colab removes last 2 amino acids

[HWaymentSteele]

Hi,

The RGN2 colab notebook removes the last 2 residues of the input sequence. The input sequence in run.json still contains the last 2 residues, but the fulinit.pdb is missing them.

I would like to benchmark RGN2 in comparison to other methods, and so I could hack a fix by adding two alanines or something, but I am not sure that will faithfully represent RGN2's embedding. Thanks!

[christinaflo]

Hi, this is actually expected behavior. Due to the Frenet-Serret framing strategy, in order to construct a frame at a given residue, two residues following it are needed to define the torsion and curvature angles required to create the rotation matrix. Therefore, we exclude the last two residues where we cannot construct a frame.