search

ReliaSolve / cctbx_project

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

Reduce2 should have a neighbor length of 4 #254

Closed russell-taylor closed 1 year ago

russell-taylor commented 1 year ago

To make it compatible with Probe, and for the correct expected operation, Reduce2 should have a neighbor length of 4 (which only applies when one of the endpoints is a hydrogen) rather than 3.

Note in the command-line argument description that changing this should only be done for regression-testing purposes.

For regression testing code against original Reduce, we should specify a value of 3 so that we're comparing apples to apples. Specifying 4 in original Reduce doesn't include the "4 only if hydrogen" check.

russell-taylor commented 1 year ago

@todo: Ensure that we use a non-hydrogen length of 3.

russell-taylor commented 1 year ago

When regression testing against the standard set of files and fragments for original Reduce, this did not make the Probe2 score of Reduce2 worse than the score for Reduce in any additional models.