Pipeline to generate complete public health microbiology reports from sequenced isolates
Public health microbiology labs receive batches of bacterial isolates whenever there is a suspected outbreak.In modernised labs, each of these isolates will be whole genome sequenced, typically on an Illumina or Ion Torrent instrument. Each of these WGS samples needs to quality checked for coverage, contamination and correct species. Genotyping (eg. MLST) and resistome characterisation is also required. Finally a phylogenetic tree needs to be generated to show the relationship and genomic distance between the strains. All this information is then combined with epidemiological information (metadata for each sample) to assess the situation and inform further action.
Feel free to browse some example reports.
Nullarbor currently only supports Illumina paired-end sequencing data; single end reads, from either Illumina or Ion Torrent are not supported. All jobs are run on a single compute node; there is no support yet for distributing the work across a high performance cluster.
- Clean reads
- remove adaptors, low quality bases and reads (Trimmomatic)
- Species identification
- k-mer analysis against known genome database (Kraken, Kraken2, Centrifuge)
- De novo assembly
- User can select (SKESA, SPAdes, Megahit, shovill, Velvet)
- Annotation
- Add features to assembly Prokka)
- MLST
- From assembly w/ automatic scheme detection (mlst + PubMLST)
- Resistome
- From assembly (abricate + Resfinder)
- Virulome
- From assembly (abricate + VFDB)
- Variants
- From reads aligned to reference (snippy)
- Core genome SNPs
- From reads (snippy-core)
- Infer core SNP phylogeny
- Pan genome
- From annotated contigs (Roary)
- Report
- Summary isolate information (HTML + Plotly.JS + DataTables + PhyloCanvas)
- More detailed per isolate pages (COMING SOON)
You need to install both the software and the databases separately.
conda install -c conda-forge -c bioconda -c defaults nullarbor
Install Homebrew (macOS) or LinuxBrew (Linux).
brew install brewsci/bio/nullarbor
This is the hardest way to install Nullarbor.
cd $HOME
git clone https://github.com/tseemann/nullarbor.git
# keep running this command and installing stuff until it says everything is correct
./nullarbor/bin/nullarbor.pl --check
# For Perl modules (eg. YAML::Tiny), use one of the following methods
apt-get install yaml-tiny-perl # ubuntu/debian
yum install perl-YAML-Tiny # centos/redhat
cpan YAML::Tiny
cpanm YAML::Tiny
You need to install a Kraken database (~8 GB).
wget https://ccb.jhu.edu/software/kraken/dl/minikraken_20171019_8GB.tgz
tar -C $HOME -zxvf minikraken_20171019_8GB.tgz
You need to install a Kraken2 database (~8 GB).
wget ftp://ftp.ccb.jhu.edu/pub/data/kraken2_dbs/minikraken2_v2_8GB_201904_UPDATE.tgz
tar -C $HOME -zxvf minikraken2_v2_8GB_201904_UPDATE.tgz
Install a Centrifuge database (~8 GB):
wget ftp://ftp.ccb.jhu.edu/pub/infphilo/centrifuge/data/p_compressed+h+v.tar.gz
mkdir $HOME/centrifuge-db
tar -C $HOME/centrifuge-db -zxvf p_compressed+h+v.tar.gz
Then add the following to your $HOME/.bashrc so Nullarbor can find the databases:
export KRAKEN_DEFAULT_DB=$HOME/minikraken_20171019_8GB
export KRAKEN2_DEFAULT_DB=$HOME/minikraken2_v2_8GB_201904_UPDATE
export CENTRIFUGE_DEFAULT_DB=$HOME/centrifuge-db/p_compressed+h+v
You should be good to go now. When you first run Nullarbor it will let you know of any missing dependencies or databases.
Nullarbor does a self-check of all binaries, Perl modules and databases:
nullarbor.pl --check
This is a file, one line per isolate, with 3 tab separated columns: ID, R1, R2.
Isolate1 /data/reads/Isolate1_R1.fq.gz /data/reads/Isolate2_R1.fq.gz
Isolate2 /data/reads/Isolate2_R1.fq /data/reads/Isolate2_R2.fq
Isolate3 /data/old/s_3_1_sequence.txt /data/old/s_3_2_sequence.txt
Isolate3b /data/reads/Isolate3b_R1.fastq /data/reads/Isolate3b_R2.fastq
This is just a regular FASTA or GENBANK file. Try and choose a reference phylogenomically similar to your isolates.
If you use a GENBANK or EMBL file the annotations will be used to annotate SNPs by Snippy.
This command will create a new folder with a Makefile in it:
nullarbor.pl --name PROJNAME --mlst saureus --ref US300.fna --input samples.tab --outdir OUTDIR
This will check that everything is okay. One of the last lines it prints is the command you need to run to actually perform the analysis e.g.
Run the pipeline with: nice make -j 4 -C OUTDIR
So you can just cut and paste that:
nice make -j 4 -C OUTDIR
The -C option just means to change into the /home/maria/listeria/nullarbor folder first, so you could
do this instead:
cd OUTDIR
make -j 4
firefox OUTDIR/report/index.html
Here are some example reports.
Once set up, a Nullarbor folder can be used in a few different ways. See what's available with this command:
make help
You should not do a full run the first time, because it will probably contain outliers and QC failures. To build a quick "rough" tree:
make preview
This will create a mini-report in the same report/ folder.
Use this to identify outliers and then comment them out (or delete)
them from the --input file. Then type the following to regenerate
the report for a second round of inspection:
make again
make preview
When you are happy with the result, proceed with the full analysis:
make again
make
Often you want to perform multiple analyses where some of the isolates
have been used in previous Nullarbor runs. It is wasteful to recompute
results you already have. The --prefill option allows you to "copy"
existing result files into a new Nullarbor folder before commencing
the run.
To set it up, add a prefill section to nullarbor.conf as follows:
# nullarbor.conf
prefill:
contigs.fa: /home/seq/MDU/QC/{ID}/contigs.fa
The {ID} will replaced for each isolate ID in your --input TAB file
and the contigs.fa copied from the source path specified. This will
prevent Nullarbor having to re-assemble the reads.
Nullarbor 2.x has a plugin system for assembly and tree building.
These can be changed using the --assembler and --treebuilder options.
Read trimming is off by default, because most sequences are now
provided pre-trimmed, and retrimming occupies much disk space.
To trim Illumina adaptors, use the --trim option.
After examining the report from your initial analysis, it is common to observe some outliers, or bad data. In this case, you want to remove those isolates from the analysis, but want to minimize the amount of recomputation needed.
Just go to the original --input TAB file and either (1) remove
the offending lines; or (2) just add a # symbol to "comment out"
the line and it will be ignored by Nullarbor.
Then go back into the Nullarbor folder and type make again
and it should make a new report. Assemblies and SNPs won't be
redone, but the tree-builder and pan-genome components will
need to run again.
As per "Removing isolates" above, you can also add in more isolates
to your original --input TAB file when you want to expand the analysis.
Then just type make again and it should only recalculate
things it needs to, saving a lot of computation.
If you don't want to cut and paste the make .... instructions to
start the analysis, just add the --run option to your nullarbor.pl command.
NULLARBOR_CONF- default--conf, the path tonullarbor.confNULLARBOR_CPUS- default--cpusNULLARBOR_ASSEMBLER- default--assemblertoolNULLARBOR_TREEBUILDER- default--treebuildertoolNULLARBOR_TAXONER- default--taxonertool
Nullarbor has many dependencies, so you are best off using a package
manager to install it. Type nullarbor.pl --check to see what you need.
Perl: Bio::Perl Time::Piece List::Util Path::Tiny YAML::Tiny Moo SVG Text::CSV List::MoreUtils IO::File
Tools: seqtk trimmomatic prokka roary mlst abricate seqret skesa megahit spades shovill snippy snp-dists newick-utils iqtree fasttree quicktree kraken kraken2 centrifuge
Databases: minikraken centrifuge-bacvirhum
Note that these are only the immediate dependencies and that the tools listed above will depend on various other tools, Perl modules, and Python modules.
The Nullarbor is a huge treeless plain that spans the area between south-west and south-east Australia. It comes from the Latin "nullus" (no) and "arbor" (tree), or "no trees". As this software will generate a tree, there is an element of Australian irony in the name.
Submit problems to the Issues Page
Seemann T, Goncalves da Silva A, Bulach DM, Schultz MB, Kwong JC, Howden BP. Nullarbor Github https://github.com/tseemann/nullarbor