flucleave 0.1.5

Deep learning prediction of Influenza A virus pathogenicity from H5 and H7 cleavage sites

FluCleave

Deep learning prediction of influenza virus pathogenicity from HA cleavage sites for H5 and H7 variants.

Features

• Analyzes H5 and H7 hemagglutinin (HA) cleavage site sequences
• Predicts high/low pathogenicity using deep learning
• Handles both DNA and protein sequences
• Command-line interface for easy use
• Trained on curated dataset of known pathogenic sequences

Installation

pip install flucleave

Usage

Predict pathogenicity from FASTA file:

flucleave predict --fasta sequences.fasta --output-dir results/ --prefix sample_name

Train new model (optional):

flucleave train --training-csv data.csv

Training Data

This model was trained on data curated from 2 sources:

• Offlu Influenza A Cleavage Sites Document for H5 and H7
• GISAID query of all LPAI HA sequences with complete cleavage sites (as of 2025-02-21)

Data Format

Input FASTA should contain HA protein sequences in either amino acid or DNA format.

License

FluCleave is licensed under the MIT License.