PhANNs was written by Adrian Cantu in the EdwardsLab and SegallLab at San Diego State University.

PhANNs is designed to address the `dark-matter`: we are unable to assign functions to 50-90% of genes using homology-based alignments. Yet phages have analogous structural proteins with a similar 3D structure that are needed to protect the nucleic acid genome and infect next host. PhANNs is a new method that classifies phage ORFs into ten classes of structural proteins or, if that fails, into the `other` category. PhANNs was trained against a database of 70,601 manually curated phage protein sequences. The PhANN classifier reached an F₁-score of 0.916 using 10-fold cross-validation. PhANNs provide a new approach for functional annotation of phage proteins.

The PhANNs technology uses Keras and Tensorflow to build a model based on the classes of phage structural proteins. You can download Model file from the link above to donwload that model and run it using the code in our https://github.com/Adrian-Cantu/PhANNs repository.