- Retrained the predictor with newly available PDB data and exactly the same procedures as in the original paper (https://doi.org/10.1038/s41598-019-43189-4).
- Added handling of sequences with non-standard amino acids (jointly denoted as 'X' residues).
- Changes did not affect the performance of our method.