Improve H placements vs. Reduce in 3boy

[russell-taylor]

From email on 7/11: Here's the image with the very wrong-answer H placements, in 3boy between the U7 2' OH of chain D that is totally out of density and A56 Thr OH (which is definitely correct for the heavy atoms, with good density overall and much stronger for this O branch). This is a pathological case of an RNA pucker outlier, but we should scream the problem, not invent impossible H-bonds. See if you can tell any better than I could what is actually going wrong with the Hs, which should just get out of the way as far as possible. The other 2 approximately symmetric cases (U14 & U21) are also fit badly but don't put the 2' as close to the Thr.

[russell-taylor]

Ran mmtbx.reduce2 add_flip_movers=True D:\data\Richardsons\3boy.pdb output.overwrite=True output.description_file_name=3boy.txt output.file_name=3boyFH.pdb to place the hydrogens so we could take a look at how it handled them.

It produces the same, or very similar, results:

Ran mmtbx.probe2 output.add_kinemage_keyword=True 3boyFH.pdb and got similar overlaps:

[russell-taylor]

This appears to be another instance that could benefit from single-hydrogen rotators having a "removed" coarse state with no corresponding fine states, as described in https://github.com/ReliaSolve/cctbx_project/issues/262

Nope -- the hydrogens should be getting out of each others' way, as Jane described.

[russell-taylor]

It looks like the Hydrogens are 0.84 units away from the Oxygens. The clashes are 0.762 or smaller distance from the other hydrogens; the nearest H-bond dot appears to be around 0.776. The dots are always 1.05 units away from the Hydrogens, so that must be their radius.

[russell-taylor]

Trimming down to just the two oxygens and hydrogens makes the situation more clear (HG1 on the bottom):

• A set of one-sided collisions near the poles come from the hydrogens, seemingly with two different lengths.
• A set of one-sided collisions from the hydrogens appear around front and sides: HG1 on the left and HO2* on the right.
• A band of O-O collisions appear around the back side of the plane bisecting them.
• Sets of unexpected one-sided hydrogen bonds (from HG1 on the top and HO2* on the bottom). The ones on the upper right are closer to the hydrogen than to the oxygen, so why are the hydrogen contacts not being shown? The larger radius of the Oxygen means that the dots are further inside the O than they are inside the H (smaller "gap"), so the O wins and its contact is selected.
• A thin line of O2' hydrogen bonds and OG1 hydrogen bonds forms along the separating plane.