Improved metagenomic analysis with Kraken 2

Abstract

Although Kraken's k-mer-based approach provides fast taxonomic classification of metagenomic sequence data, its large memory requirements can be limiting for some applications. Kraken 2 improves upon Kraken 1 by reducing memory usage by 85%, allowing greater amounts of reference genomic data to be used, while maintaining high accuracy and increasing speed five-fold. Kraken 2 also introduces a translated search mode, providing increased sensitivity in viral metagenomics analysis.Assigning taxonomic labels to sequencing reads is an important part of many computational genomics pipelines for metagenomics projects. Recent years have seen several approaches to accomplish this task in a time-efficient manner 1-3 . Kraken 4 used a memory-intensive algorithm that associates short genomic substrings (k-mers) with lowest common ancestor (LCA) taxa. Kraken and related tools like KrakenUniq 5 have proven highly efficient and accurate in other tool comparisons 6,7 . But Kraken's high memory requirements force many researchers to either use a reduced-sensitivity MiniKraken database 8,9 , or to build and use many indexes over subsets of the reference sequences 10,11 . Its memory requirements can easily exceed 100 GB 7 , especially when the reference data includes large eukaryotic genomes 12,13 . Here we introduce Kraken 2, which provides a major reduction in memory usage as well as faster classification, a spaced-seed searching scheme, a translated search mode for matching in amino acid space, and continued compatibility with the Bracken 14 species-level quantification algorithm.Kraken 2 addresses the issue of large memory requirements through two changes to Kraken 1's data structures and algorithms. While Kraken 1 used a sorted list of k-mer/LCA pairs indexed by minimizers 15 , Kraken 2 introduces a probabilistic, compact hash table to map minimizers to LCAs. This table uses one-third of the memory of a standard hash table, at the cost of some specificity and accuracy. Additionally, Kraken 2 only stores minimizers (of length ℓ, ℓ ≤ k) from the reference sequence library in the data structure, whereas Kraken 1's stored all k-mers. Kraken 2's index for a reference database consisting of 9.1 Gbp of genomic sequence uses 10.6 gigabytes of memory at classification time. Kraken 1's index for the same reference set uses 72.4 gigabytes of memory for classification (Figure 1a, Supplementary Table S1). In general, a Kraken 2 database is about 15% as large as a Kraken 1 database over the same references (Supplementary Figure S1).Kraken 2's approach is faster than Kraken 1's because only the distinct minimizers from the query (read) trigger accesses to the hash table. A similar minimizer-based approach has proven useful in accelerating read alignment 16 . Kraken 2 additionally provides a hash-based filtering approach that subsamples the set of minimizer/LCA pairs included in the table, allowing the user to specify a target hash table size; smaller hash tables yield lower memory footprint and higher classification throughput at the expens...

Long-Read Metagenomics to Retrieve High-Quality Metagenome-Assembled Genomes from Canine Feces

“…To obtain the rst taxonomic assignment directly from the raw reads, we processed the data using Kraken2 2.0.8 [17] with the maxikraken2 database (Loman Lab, from March 2019) that includes all the genomes from RefSeq. We visualized Kraken2 reports using Sankey diagrams with pavian 1.0.0 R package [18].…”

Hybridization capture of larch (LarixMill) chloroplast genomes from sedimentary ancient DNA reveals past changes of Siberian forests

“…Reads were classified using kraken2, (v. 2.0.7-beta, Wood, Lu, & Langmead, 2019) with a conservative confidence threshold (--confidence 0.9) against the non-redundant nucleotide database (nt) from NCBI (ftp://ftp.ncbi.nlm.nih.gov/blast/db/FASTA/nt.gz; downloaded in May 2019) and the NCBI taxonomy (retrieved via the kraken2-build command in June 2019). This classification was used for the description and comparison of the shotgun as well as the capture dataset.…”

Revealing Oral Microbiota Composition and Functionality Associated with Heavy Cigarette Smoking.

“…At the end of quality control, the median number of quality-ltered reads per samples was 5062550. The remaining reads were taxonomically classi ed using Kraken2 with the MiniKraken2_v1 database [23] and with the Genome Taxonomy Database (v. 89).…”

“…We analyzed buccal swab samples from 105 participants for taxonomic composition, differential abundance and functional pro ling of their oral microbiota. The subjects' characteristics in this study such as age, gender, body mass index (BMI), ethnicity, and medical history among others have been provided (Table 1) Table 1 Demographics of Study Cohort First, we evaluated the taxonomic composition generated from high-quality reads and classi ed them using the MiniKraken2_v1 database [23] as the reference database for bacteria. We aggregated taxa abundances into genera and plotted the relative abundances of the most abundant ones (Fig.…”