Greengenes Dataset (modified for deeptaxa)

This dataset contains 16S rRNA gene sequences with hierarchical taxonomic annotations, designed for training and evaluating models like DeepTaxa. It is a processed version of the Greengenes database, widely used in microbiome research.

Dataset Details

The dataset includes the following files:

File Name	Type	Number of Sequences	Size
gg_2024_09_training.fna.gz	FASTA (sequences)	277,336	~96.4 MB
gg_2024_09_training.tsv.gz	TSV (taxonomy labels)	277,336	~2.6 MB
gg_2024_09_testing.fna.gz	FASTA (sequences)	69,335	~24.1 MB
gg_2024_09_testing.tsv.gz	TSV (taxonomy labels)	69,335	~0.8 MB

• Format: FASTA for sequences (compressed), TSV for labels with 7 taxonomic levels (compressed).
• Source: Derived from Greengenes (DeSantis et al., 2006), downloaded from https://ftp.microbio.me/greengenes_release/2022.10/

Usage

Use with DeepTaxa:

deeptaxa train --fasta-file gg_2024_09_training.fna.gz --taxonomy-file gg_2024_09_training.tsv.gz

Models

This dataset is compatible with the DeepTaxa.

License

This dataset is a modified version of the Greengenes2 dataset, used for training and testing DeepTaxa. It is distributed under the terms of the Modified BSD License, as required by the original Greengenes2 license.

Original Greengenes2 License

Copyright (c) 2022-, Greengenes2 [email protected]

All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: * Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. * Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. * Neither the name of Greengenes2 nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE BIOM FORMAT DEVELOPMENT TEAM BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Modifications

This modified version (gg_2024_09) was created by the Systems Genomics Lab for use with DeepTaxa. The Greengenes database, downloaded from https://ftp.microbio.me/greengenes_release/2022.10/, was processed as follows:

1. The original dataset, containing 23,450,269 sequences, was downloaded.
2. Missing taxonomic classifications were filled with "Unclassified" or derived labels. For each sequence, the taxonomy table's ranks (e.g., domain, phylum, class) were processed sequentially. If a rank was empty or contained only whitespace:
   • If no prior rank had a valid value, the field was left empty.
   • If the previous rank's value included "Unclassified," that value was copied to the current rank.
   • Otherwise, the field was set to "Unclassified_" followed by the previous rank's value (e.g., "Unclassified_Proteobacteria" for a missing class after phylum "Proteobacteria"). This ensured hierarchical consistency in the taxonomy labels.
3. Sequences were filtered to retain only those with at least 300 nucleotides, resulting in 354,023 sequences.
4. Sequence IDs were extracted, and the taxonomy table was filtered to match these IDs, yielding 346,671 entries (7,352 IDs not found were noted).
5. A clean ID list was generated from the filtered taxonomy table (346,671 IDs).
6. Sequences were filtered to match the clean ID list, retaining 346,671 sequences.
7. The dataset was split into training (277,336 sequences) and testing (69,335 sequences) sets using an 80:20 ratio and a fixed random seed.
8. Training and testing sequence IDs were extracted (277,336 and 69,335 IDs, respectively).
9. The taxonomy table was filtered to create separate training (277,336 rows) and testing (69,335 rows) label files.

These steps ensured the dataset was cleaned, filtered, and split for use with DeepTaxa.

Copyright (c) 2025, Systems Genomics Lab

Distributed under the same Modified BSD License terms as the original Greengenes2 dataset.

Citation

For the dataset:

@article{DeSantis2006,
  author = {DeSantis, T. Z. and Hugenholtz, P. and Larsen, N. and Rojas, M. and Brodie, E. L. and Keller, K. and Huber, T. and Dalevi, D. and Hu, P. and Andersen, G. L.},
  title = {Greengenes, a Chimera-Checked 16S rRNA Gene Database and Workbench Compatible with ARB},
  journal = {Applied and Environmental Microbiology},
  year = {2006},
  doi = {10.1128/AEM.03006-05}
}

For DeepTaxa: See GitHub.

Contact

Report issues on GitHub.