Custom databases
VirPipe employs Kraken2 for taxonomic classification of Illumina reads.
Kraken2 database can be downloaded at Kraken2 Refseq index page or built as explained at official kraken2 github.
VirPipe employs Centrifuge for taxonomic classification of Nanopore reads.
Some prebuilt centrifuge databases can be downloaded at the program's homepage. But no database only for viral sequences is provided and should be built on the user's end. We have built a database only with refseq viral sequences and shared it via Zenodo.
VirPipe runs BLAST as a post-assembly analysis.
As NCBI uploads the prebuilt BLAST databases to its ftp server, you could download whichever database you want to deploy.
Instead, you could build new BLAST database with your own sequences. In this case, you need to install blast+ locally. Follow the instructions below:
- Install BLAST+ (if you have not done before)
conda create -n blast blast && conda activate blast- Download accession2taxid from NCBI and decompress the file
# Download can take several minutes
wget ftp://ftp.ncbi.nih.gov/pub/taxonomy/accession2taxid/nucl_gb.accession2taxid.gz && tar -xzvf nucl_gb.accession2taxid.gz- Parse accession2taxid to make taxidmap
sed '1d' nucl_gb.accession2taxid | awk '{print $2" "$3}' > taxidmapfile # can take several hours- Collect sequences to be contained in database to one fasta file
cat virus_1.fasta virus_2.fasta > viruses.fasta- Set database name.
blast_dbname=viruses
- Build BLAST database
makeblastdb -in viruses.fasta -parse_seqids -taxid_map taxidmapfile -dbtype nucl -out $blast_dbnameVirPipe annotates BLAST results with taxonomy information. This is possible by utilizing taxnomizr package in R. A database that associates accessions with taxonomies is needed in the process, and you can download it on Zenodo repository.