ViTax is a virus taxonomy classification tool powered by the HyenaDNA foundation model, a large language model for long-range genomic sequences at single nucleotide resolution. Aiming to the specific challenges of virus taxonomy classification, ViTax integrates supervised prototypical contrastive learning to address the challenge of highly imbalanced distributions across various taxonomic clades and employs a belief mapping tree on the Least Common Ancestor algorithm to achieve the taxonomy level with the most confidence adaptively. ViTax support the classification of up to 631 genera.
ViTax is a Python package. To install it, run the following command in your terminal:
git clone https://github.com/Ying-Lab/ViTax.git
pip install -r requirements.txt
python ViTax.py [--contigs INPUT_FA] [--out OUTPUT_TXT]
--contigs INPUT_FA input fasta file
--out OUTPUT_TXT The output csv file (prediction_output.txt default)
--confidence The confidence threshold of the prediction (0.6 default)
--rc include reverse complement prediction (default True)
--window_size The sliding window size (default 400)
python ViTax.py --contigs test.fa --out prediction_output.txt --confidence 0.6 --rc False --window_size 400