Pipeline PTM_neighborhood calculations may be too distance-tolerant

[ChrisMoth]

Alican's scripts used an 8A cutoff to enumerate predicted PTMs predicted from the sequeence, and mapped onto the alphafold models.

His code used Rosetta library routines. Chris' code uses biopython.

As a result, there are small differences in candidate residue sets, with Chris's code tending to be more permissive.

This needs further exploriation

[ChrisMoth]

WHEN PTM-predictions are found close to a VUS, the output should include the closest atom pairs between the residues, as audit trail of selection.