTree is available as an executable jar file from: http://www.manchester.ac.uk/bioinformatics/ctree

BACKGROUND: Alternative representations of biochemical networks emphasise different aspects of the data and contribute to the understanding of complex biological systems. In this study we present a variety of automated methods for visualisation of a protein-protein interaction network, using the basic helix-loop-helix (bHLH) family of transcription factors as an example. RESULTS: Network representations that arrange nodes (proteins) according to either continuous or discrete information are investigated, revealing the existence of protein sub-families and the retention of interactions following gene duplication events. Methods of network visualisation in conjunction with a phylogenetic tree are presented, highlighting the evolutionary relationships between proteins, and clarifying the context of network hubs and interaction clusters. Finally, an optimisation technique is used to create a three-dimensional layout of the phylogenetic tree upon which the protein-protein interactions may be projected. CONCLUSION: We show that by incorporating secondary genomic, functional or phylogenetic information into network visualisation, it is possible to move beyond simple layout algorithms based on network topology towards more biologically meaningful representations. These new visualisations can give structure to complex networks and will greatly help in interpreting their evolutionary origins and functional implications. Three open source software packages (InterView, TVi and OptiMage) implementing our methods are available.