Whokaryote

Whokaryote uses a random forest classifier that uses gene-structure based features and optionally Tiara (https://github.com/ibe-uw/tiara) predictions to predict whether a contig is from a eukaryote or from a prokaryote.

You can use Whokaryote to determine which contigs need eukaryotic gene prediction and which need prokaryotic gene prediction.

Issues can be created via GitHub: https://github.com/LottePronk/whokaryote

Publication

If you have used Whokaryote in your work, please cite our paper:

Pronk, Lotte J U, and Marnix H Medema. “Whokaryote: distinguishing eukaryotic and prokaryotic contigs in metagenomes based on gene structure.” Microbial genomics vol. 8,5 (2022): https://doi.org/10.1099/mgen.0.000823

---

Installation

Whokaryote was developed to run from the commandline on a UNIX-based system such as Ubuntu or MacOS. If you want to use Whokaryote on your Windows PC, we recommend you install Windows Subsystem for Linux (WSL). See https://docs.microsoft.com/en-us/windows/wsl/about

Recommended installation (Linux/Mac):

You can now easily install whokaryote via conda (or mamba). Simply use the following command while in a (dedicated) conda environment:

conda create -c bioconda -n whokaryote whokaryote

Are you new to using conda? You can download miniconda here: https://docs.conda.io/en/latest/miniconda.html

Alternative installation for Windows:

Alternatively, you can install Whokaryote and its dependencies in a conda environment in Windows, see the instructions. You can download miniconda here: https://docs.conda.io/en/latest/miniconda.html

1. Make a new and empty conda environment, and activate it:

   • conda create -n whokaryote python==3.8
   • conda activate whokaryote

2. Install dependencies with these commands:

   • python -m pip install tiara (see: https://github.com/ibe-uw/tiara)
   • download the windows binary of prodigal from the following link: https://github.com/hyattpd/Prodigal/releases/tag/v2.6.3
   • change the file name: move prodigal.windows.exe prodigal.exe
   • move the file to the directory from where you run whokaryote.

3. Install whokaryote:

   • Navigate to a directory you want to install whokaryote in.
   • Clone whokaryote to this directory: git clone https://git.wur.nl/lotte.pronk/whokaryote.git (or git clone https://github.com/LottePronk/whokaryote.git )
   • Go to the whokaryote directory: cd whokaryote
   • Install whokaryote in your (conda) environment withpython setup.py install

---

Using whokaryote

Use whokaryote.py --help to see all the options:

-h, --help            show this help message and exit
--contigs CONTIGS     The path to your contigs file. It should be one (multi)fasta (DNA).
--outdir OUTDIR       Specify the path to your preferred output directory. No / at the end.
--prodigal_file PRODIGAL_FILE
                        If you already have prodigal gene predictions, specify path to the .genes or .gff file
--f                   If you want new multifastas with only eukaryotes and only prokaryotes. This can take a long time.
--minsize MINSIZE     Select a minimum contig size in bp, default = 5000. Accuracy on contigs below 5000 is lower.
--model MODEL         Choose the stand-alone model or the tiara-integrated model: S or T. Option 'T' only works with argument --contigs

Note: When using Windows, you may need to run whokaryote from the commandline as follows: python whokaryote\bin\whokaryote.py [options]

Example:

whokaryote.py --contigs contigs.fasta --outdir whokaryote_output --prodigal_file contigs_genes.gff

This is the standard way to run Whokaryote. If you don't specify --model, it will run the model that integrates Tiara predictions. This is recommended.

If you already have gene annotation files (.gff or gene coordinates file from prodigal), use the --prodigal_file option.

If you don't have an annotation file yet, you can use only --contigs and --outdir. Prodigal will then be run on your contigs file, which may take a while.

With the example command, you will get the following output files:

• A file with all contig headers that were classified as eukaryotic (eukaryote_contig_headers.txt), and a similar file for prokaryotic contig headers (prokaryote_contig_headers.txt)
• A fasta file with only contigs that were longer than 5000 bp, called contigs5000.fasta
• A gene coordinates file generated by prodigal called contigs_genes.genes
• A .csv file with all the calculated features called featuretable.csv
• A file with the tiara predictions called tiara_pred.txt
• A .csv file with the calculated features and the predictions called featuretable_predictions_[model].tsv
• A .tsv file with only the contig ID and the predictions called whokaryote_predictions_[model].tsv

Note that contigs with less than 2 genes or shorter than the minimum size are not classified and do not appear in the output.