Similarity search is one of the most fundamental computations that are regularly performed on ever-increasing protein datasets. Scalability is of paramount importance for uncovering novel phenomena that occur at very large scales. We unleash the power of over 20,000 GPUs on the Summit system to perform all-vs-all protein similarity search on one of the largest publicly available datasets with 405 million proteins, in less than 3.5 hours, cutting the time-to-solution for many use cases from weeks. The variability of protein sequence lengths, as well as the sparsity of the space of pairwise comparisons, make this a challenging problem in distributed memory. Due to the need to construct and maintain a data structure holding indices to all other sequences, this application has a huge memory footprint that makes it hard to scale the problem sizes. We overcome this memory limitation by innovative matrix-based blocking techniques, without introducing additional load imbalance.

MOTIVATION: Two key steps in the analysis of uncultured viruses recovered from metagenomes are the taxonomic classification of the viral sequences and the identification of putative host(s). Both steps rely mainly on the assignment of viral proteins to orthologs in cultivated viruses. Viral Protein Families (VPFs) can be used for the robust identification of new viral sequences in large metagenomics datasets. Despite the importance of VPF information for viral discovery, VPFs have not yet been explored for determining viral taxonomy and host targets. RESULTS: In this work, we classified the set of VPFs from the IMG/VR database and developed VPF-Class. VPF-Class is a tool that automates the taxonomic classification and host prediction of viral contigs based on the assignment of their proteins to a set of classified VPFs. Applying VPF-Class on 731K uncultivated virus contigs from the IMG/VR database, we were able to classify 363K contigs at the genus level and predict the host of over 461K contigs. In the RefSeq database, VPF-class reported an accuracy of nearly 100% to classify dsDNA, ssDNA and retroviruses, at the genus level, considering a membership ratio and a confidence score of 0.2. The accuracy in host prediction was 86.4%, also at the genus level, considering a membership ratio of 0.3 and a confidence score of 0.5. And, in the prophages dataset, the accuracy in host prediction was 86% considering a membership ratio of 0.6 and a confidence score of 0.8. Moreover, from the Global Ocean Virome dataset, over 817K viral contigs out of 1 million were classified. AVAILABILITY AND IMPLEMENTATION: The implementation of VPF-Class can be downloaded from https://github.com/biocom-uib/vpf-tools. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

We review the establishment of computational biology in Greece and Cyprus from its inception to date and issue recommendations for future development.We compare output to other countries of similar geography, economy, and size-based on publication counts recorded in the literature-and predict future growth based on those counts as well as national priority areas.Our analysis may be pertinent to wider national or regional communities with challenges and opportunities emerging from the rapid expansion of the field and related industries.Our recommendations suggest a 2-fold growth margin for the 2 countries, as a realistic expectation for further expansion of the field and the development of a credible roadmap of national priorities, both in terms of research and infrastructure funding.

BACKGROUND: Computing of sequence similarity results is becoming a limiting factor in metagenome analysis. Sequence similarity search results encoded in an open, exchangeable format have the potential to limit the needs for computational reanalysis of these data sets. A prerequisite for sharing of similarity results is a common reference. DESCRIPTION: We introduce a mechanism for automatically maintaining a comprehensive, non-redundant protein database and for creating a quarterly release of this resource. In addition, we present tools for translating similarity searches into many annotation namespaces, e.g. KEGG or NCBI's GenBank. CONCLUSIONS: The data and tools we present allow the creation of multiple result sets using a single computation, permitting computational results to be shared between groups for large sequence data sets.

MOTIVATION: Shotgun sequencing generates large numbers of short DNA reads from either an isolated organism or, in the case of metagenomics projects, from the aggregate genome of a microbial community. These reads are then assembled based on overlapping sequences into larger, contiguous sequences (contigs). The feasibility of assembly and the coverage achieved (reads per nucleotide or distinct sequence of nucleotides) depend on several factors: the number of reads sequenced, the read length and the relative abundances of their source genomes in the microbial community. A low coverage suggests that most of the genomic DNA in the sample has not been sequenced, but it is often difficult to estimate either the extent of the uncaptured diversity or the amount of additional sequencing that would be most efficacious. In this work, we regard a metagenome as a population of DNA fragments (bins), each of which may be covered by one or more reads. We employ a gamma distribution to model this bin population due to its flexibility and ease of use. When a gamma approximation can be found that adequately fits the data, we may estimate the number of bins that were not sequenced and that could potentially be revealed by additional sequencing. We evaluated the performance of this model using simulated metagenomes and demonstrate its applicability on three recent metagenomic datasets. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: A rapidly increasing number of microbial genomes are sequenced by organizations worldwide and are eventually included into various public genome data resources. The quality of the annotations depends largely on the original dataset providers, with erroneous or incomplete annotations often carried over into the public resources and difficult to correct. RESULTS: We have developed an Expert Review (ER) version of the Integrated Microbial Genomes (IMG) system, with the goal of supporting systematic and efficient revision of microbial genome annotations. IMG ER provides tools for the review and curation of annotations of both new and publicly available microbial genomes within IMG's rich integrated genome framework. New genome datasets are included into IMG ER prior to their public release either with their native annotations or with annotations generated by IMG ER's annotation pipeline. IMG ER tools allow addressing annotation problems detected with IMG's comparative analysis tools, such as genes missed by gene prediction pipelines or genes without an associated function. Over the past year, IMG ER was used for improving the annotations of about 150 microbial genomes.

BACKGROUND: Clustered Regularly Interspaced Palindromic Repeats (CRISPRs) are a novel type of direct repeat found in a wide range of bacteria and archaea. CRISPRs are beginning to attract attention because of their proposed mechanism; that is, defending their hosts against invading extrachromosomal elements such as viruses. Existing repeat detection tools do a poor job of identifying CRISPRs due to the presence of unique spacer sequences separating the repeats. In this study, a new tool, CRT, is introduced that rapidly and accurately identifies CRISPRs in large DNA strings, such as genomes and metagenomes. RESULTS: CRT was compared to CRISPR detection tools, Patscan and Pilercr. In terms of correctness, CRT was shown to be very reliable, demonstrating significant improvements over Patscan for measures precision, recall and quality. When compared to Pilercr, CRT showed improved performance for recall and quality. In terms of speed, CRT proved to be a huge improvement over Patscan. Both CRT and Pilercr were comparable in speed, however CRT was faster for genomes containing large numbers of repeats. CONCLUSION: In this paper a new tool was introduced for the automatic detection of CRISPR elements. This tool, CRT, showed some important improvements over current techniques for CRISPR identification. CRT's approach to detecting repetitive sequences is straightforward. It uses a simple sequential scan of a DNA sequence and detects repeats directly without any major conversion or preprocessing of the input. This leads to a program that is easy to describe and understand; yet it is very accurate, fast and memory efficient, being O(n) in space and O(nm/l) in time.

UNLABELLED: GOLD (Genomes On Line Database) is a World Wide Web resource for comprehensive access to information regarding complete and ongoing genome projects around the world. AVAILABILITY: GOLD is based at the University of Illinois at Urbana-Champaign and is available at http://geta.life.uiuc.edu/ approximately nikos/genomes. html. It is also mirrored at the European Bioinformatics Institute at http://www.ebi.ac.uk/research/cgg/genomes.html. CONTACT: [email protected]