Hierarchical Edge Bundling of FracMinHash dN/dS (also a dN/dS genomic level estimation)

Hierarchical edge bundling is an easy way to display quantitative pairwise relationships among thousands of genomes [1]. Here, we use the Hierarchical Edge Bundling R Package to present genomic dN/dS estimations between archaeal and bacterial genomes from the Genome Taxonomy Database, where red and blue represent positive and negative selection, respectively. Please find the reproducible script to generate the Hierarchical Edge Bundling Figure presented in manuscript for "Leveraging FracMinHash Containment for Genomic dN/dS".

Table of contents

• Environment setupe
  • Conda
  • Pip
• Quick start
• Dataset Information
  • Taxonomic datasets
  • Input
• Usage
  • Output
• Acknowledgements
• References
• Our paper

Environment Setup

Create either a conda or pip environment to reproduce this figure.

Conda environment

conda env create -f environment.yml
conda activate hierarchical_env

Pip environment

python3 -m venv hierarchical_pip_env
source hierarchical_pip_env/bin/activate
pip install -r requirements.txt

Quick start

Please execute the following script to generate figure:

Rscript script.r --input fmh_omega_7.csv --taxonomy species

Note: Currently, this file is not available on the repo, if you plan to run this please reach out to bioinfwithjudith.

Dataset Information

Taxonomic datasets

This file is to help you out on what the files mean and there purpose. If you have any questions, feel free to reach out!

Download the following files:

wget https://data.ace.uq.edu.au/public/gtdb/data/releases/release214/214.0/ar53_taxonomy_r214.tsv
wget https://data.ace.uq.edu.au/public/gtdb/data/releases/release214/214.0/bac120_taxonomy_r214.tsv

These files contain taxonomic information of archaeal and bacterial genomes that are used to produce a phylogenetic tree and labels.

Column Name	Description
NCBI_fasta_name	the identifier also known as an accession number used by ncbi.com
custom_fasta_name	a name Judith gave to the sequence based on published data
phylum	taxonomic classifier
class	taxonomix classifier
order	taxonomix classifier
family	taxonomix classifier
genus	taxonomix classifier
species	taxonomix classifier
strain	taxonomix classifier

Input

We include the a FracMinHash dN/dS result to generate this figure.

Filename: fmh_omega_7.csv

Purpose: This file contains the dN/dS values. dN/dS above 1 is postive selection (generally shown in red) and a dN/dS below 1 is negative selection (generally shown in blue). The most important columns here are A, B, and dNdS_ratio_constant.

Note: Currently, this file is not available on the repo, if you plan to run this please reach out to bioinfwithjudith.

Column Name	Description
A	genome A
B	genome B
containment_nt	containment estimated among nucleotide sequences of genomes A and B
ksize	the size of the subsets used for sketching
containment_protein	containment estimated among protein sequences of genomes A and B
sequence_comparison	comparison between genome name A and B
dN_constant	estimated nonsynonymous mutations with constant
dS_constant	estimated synonymous mutations with constant
dN	estimated nonsynonymous mutations
dS	estimated synonymous mutations
dNdS_ratio	the estimated dN/dS without constant
dNdS_ratio_constant	the estimated dN/dS with constant

Usage

You can always find usage instructions here:

Rscript script.R --help

Argument	Description
Input	This is a FracMinHash dN/dS csv file
Taxonomy	The taxonomy information for species in input file

Output

An output_[taxonomic level].png file of the figure and a output_[taxonomic level].csv with the names of genomes and their pairwise FracMinHash dN/dS estimations will be produced. This is done because the script currently randomly chooses 100 estimations (50 dN/dS estimations < 1 and 50 dN/dS estimations > 1) to be represented in the figure. The blue represents negative selection (dN/dS<1) and the dark orange represents positive selection (dN/dS>1).

Acknowledgements

Thank you Jinglin Feng for your contributions to this figure!

References

[1] D. Holten (2006). Hierarchical Edge Bundles: Visualization of Adjacency Relations in Hierarchical Data. IEEE Transactions on Visualization and Computer Graphics, 12(5)

Our paper

Leveraging FracMinHash Containment for Genomic dN /dS. Judith S. Rodriguez, Mahmudur Rahman Hera, and David Koslicki. In preparation.