blast_getLCA

The blast_getLCA algorithm parses a blast file, typically generated from amplicon data, and assigns each read to the lowest common ancestor of the best hit(s) in the blast file.

Getting started

Download and Install

1) Download and unpack the blast_getLCA master

wget https://github.com/frederikseersholm/blast_getLCA/archive/master.tar.gz
tar xpvf master.tar.gz
cd blast_getLCA-master

2) Run the script prepare_getLCA_script.sh

This script downloads the NCBI taxonomy files nodes.dmp and names.dmp to the folder 'taxdump', and adds the paths to the files in the python script (expected run time ~1 min).

bash prepare_getLCA_script.sh

3) Run the script on the supplied test blast-file, modified for the blast_getLCA algorithm

Please make sure that you are using python 2.7 for this step. This can be specified by calling 'python2.7' instead of just 'python'. Expected run time ~1 min.

python blast_getLCA.py test.taxid.blast

Generate a suitable blast file for analysis

1) Blast fasta file with custom output format 6

In order for the blast_getLCA script to work, fasta files should be blasted with the following output format:

• outfmt "6 qseqid sacc sseqid pident qlen length mismatch gapopen gaps evalue bitscore nident"

For example as follows:

blastn -outfmt "6 qseqid sacc sseqid pident qlen length mismatch gapopen gaps evalue bitscore nident" -max_target_seqs 100 -reward 1 -db $DB -query ${FILENAME}.fasta -out ${FILENAME}.blast

2) Run the script prepare_add_taxid2blast_script.sh (only required the first time the script is run)

This script downloads NCBI lists with gi-numbers and their associated taxIDs, which is used when adding taxIDs to blast files. After downloading, the script adds the paths of the NCBI files to the python script add_taxid2blast.py. This step might take several hours, but is only required once.

bash prepare_add_taxid2blast_script.sh

3) Add taxid information to blast file by running the script add_taxid2blast.py

This script parses the blast file and appends the taxID to the subject sequence identifier (column 2 in blast file). Expected run time ~15 min.

python add_taxid2blast.py test2.blast

Output

The output is a tsv file, with one row for each read, and the following columns:

• qseqid The identifier of the query sequence
• LCA The lowest common ancestor of the read, with info on genus, family, order and class of the assigned taxon
• rank Rank of the lowest common ancestor
• TaxIDs TaxIDs of the the best matching reference sequences in the database used to infer the lowest common ancestor, separated by colon
• Stats Basic statistics of the assignment (separated by underscore):
  • Tothits = total number of alignments for the read
  • accepted-hits = total number of best (equally good) hits for the read
  • Min-Nm = Edit distance for the best hit(s)
  • IDp = Identity percentage for the best hit(s)
• qseqlength Query sequence length
• IDp Identity percentage for the best hit(s)
• gapmm Gaps, mismatches and length of the best alignment(s), separated by underscore
• drop Information on whether the LCA have been dropped to genus or family level based on a low identity percent to the best hit(s)

Options

Options:
  -h, --help            show this help message and exit
  -i WRONGTAX, --ignoretaxid=WRONGTAX
                        csv file of taxids to ignore, first column should
                        contain taxids
  -t THRESHOLD, --threshold=THRESHOLD
                        Ignores reads where the best alignment has less than a
                        certain percentage identity to the reference
                        [default=95]
  -l LIMITS, --limits=LIMITS
                        List of identity thresholds which will inform the
                        algorithm to drop lowest common ancestor to genus
                        (default: 95-98%), family level (default: 90-95%) or
                        order level (default: below 90%). Separated by dash
                        '-' [default=98-95-90]

System Requirements

This software has been tested on the following operating systems:

• macOS Mojave version 10.14.6
• CentOS Linux 7 (Core)
• SUSE Linux Enterprise Server 12 SP3

Software dependencies:

• python2.7