Putative genome contamination has minimal impact on the GTDB taxonomy

This repository contains content for an analysis that evaluates the impact of contamination on taxonomy. It is not designed for conducting new analyses. Instead, it serves as a documented record of the methods used in the manuscript and aids in reproducing the results.

For more information, read the manuscript at ... (pending publication).

1. Installation

This repository is designed to be run using Luigi on a Unix-based system.

1.1 Conda environment

To create the conda environment run the following command to install the required dependencies specified in conda/env.yaml:

bash conda env create -f conda/env.yaml

1.2 Reference data

This analysis requires the GTDB R207 database, including called genes. These must be downloaded using the published data from here: https://doi.org/10.48610/85c83e3

1.3 Configuration

The following variables must be set in workflow/config.py:

• DIR_PROJECT: This is the absolute path to the project directory, it is expected that the reference data are stored in DIR_PROJECT/output.

Other variables do not need to be modified assuming the reference data is setup correctly.

2. Running the analysis

After you have set the PYTHONPATH to the cloned repository, the full analysis can be run using the following command:

bash python -m workflow run

2.1 - Running individual tasks

To run individual tasks, you can modify the __main__.py file under the run target to specify the output file you want to generate. If you are not familiar with Luigi, you should familiarise yourself with the output target.

2.2 - Notable output targets for manuscript

There are a number of key output targets used in generating figures for the manuscript, these are detailed below:

Figure 1: - This is an illustrative example.

Figure 2: - 2a: Data are determined by running GUNC (see methods). - 2b: ./notebook/manuscript/x_n_contigs_y_frequency_hue_pass.ipynb - 2c: ./notebook/manuscript/taxonomic_novelty_v2.ipynb - 2d: ./notebook/manuscript/gunc_stats_d.ipynb - 2e: ./notebook/manuscript/hist_of_top_failed_sp_gunc_stats_e.ipynb

Figure 3: - This is an illustrative example.

Figure 4: - Directed/random removal: workflow/fastani_random/e_analyse_results.py - Inter-rep distances: ./notebook/manuscript/figure_ani_outcome_rep_rep_dist.ipynb

Figure 5: - ./workflow/circos_plot/create_circos_plot.py

Supplementary tables and main text data: - ./workflow/final/v2_generate_master_tsv.py

FastANI analysis: - ./workflow/fastani_random

BAC120 analysis: - ./workflow/v2_fasttree_marker_split

2.3 - Notable methods

The majority of the key methods used in this analysis are abstracted into the Genome class, this can be found in ./workflow/model/genome.py. Note that these methods are called in their respective Luigi tasks.

Contig ranking: - get_gunc_max_css_contig_ranking

Marker congruence: - get_marker_congruence

3 - Reading individual results

All genome-specific results are stored in the output of the reference data. To view a specific result, download the reference data and view the file under each genome directory.