functional_dark_matter_code

This repository contains script used for the manipulation of sequence alignments, preparation for 3D modeling and mass-scale 3D structure search of PDB models against reference datasets (e.g. PDB, SCOP) with TMalign/MMalign.

The aforementioned code was used in Pavlopoulos, G.A., Baltoumas, F.A., Liu, S. et al. Unraveling the functional dark matter through global metagenomics. Nature 622, 594–602 (2023), doi: https://doi.org/10.1038/s41586-023-06583-7

Contents:

• generate_seed.py : parses a multiple sequence alignment in FASTA format and produces its consensus sequence and a non-redundant dataset ("seed" alignment)

• prepare_MSA_for_alphafold.py: parses an alignment in FASTA or A3M format and produces a refined MSA output that can be used as input for AlphaFold2 or ColabFold / LocalColabFold.

• structure_search.py: performs structural alignment for a set of query PDB/mmCIF files against a set of reference PDB/mmCIF files with TMalign and/or MMalign.

System Requirements:

• OS: Linux, MacOS or Windows 10/11 with WSL (Windows Subsystem for Linux) installed
• Python v. 3.8 or newer
• Python modules: numpy, scipy, scikit-learn, biopython, prody
• TMalign & MMalign
• HH-suite 3

Installation instructions:

Note: Examples are shown for Ubuntu/Debian and similar Linux distributions. For rpm-based distributions, please use the equivalent package managers

1. Download and install software dependencies:

sudo apt install python3 python3-pip
sudo apt install hhsuite
sudo apt install tmalign
sudo apt install mmalign

pip3 install numpy scipy scikit-learn biopython prody

Optimally, this will take around 5 minutes to set up.

Example runs:

• Generation of seed alignments

cd example_data/MSAs/

python ../../generate_seed.py MSA_list.txt

• Prepare an alignment for 3D modeling with AlphaFold2:

cd example_data/MSAs/seed/
python ../../../prepare_MSA_for_alphafold.py --input F040820.fasta --hhfilter_bin /opt/hh-suite/bin/hhfilter --id 90 --cov 75

• Pairwise structural alignment of NMPF 3D models against SCOP domains

cd example_data/structures

python ../../structure_search.py --query NMPF_list.txt --target SCOP_list.txt --method tmalign --cpus 4 1> results_table.tsv 2>stderr.log