This repository hosts the code for NTM-Profiler. A tool to predict species and drug resistance from NTM WGS data.

Please beware that this tools is in alpha testing and should not yet be considered for production use. If you would like to get involved and help out with testing or development please drop me a line through the issues tab.

Installation is available through conda:

conda install -c bioconda ntm-profiler=0.2.1

After installing, the relevant species and resistance databases can be downloaded by running:

ntm-profiler update_db

NTM-Profiler species prediciton is currently available to run on a fastq, bam, cram, fasta or vcf data. The output is a txt file with the species prediction and if there is a resistance database then it will also output a list of variants and if they have been associated with drug resistance.

Raw sequencing data in fastq format can been used as input using the following command. The second read is optional.

ntm-profiler profile -1 /path/to/my/reads_1.fastq.gz -2 /path/to/my/reads_2.fastq.gz -p my_sample_name

Aligned data in the form of bam or cram files can be used. Please note that the alignment files must have been generated with the same reference genome (even the chromosome names) as those used by ntm-profiler database.

ntm-profiler profile -a /path/to/my/bam -p my_sample_name

Assembled genomes or gene sequencves in fasta format can been used as input using the following command.

ntm-profiler profile -f /path/to/my/fasta -p my_sample_name

Varaints stored in VCF format can been used as input using the following command. Again the chromosome names must match those in the species-specific database

ntm-profiler profile --vcf /path/to/my/vcf -p my_sample_name

If you have used a reference genome with different sequence names that you have used to generate a bam/cram/vcf then it is possible to align the ntm-profiler databases to use the same sequence names. Please go to the custom databases section to find out more.

Other useful options arguments include

• --threads - sets the number of parallel threads
• --platform - sets the platform that was used to generate the data (default=illumina)
• --txt - outputs a text based report

A full list of arguments can be found by running ntm-profiler profile -h

Species prediction is performed by looking for pre-detemined kmers in read files which belong to a specific species. If no species is found using this method, mash is run using a database of all Mycobacteria sequences from GTDB to find the top 10 closest genomes.

Resistance prediction is performed by aligning the read data to a species-specific reference genome and looking for resistance associated genes and variants. The reference and resistance database is stored in the ntmdb github repo. At the moment resistance prediction is available for:

• Mycobacterium leprae
• Mycobacteroides abscessus subsp. abscessus
• Mycobacteroides abscessus subsp. bolletii
• Mycobacteroides abscessus subsp. massiliense

If you would like to suggest another organism please leave a comment in this thread.

Custom databases can be generated for both species and resistance prediction. The simplest way is to get a copy of ntmdb and modify to alter or add databases.

git clone https://github.com/jodyphelan/ntmdb.git

The species database consists of a two column tab delimited file with kmers and associated aspecies. To alter the current list of kmers you can modify the db/ntm_db.kmers.txt. After this, navigate to the db directory and create the database.

cd db
ntm-profiler create_species_db --kmers ntm_db.kmers.txt --prefix ntmdb --load

The --prefix argument species the name of the database. ntm-profiler will automatically look for a database with the name ntmdb if no other is provided on the commandline, so it is a good idea to keep is like this. The --load parameter will lead to the library being automatically copied to the right locations so that ntm-profiler will be able to automatically detect it.

You can find an example of the files needed to create a resistance database by navigating to db/Mycobacteroides_abscessus. You will need a reference genome, gff file, a json file containing some variables and the mutations/drug associations stored in CSV format.

.
├── Mycobacteroides_abscessus.csv
├── genome.fasta
├── genome.gff
└── variables.json

The reference genome and gff giles should be named genome.fasta and genome.gff respectively. The variables.json file should at least contain the snpEff_database used by ntm-profiler to annotate mutations.

{
    "snpEff_db":"Mycobacterium_abscessus_atcc_19977"
}

The CSV file should contain the columns as shown in the example below. For more information on the format of this file see here.

Finally, after setting these files up you can create and load the database using the following command.

ntm-profiler create_resistance_db --csv myfile.csv --prefix myspecies 

As before you can add the --load to automate loading of the database. Some important points to keep in mind:

• If you want ntm-profiler to automatically select the right database for resistance prediction after speciation, then the --prefix should match the desired species as it appears in the kmers file.
• If you are planning to use ntm-profiler on bam files, it is important to use --match_ref myref.fa to match the chromosome names used in your alignment pipeline, as it might differ with the ones used in the github repo.