GeNet

Genet is a convolutional model for shotgun metagenomic classification. Code is provided for training GeNet using the dataset from [1].

Quick guide

Install required dependencies

To install the required Python packages, install Anaconda and create a virtual environment by executing:

bash install_requirements.sh

Extact the NCBI taxonomy:

cd data
tar -xvzf nodes.dmp.tar.gz
rm nodes.dmp.tar.gz

Train the model

To train the model, run the following:

source activate genet
cd code
python genet_train.py --dir_path=../data

The first execution of the code will download the required genomes from NCBI.

The resulting weights and tensorboard log will be saved in GeNet/saved_models.

References

*[1] Rojas-Carulla, M., Tolstikhin, I., Luque, G., Youngblut, N., Ley, R. and Schölkopf, B. GeNet: Deep Representations for Metagenomics. arXiv e-print: 1901.11015. 2019. *