Installation

BACoN is a Python package (standard library only) that runs command-line programs, which are installed with conda.

bioconda (recommended)

conda create -n bacon -c conda-forge -c bioconda bacon-nanopore
conda activate bacon
bacon --version

The package is named bacon-nanopore because conda-forge already has an unrelated package called bacon; the command is bacon. It installs BACoN with every program below except Bandage.

From the source code

For development, or the latest code:

git clone https://github.com/duceppemo/BACoN
cd BACoN
conda env create -f environment.yml
conda activate BACoN
pip install .
bacon --version

environment.yml lists the programs and their minimum versions:

Program	Used for	Minimum
minimap2	baiting; templated assembly; off-reference share of de novo assemblies; SNPs (--snp-method minimap2, the default for templated assemblies)	2.24
samtools	templated assembly (samtools consensus -X r10.4_sup)	1.21
Filtlong	read filtering	0.2.1
Flye	de novo assembly (-a flye)	2.9.5
myloasm	de novo assembly (-a myloasm)	0.7
SKA2	SNPs (--snp-method ska; the default for de novo assemblies)	0.5
Parsnp, HarvestTools	SNPs (--snp-method parsnp)	2.1.2 (with 2.1.1, every SNP is one base off)
FastTree	tree (default)	2.1.11
IQ-TREE	tree (--tree iqtree)	2.2
BBMap (BBDuk)	baiting (-b bbduk)	39
Bandage	optional: pictures of the assembly graphs
python-isal	optional: faster reading of gzipped reads

BACoN checks at start-up that the programs needed by the chosen options are on the PATH, and lists the missing ones with the command to install them.

Checking the installation

The example is in the repository (example/), not in the conda package. From a clone, or after downloading the example folder of the release:

curl -sL https://github.com/duceppemo/BACoN/archive/refs/tags/v0.4.0.tar.gz | tar -xz --strip-components=1 BACoN-0.4.0/example
bash example/run_example.sh

generates a small simulated dataset (an annotated 30 kb circular reference, four samples with known SNPs and their metadata), runs BACoN with the default settings and checks the results against the truth: every pairwise SNP distance, the region, gene, context and effect of every SNP, and the report's annotation and metadata (Example); it prints one OK line per check. It takes a few seconds.

pip only

If the programs are already installed (for example in an HPC module system), BACoN itself installs with pip install https://github.com/duceppemo/BACoN/archive/refs/tags/v0.4.0.tar.gz (or pip install git+https://github.com/duceppemo/BACoN for the latest code).

Troubleshooting

• The environment takes long to solve, or mamba reports "nothing provides ..." for packages that exist: some mamba 2.x versions misreport conflicts; conda env create (with the libmamba solver) solves it.
• samtools consensus: unrecognised option -X: samtools is older than 1.17 (which added -X); update it. BACoN is tested with samtools 1.21 and later.
• Upgrading from BACoN 0.2: create a new environment; the old requirements.txt environment pinned programs that are no longer used (Porechop, Shasta, Rebaler, Snippy, PhaME, RAxML, ete3).