Swap sidechain occasionally yields incorrect geometry

[Tristan Croll]

If I load up 5f4y with added hydrogens (not sure if that's relevant), select the "swapaa" right mouse mode and start rapidly dragging up and down on residue /A:1593 (or any residue in a helix), the geometry will fairly quickly break. The amide hydrogen starts getting misplaced almost immediately (particularly on passing through proline - for some reason on alanine it disappears entirely) and the direction of the CA-CB bond starts varying more and more dramatically (often ending up pointing into the helix). Continuing to scrub in this manner will eventually trigger it to stop adding hydrogens at all. The serious deviations in sidechain geometry only *seem* to show up after many iterations (and so may not be a problem in most cases), but it does point to something funny happening.

[Tom Goddard]

Fixed.

The bizarre behavior where the added residue position would change as you cycled repeatedly through residues was caused by the new residue alignment using atom positions from the previously placed residue. Now it only uses N, CA, C positions for the alignment and those are not moved when adding new side chains (nor are O and H moved). Side chain orientation is much better now. Alanine missing the H atom is fixed (template did not have it), and code that looked for whether to add hydrogens was fixed (was checking for presence of a hydrogen after stripping of side chain).