TaxoTagger is an open-source Python library for DNA taxonomy identification, which involves categorizing DNA sequences into their respective taxonomic groups. It is powered by deep learning and semantic search to provide efficient and accurate results.
Key Features:
- 🚀 Build vector databases from DNA sequences with ease
- ⚡ Conduct efficient semantic searches for precise results
- 🛠 Extend support for custom embedding models effortlessly
- 🌐 Interact seamlessly through a user-friendly web app
TaxoTagger requires Python 3.10 or later.
# create an virtual environment
conda create -n venv-3.10 python=3.10
conda activate venv-3.10
# install the `taxotagger` package
pip install --pre taxotaggerfrom taxotagger import ProjectConfig
from taxotagger import TaxoTagger
config = ProjectConfig()
tt = TaxoTagger(config)
# creating the database will take ~30s
tt.create_db('data/database.fasta')By default, the ~/.cache/mycoai folder is used to store the vector database and the embedding model. The MycoAI-CNN.pt model is automatically downloaded to this folder if it is not there, and the vector database is created and named after the model.
from taxotagger import ProjectConfig
from taxotagger import TaxoTagger
config = ProjectConfig()
tt = TaxoTagger(config)
# semantic search and return the top 1 result for each query sequence
res = tt.search('data/query.fasta', limit = 1)The data/query.fasta file contains two query sequences: KY106088 and KY106087.
The search results res will be a dictionary with taxonomic level names as keys and matched results as values for each of the two query sequences. For example, res['phylum'] will look like:
[
[{"id": "KY106088", "distance": 1.0, "entity": {"phylum": "Ascomycota"}}],
[{"id": "KY106087", "distance": 0.9999998807907104, "entity": {"phylum": "Ascomycota"}}]
]The first inner list is the top results for the first query sequence, and the second inner list is the top results for the second query sequence.
The id field is the sequence ID of the matched sequence. The distance field is the cosine similarity between the query sequence and the matched sequence with a value between 0 and 1, the closer to 1, the more similar. The entity field is the taxonomic information of the matched sequence.
We can see that the top 1 results for both query sequences are exactly themselves. This is because the query sequences are also in the database. You can try with different query sequences to see the search results.
Please visit the official documentation for more details.
Please submit an issue if you have any question or feedback.
If you use TaxoTagger in your work, please cite it by clicking the Cite this repository on right top of this page.