phold is a sensitive annotation tool for bacteriophage genomes and metagenomes using protein structural homology.

phold uses the ProstT5 protein language model to rapidly translate protein amino acid sequences to the 3Di token alphabet used by Foldseek. Foldseek is then used to search these against a database of over 1 million phage protein structures mostly predicted using Colabfold.

Alternatively, you can specify protein structures that you have pre-computed for your phage(s) instead of using ProstT5 with phold compare.

The phold databse consists of over 1 million protein structures generated using Colabfold from the following databases:

• PHROGs — 440549 de-deuplicated proteins. Proteins over 1000AA were chunked into 1000AA components.
• ENVHOGs — 315k representative proteins from the 2.2M ENVHOGs that have a PHROG function that is not ‘unknown function’. Proteins over 1000AA were chunked into 1000AA components.
• EFAM - 262k efam “extra conservative” proteins. Proteins over 1000AA were chunked into 1000AA components.
• DGRs - 12683 extra diversity generating element proteins from Roux et al 2021.
• VFDB — over 28k structures of putative bacterial virulence factors from the VFDB.
• CARD — nearly 5k structures of anitbiotic resistant proteins from CARD.
• acrDB - nearly 3.7k anti-crispr proteins predicted in this study.
• Defensefinder - 455 monomer prokaryotic defense proteins.
• Netflax - 7153 toxin-antitoxin proteins.

Google Colab Notebook

If you don’t want to install phold locally, you can run it without any code using one this Google Colab notebook