MSA Data Collection and Filtering

This folder contains scripts and files related to the generation, filtering, and usage of multiple sequence alignments (MSAs) for the OLF domain.

Step 1: Get MSA from hmmer

Note: we provide the finalized dataset at subsampled_MSA.a2m. These steps are only to reproduce our results.

Ensure you have the EVcouplings repository cloned into this directory, as we use it for running hmmer.

git clone https://github.com/debbiemarkslab/EVcouplings.git

Ensure that you have hmmer installed and available at the folder: profilehmm/hmmer-3.4. You also have to download the UniRef100 database and place it at profilehmm/hmmer-3.4/bin/uniref100.fasta.

You also need to create a conda environment using the msagen.yml file provided:

conda env create -f msagen.yml
conda activate msagen

For an example of how to run hmmer through EVcouplings, refer to scripts/msagen.py.

The MSA generated from this is located at results/OLF_HUMAN/align/OLF_HUMAN.a2m.

Step 2: First round of filtering

Run filtering.py to remove short/duplicate sequences and filter the MSA to only include sequences with 200 residues or more. This generates a filtered MSA at OLF_filtered.a2m.

Step 3: MMseqs2 clustering and subsampling

We use MMseqs2 to cluster the MSA.

See the main README.md for additional details on running MMseqs2.

This generates the subsampled MSA at subsampled_MSA.a2m.

File Descriptions

File/Folder	Description
fasta/OLF.fasta	FASTA file for the OLF domain.
scripts/msagen.py	Script to run hmmer through EVcouplings.
scripts/config_files/OLF_HUMAN.txt	Configuration file specifying hmmer parameters.
results/OLF_HUMAN/align/OLF_HUMAN.a2m	Output MSA from EVcouplings.
filtering.py	First round of filtering: removes duplicate sequences and drops those shorter than 200 residues.
OLF_filtered.a2m	Filtered MSA after removing short/duplicate sequences.
subsampled_MSA.a2m	Subsampled MSA after MMseqs2 clustering.

For Phylogenetic Analysis

To see the sequences used for phylogeny, refer to:

../phylo/MSA_80cluster.fas