Joint Shantou International Eye Centre of Shantou University and The Chinese University of Hong Kong


To facilitate the biology investigations and structure-based drug development for ocular diseases, we aimed to establish a protein structure database for ocular diseases. Based on the source code and trained models of AlpbhaFold2 and RoseTTAFold, we built a protein structure prediction platform for the protein structure prediction of wildtype and variant amino acid sequences for the ocular diseases. To date, 745 genes with phenotypes and Online Mendelian Inheritance in Man (OMIM) numbers have been collected, and 1684 amino acid sequences (with the length of 11 - 2843 residues) of both wildtype and variants ophthalmic genes have been predicted. These predicted genes were summarized according to the ocular tissues, the type of diseases, including strabismus, lens, vitreous, ocular deformity, eyelid, neuro-ophthalmology, choroid, iris, retina, cornea, myopia, and glaucoma.

User tutorials

The predicted structures were listed in alphabetical orders according to the genes or the phenotypes. Users could search by the gene/protein name or phenotype. Each protein has a prediction-list page that showed the sequence length, pLDDT score and B-factor for each of the listed wildtype and its mutants. Each wildtype/mutant has a dedicated structure page linked to OMIM for protein sequence with length. Five predictions of the AlphaFold model and the RoseTTAFold model were listed by the pLDDT score and estimated CA rms error at the B-factor column respectively. Tertiary protein structures of each prediction could be visualized in a full-screen view with a built-in 3D Pdb viewer, with pLDDT or B-factor. Customized service for protein structure prediction of specific genes could be obtained on a request page. By providing protein sequence with gene information and contact information, users could submit a service request to our platform and obtained a respond within a few days.