Vamb is a metagenomic binner which feeds sequence composition from a FASTA file of contigs, and abundance information from e.g. BAM files into a variational autoencoder and clusters the latent representation. It performs excellently with multiple samples, and pretty good on single-sample data.
The Vamb package contains several programs, including three binners:
mmseqs taxonomy
. Article.
TaxVamb produces superior bins, but requires you have run a taxonomic annotation workflow.And a taxonomy predictor:
mmseqs taxonomy
) using kmer composition and co-abundance. Go to the release branch for the instructions. PreprintSee also our tool BinBencher.jl for evaluating metagenomic bins when a ground truth is available, e.g. for simulated data or a mock microbiome.
Vamb is in continuous development. Make sure to install the latest version for the best results.
Recommended: Vamb can be installed with pip (thanks to contribution from C. Titus Brown):
pip install vamb
Note: An active Conda environment can hijack your system's linker, causing an error during installation. Either deactivate conda
, or delete the ~/miniconda/compiler_compats
directory before installing with pip.
If you want to install the latest version from GitHub, or you want to change Vamb's source code, you should install it like this:
# clone the desired branch from the repository, here master
git clone https://github.com/RasmussenLab/vamb -b master
cd vamb
pip install -e .
Note that the master branch is work-in-progress and is expected to have more bugs
If you can't/don't want to use pip/Conda, you can do it the hard way: Install the dependencies listed in the pyproject.toml
file. Compile src/_vambtools.pyx
then move the resulting binary to the inner of the two vamb
directories. Check if it works by importing vamb
in a Python session.
First, figure out what program you want to run:
vamb bin taxvamb
vamb bin default
vamb taxometer
For more command-line options, see the command-line help menu:
vamb -h
For details about how to run Vamb, see the documentation