Protein sequence similarity network (SSN) and clustering

This repository contains scripts to construct Sequence Similarity Networks (SSNs) of proteins and to cluster them by sequence similarity.

Software requirements

• NCBI BLAST+: https://blast.ncbi.nlm.nih.gov
• R v3.6 with packages seqinr, argparse, dplyr, tidyr, igraph, data.table, ggplot2 and tidyr.

Scripts

This repository contains three scripts, listed below. Each script can be run from the command line with custom input and output options. Use the help option (-h/--help) on each script to see all the available options.

Rscript script_name.R -h

• blastp_clustering.R: plot a sequence similarity network (SSN) for a set of protein sequences using all-vs-all BLAST hits and cluster them extracting the connected components.
• pfam_ssn.R: plot a sequence similarity network (SSN) of domain sequences from a subset of related Pfam families (e.g. same clan) and color by Pfam family.
• pfam_pidmat.R: create an all-vs-all sequence similarity matrix of domains in a protein (or few proteins) from a Pfam alignment, and extract the list of high-identity tandem repeat domains (HITRDs).

Results

Each script takes as input protein sequences in a FASTA format and/or protein alignments in Pfam format, plus other options and parameters, and produces figures and/or results tables. Results and figures for example input files are provided in the examples folder.

Reference

These scripts were created and used for my PhD research work at the European Bioinformatics Institute (EMBL-EBI). More details are provided in my Thesis (insert link when available).

Aleix Lafita - April 2021