Sequence Alignment

BackgroundTaxonomic classification of marker-gene sequences is an important step in microbiome analysis.ResultsWe present q2-feature-classifier ( https://github.com/qiime2/q2-feature-classifier ), a QIIME 2 plugin containing several novel machine-learning and alignment-based methods for taxonomy classification. We evaluated and optimized several commonly used classification methods implemented in QIIME 1 (RDP, BLAST, UCLUST, and SortMeRNA) and several new methods implemented in QIIME 2 (a scikit-learn naive Bayes machine-learning classifier, and alignment-based taxonomy consensus methods based on VSEARCH, and BLAST+) for classification of bacterial 16S rRNA and fungal ITS marker-gene amplicon sequence data. The naive-Bayes, BLAST+-based, and VSEARCH-based classifiers implemented in QIIME 2 meet or exceed the species-level accuracy of other commonly used methods designed for classification of marker gene sequences that were evaluated in this work. These evaluations, based on 19 mock communities and error-free sequence simulations, including classification of simulated "novel" marker-gene sequences, are available in our extensible benchmarking framework, tax-credit ( https://github.com/caporaso-lab/tax-credit-data ).ConclusionsOur results illustrate the importance of parameter tuning for optimizing classifier performance, and we make recommendations regarding parameter choices for these classifiers under a range of standard operating conditions. q2-feature-classifier and tax-credit are both free, open-source, BSD-licensed packages available on GitHub.

The MUMmer system and the genome sequence aligner nucmer included within it are among the most widely used alignment packages in genomics. Since the last major release of MUMmer version 3 in 2004, it has been applied to many types of problems including aligning whole genome sequences, aligning reads to a reference genome, and comparing different assemblies of the same genome. Despite its broad utility, MUMmer3 has limitations that can make it difficult to use for large genomes and for the very large sequence data sets that are common today. In this paper we describe MUMmer4, a substantially improved version of MUMmer that addresses genome size constraints by changing the 32-bit suffix tree data structure at the core of MUMmer to a 48-bit suffix array, and that offers improved speed through parallel processing of input query sequences. With a theoretical limit on the input size of 141Tbp, MUMmer4 can now work with input sequences of any biologically realistic length. We show that as a result of these enhancements, the nucmer program in MUMmer4 is easily able to handle alignments of large genomes; we illustrate this with an alignment of the human and chimpanzee genomes, which allows us to compute that the two species are 98% identical across 96% of their length. With the enhancements described here, MUMmer4 can also be used to efficiently align reads to reference genomes, although it is less sensitive and accurate than the dedicated read aligners. The nucmer aligner in MUMmer4 can now be called from scripting languages such as Perl, Python and Ruby. These improvements make MUMer4 one the most versatile genome alignment packages available.

Orthology assignment is ideally suited for functional inference. However, because predicting orthology is computationally intensive at large scale, and most pipelines are relatively inaccessible (e.g., new assignments only available through database updates), less precise homology-based functional transfer is still the default for (meta-)genome annotation. We, therefore, developed eggNOG-mapper, a tool for functional annotation of large sets of sequences based on fast orthology assignments using precomputed clusters and phylogenies from the eggNOG database. To validate our method, we benchmarked Gene Ontology (GO) predictions against two widely used homology-based approaches: BLAST and InterProScan. Orthology filters applied to BLAST results reduced the rate of false positive assignments by 11%, and increased the ratio of experimentally validated terms recovered over all terms assigned per protein by 15%. Compared with InterProScan, eggNOG-mapper achieved similar proteome coverage and precision while predicting, on average, 41 more terms per protein and increasing the rate of experimentally validated terms recovered over total term assignments per protein by 35%. EggNOG-mapper predictions scored within the top-5 methods in the three GO categories using the CAFA2 NK-partial benchmark. Finally, we evaluated eggNOG-mapper for functional annotation of metagenomics data, yielding better performance than interProScan. eggNOG-mapper runs ∼15× faster than BLAST and at least 2.5× faster than InterProScan. The tool is available standalone and as an online service at http://eggnog-mapper.embl.de.

Pangenome indexes reduce reference bias in sequencing data analysis. However, bias can be reduced further by using a personalized reference, for example, a diploid human reference constructed to match a donor individual's alleles. Here, we present a new impute-first alignment framework that combines elements of genotype imputation and read alignment. We first genotype the individual using a subsample of the input reads. Using a reference panel and an efficient imputation algorithm, we impute a personalized diploid reference. Finally, we index the personalized reference and apply a read aligner (either linear or graph) to align the full read set to the personalized reference. On the HG002 sample, this framework achieves a higher variant-calling F1-score (99.77%) compared with the traditional linear aligner (99.62%), graph pangenome aligner (99.72%), and graph personalized-pangenome aligner (99.75%), with substantial reduction in the number of errors (38.73% vs. a linear aligner, 14.97% vs. a graph aligner, and 6.05% vs. a personalized graph). An imputed reference can have comparable efficiency to a pangenome reference, making it an overall advantageous choice for whole-genome DNA sequencing experiments. Advantages of our impute-first approach include that (1) it fully considers linkage disequilibrium and produces a phased diploid reference as an output; (2) it produces accurate personalized references even from low-coverage data; and (3) it is compatible with both graph and linear reference representations, achieving its highest variant-calling F1 accuracy using a standard linear aligner (BWA-MEM).