This is a standalone classifier for marker gene sequences, like those produced in microbiome studies, that form a .fasta file that outputs a taxonomy table with Kingdom, Phylum, Class, Order, Family, Genus, Species and confidence scores at each rank. The classification is made using BLAST against the whole NCBI nt (nucleotide) sequence database that must be available locally.
blasTAX calculates the score at each rank and uses the same principles adopted by QIIME2 for blast classified. The hits are filtered by those which pass thresholds for percent ID (percent identity) and e-value. After that, the confidence of each rank is the percent of hits which agree with the most common taxon for that rank. For example, a confidence of 0.8 for Tuber means that 4 of 5 hits passing the filters have Tuber as the genus.
To start, clone this repository using: git clone git@github.com:Gian77/blasTAX.git
Once cloned, download the whole NCBI nt database. Code to download the NCBI nt is provided in the NCBI_DB directory which are made to work in the SLURM system but they are easily convertible work directly from the terminal.
After that, you should install the two necessary softwares: BLAST and taxonkit. blasTAX is written to use a conda environemnt for this purpose that is called BLAST. To generate the conda environment you need to have a version of anaconda or miniconda installed. Please refer to this link. In the condaEnv/ directory you can find a file called BLAST.yml that you can use to regenerate the same environment I used to develop blasTAX. Both BLAST and taxonkit may be updated in the future so it is up to you if you want to use the newwest versions.
To generate the required conda environment, from the cloned blasTAX repo, use:
conda env create -f condaEnv/Blast.yml
An example run of blasTAX is like this below, assuming that the NCBI database is in my home/, my sequences are in the sequences/ directory as test.fasta and that I want all the output in a directory called output.
bash blasTAX.sh -i /mnt/home/benucci/blasTAX/sequences/test.fasta -d /mnt/home/benucci/blasTAX/NCBI_DB/nt -t 16 --max_hits 25 -c 0.65 -e 0.001 -p 90.0 -o /mnt/home/benucci/blasTAX/outputs
An example of 20 sequences called test.fasta is given in the sequences/ directory. All the generated blasTAX results are also provided in the outputs/ directory.
By default, blasTAX renames your sequeunce headers to Query_1, Query_2, ..., etc. The orginale headers as well as the new headers are written in the file outputs/name_mapping.txt.
If you use this, please add a star to the repo and drop me a comment.
- It is easy to install and run.
- It is a standalone program.
- It can be used with any DNA sequence.
- It could be easily integerate into a pipeline.
- It is scalabale.
A huge thanks goes to Julian Liber since part of the code is from the software CONSTAX.