Structural and functional annotation of Glaciozyma antarctica hypothetical proteins using integrated modeling pipelines.

Glaciozyma antarctica is a psychrophilic yeast that can withstand extreme temperatures by adapting several mechanisms, such as preventing the formation of ice around the cell, maintaining the flexibility of the membrane, preventing the formation of reactive oxygen species, and producing cold-adapted enzymes. The G. antarctica genome encodes 7857 protein sequences, however, 2911 are hypothetical with no annotated functions and only five proteins have been experimentally characterised with structure coordinates in the Protein Data Bank (PDB). In this work, we modelled the of G. antarctica hypothetical proteins (HPs) using the FunMod pipeline and ColabFold, and carried out structural characterisation of the resulting models. We have successfully modelled 1517 (52%) of the hypothetical proteins with good quality structures with an average MolProbity score of 2.6. By integrating annotations from both sequence-based (InterProScan and CATH-Gene3D Functional Family), structure-based (Foldseek) methods, and removing overlapping annotated protein IDs, we predicted functions or domains to a total of 1866 (64%) unique HPs. From this annotated set, 814 proteins (44%) were predicted as probable enzymes, based on keyword parsing of their assigned functions (e.g., ‘enzyme’ or ‘-ase’). The remaining 1052 proteins (56%) were assigned non-enzymatic functions. Notably, several enzymes, including lipases and cellulases, have potential industrial applications such as detergent formulation, while increasing our understanding of adaptation mechanisms in cold temperatures. All generated data is available in the G. antarctica protein structure database as a resource of information for future research on psychrophilic organisms, especially G. antarctica.