- TB-annotator: Collect and annotate Mycobacterium tuberculosis WGS data.
- CRISPRbuilder-TB: de novo reconstruction of the CRISPR locus of tuberculosis (coming soon).
The main objective of these tools is to provide various genotyping tools for the Mycobacterium tuberculosis complex. These tools work only for Whole Genome Sequencing (WGS) data in fasta files. The choice to consider only WGS data comes from the fact that the assembled genomes of M. tuberculosis often have many errors, either from areas difficult to reconstruct (e.g., CRISPR locus), multiple repeated sequences, or the use of an unrepresentative strain as a reference (h37Rv). Both programs in this package receive either a Sequence Read Archive (SRA) accession number, or a list of such SRAs in a text file (one number per line), or the ID of an NCBI BioProject.
TB-annotator is used to collect information from a given WGS tuberculosis genome. SRA files are downloaded if needed, and name of the strain, location, bioproject, biosample, etc. are first collected from the NCBI server, when available: Read files are then systematically studied, to provide the following information:
- number of reads, their length, and the mean coverage;
- 43-spacers based and 98-spacers based in silico, and in vitro inferred, spoligotypes, plus the SIT number;
- lineages based on various state-of-the-art SNP collections: PGG, Coll, Palittapongarnpim (L1), Shitikov (L2), and Stucki (L4);
- the number and position (in h37Rv) of numerous Insertion Sequences, including IS6110.
CRISPRbuilder-TB reconstructs the whole CRISPR locus starting from WGS data. This allows to deduce the true spoligotype, to detect mutants of spacers and direct repeats, insertion of mobile elements, and duplications. The Cas locus is reconstructed too. This is a semi automatic approach that leads to a set of contigs to assemble manually. Depending on the number, length, and quality of SRAs, the number of contigs can range from 1-2 patterns, in the best case scenario where the good quality of sequences allows an automatic reconstruction of the CRISPR cut in mobile element positions, to several contigs difficult to process, for too short or polluted reads.
TB-tools needs Python 3, and the following dependencies to work:
- fastq-dump from sra-tools (NCBI) : https://github.com/ncbi/sra-tools
- blastn and makeblastdb from blast+ (NCBI) : https://blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE_TYPE=BlastDocs&DOC_TYPE=Download
A 'bin' directory as been added to the project with the 3 required executables, and for the GNU/Linux, Mac and Windows platforms, but the preferred method is a clean installation from the source.
You can either download a zip file by clicking on the 'Clone or download' green button, or clone the repository by writing in a terminal:
git clone https://github.com/cguyeux/TB-tools.git
To install the required python libraries:
pip3 install -r requirements.txt
To launch a first analysis:
python tbannotator.py -sra ERR037527
Help about this script:
python tbannotator.py -h
By default, only basic information (e.g., Coll et al. lineage) is printed in the terminal (to save execution time), and results are stored in a pickled python dictionary in the 'sequence' directory. To read such a data using Python:
from pickle import load
with open('sequences/ERR037527/ERR037527.pkl', 'rb') as f:
dico = load(f.read())
Christophe Guyeux, Christophe Sola, Guislaine Refrégier. CRISPRbuilder-TB: “CRISPR-Builder for tuberculosis”. Exhaustive reconstruction of the CRISPR locus in Mycobacterium tuberculosis complex using SRA. Submitted article (2020).