search

ReliaSolve / cctbx_project

Computational Crystallography Toolbox
https://cctbx.github.io
Other
0 stars 0 forks source link

Large differences in Probe vs. Probe2 for 1rrr.pdb #199

Closed russell-taylor closed 2 years ago

russell-taylor commented 2 years ago

(done) Large summary differences: no H contacts in Probe2 but many in Probe. (done) Large kinemage differences: Bright contacts in Probe not present in Probe2.

russell-taylor commented 2 years ago

This difference appears for each individual model.

russell-taylor commented 2 years ago

There are two Hydrogen atoms that the original Probe marks as not being bonded to anything, which is why we're seeing the entries for Hydrogens in the original code. They are stealing from their bonded neighbor's results.

The Probe2 bonding is better than the original, so we'll go with its behavior.

russell-taylor commented 2 years ago

(done) The contact and H-bonded scores are similar between the two runs for the first model. The overlap scores are very different (more than twice as large negative in Probe). There are around the same number of C small_overlaps (a bit fewer in Probe2). There are the same number of N small_overlaps. For Probe2, there are only 189 O small_overlaps and no other types; Probe has 279 small, 18 bad, and 46 worse overlaps.

There are two sets of these, each between an O and its corresponding H (which is marked as not bonded). They cause a lot of big clashes, in both directions. This is the major difference between Probe and Probe2.

The Probe2 bonding is better than the original, so we'll go with its behavior.