Kaiju is a program for the taxonomic classification of metagenomic high-throughput sequencing reads. Each read is directly assigned to a taxon within the NCBI taxonomy by comparing it to a reference database containing microbial and viral protein sequences.
By default, Kaiju uses either the available complete genomes from NCBI RefSeq or the microbial subset of the non-redundant protein database nr used by NCBI BLAST, optionally also including fungi and microbial eukaryotes.
Kaiju translates reads into amino acid sequences, which are then searched in the database using a modified backward search on a memory-efficient implementation of the Burrows-Wheeler transform, which finds maximum exact matches (MEMs), optionally allowing mismatches in the protein alignment. The search can process up to millions of reads per minute using, for example, only 10 GB RAM with a protein database comprising 4821 microbial genomes. Kaiju can also be used for querying any other protein database without taxonomic classification, using either protein or nucleotide queries.
Kaiju is described in Menzel, P. et al. (2016) Fast and sensitive taxonomic classification for metagenomics with Kaiju. Nat. Commun. 7:11257 (open access).
