-
Notifications
You must be signed in to change notification settings - Fork 13
Installation
The LIMIX QTL pipeline can be downloaded directly from github and placed where you want to use it.
Direct download link: git clone https://github.com/single-cell-genetics/limix_qtl.git
Next to the github package, you need install its dependencies. We recommend you either use conda or docker / singularity for this. Instructions below.
We recommend you install the limix based QTL mapping pipeline in a separate conda environment. Here we create the environment "limix_qtl", and use limix version 2. For faster installations consider using mamba instead of conda, (by replacing conda with mamba).
conda create -n limix_qtl python=3.8 anaconda
source activate limix_qtl
conda install -c anaconda pytest pytables scikit-learn
conda install -c conda-forge bgen=4 pandas-plink
pip install bgen-reader
pip install glimix-core
Save the following lines as a docker "recipe/definition" file or use this the "limix.def" file in the repository.
Bootstrap: docker
From: continuumio/miniconda3
######################## Limix based QTL mapping ###########################
%runscript
export PATH=/opt/conda/bin:${PATH}
/bin/bash
%post
export PATH=/opt/conda/bin:${PATH}
git clone https://github.com/single-cell-genetics/limix_qtl.git
# Initiate conda
conda package
conda config --add channels conda-forge
# Install C & Python packages with conda.
conda install -c conda-forge python=3.8 'blas=*=*mkl' 'numpy==1.21' pytest pytables scikit-learn matplotlib-venn bgen=4 pandas-plink h5py
# Install remaining Python packages with pip
pip install bgen-reader
pip install glimix-core
# Install R
apt-get update
apt-get -y install r-base
# Install R packages (from within R)
R --slave -e 'install.packages(c("BiocManager","dplyr","readr"))'
R --slave -e 'BiocManager::install(c("qvalue","multtest","rhdf5"))'
##############################################################################
To create the image you need to run sudo singularity build limix.simg ./limix.def
Some filesystems have file locking disabled, to be able to use the tool, which by default tries to lock the output file, use: export HDF5_USE_FILE_LOCKING=FALSE
We only support limix v3, so please check you have installed the correct version.
Please check and make sure your numpy is bound to intel MKL, which makes the analyses much faster (especially on intel machines). Please find further information here