Snakemake workflow for BLASSO-for-Rhodopsins

The workflow provides a wrapper for the BLASSO-for-Rhodopsins code for prediction of microbial rhodopsin absorption spectra. All the credit goes to the authors of the original software, see Exploration of natural red-shifted rhodopsins using a machine learning-based Bayesian experimental design. BLASSO-for-Rhodopsins was downloaded from http://www-als.ics.nitech.ac.jp/~karasuyama/BLASSO-for-Rhodopsins/BLASSO-Rhodopsin.zip, archived on 2021-02-13. The original Data folder is located in original/Data, the seqBlasso.R script is located under workflow/scripts/ and is used verbatim.

The workflow is run as follows: snakemake {rule} -c{number of cores} --use-conda --config set={set name} where number of cores is the number of the CPU cores you want to allocate to the workflow, set name is name of the training set and rule is the rule name (see below).

Currently supported sets are:

• original -- the original alignment published alongside the paper. The target sequences are added to this alignment with mafft.
• original-profiles -- the sequences from the original dataset and the target sequences are aligned with profile-profile matches using hhalign.

There are three main rules:

• predict -- predicts wavelengths for the targets in targets/{target}.fasta, the results will be in output/{set}/{target}.tsv
• train -- only does the model training to generate the file training/{set}/blasso.RData
• positions -- produces a list of position correspondences between the protein sequences in targets/{target}.fasta and the training alignment, the results are saved in output/{set}/{target}.pos

The workflow uses conda to take care of the dependencies. Note that if when using conda you encounter an error about missing libRlapack.so, you have to create the missing symlink manually, e.g. with GNU parallel: find .snakemake/conda -name liblapack.so | parallel ln -s liblapack.so {//}/libRlapack.so.