ReliaSolve / cctbx_project

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

Undowser2 #286

Closed russell-taylor closed 3 months ago

russell-taylor commented 5 months ago

Make an undowser2 that uses probe2 https://github.com/cctbx/cctbx_project/blob/master/mmtbx/programs/undowser.py https://github.com/cctbx/cctbx_project/blob/master/mmtbx/validation/undowser.py https://github.com/cctbx/cctbx_project/blob/master/mmtbx/validation/regression/tst_undowser.py

russell-taylor commented 4 months ago

@chrissciwilliams what are good PDB files to run on to test whether we're getting the same water collisions in undowser2 as undowser?

russell-taylor commented 4 months ago

(done) Test on both single-model and multi-model files. Probably 1LPL, according to tst_undowser.py

The tst_undowser.py fails, and the program returns no clashes for 1LPL. Asked Christopher and Jane about this. Running in 1LPL.cif under Linux does show collisions, but tst_undowser.py still fails (maybe it can't find probe?).

The phenix.undowser_validation program also returns nothing for 1ssx on Windows, but it returns a number of clashes on Linux (last update Jan 29), also when rebuilt in the undowser2 branch. This is a file that Christopher thought should exhibit all of the types of clashes.

Running on 1LPL produces 6 waters of 86 with clashes in undowser and 7 on undowser2.

russell-taylor commented 4 months ago

(done) Make a regression test for it once it is working and add it to the list of those that are run.

chrissciwilliams commented 4 months ago

1ssx is a rich source of water clashes - it populates all of UnDowser's current categories. It's probably the most useful single file I have on hand.

Some other useful files: 1gwe has many clashes with alternate-conformation waters 1eb6 has lots of water-water clashes 1LpL should work, and is a simpler file with no alternates-conformation/water interactions.

-Christopher Williams ---Richardson Lab, Duke University

On Mon, Apr 22, 2024 at 2:36 PM Russell Taylor @.***> wrote:

@chrissciwilliams https://github.com/chrissciwilliams what are good PDB files to run on to test whether we're getting the same water collisions in undowser2 as undowser?

— Reply to this email directly, view it on GitHub https://github.com/ReliaSolve/cctbx_project/issues/286#issuecomment-2070582501, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACLEREBC6NW36MUGIZCMLQ3Y6VKB3AVCNFSM6AAAAABF6SBQU6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDANZQGU4DENJQGE . You are receiving this because you were mentioned.Message ID: @.***>

russell-taylor commented 4 months ago

(done) When we try to run on 1SSX, it has a Sorry with a Hydrogen with 3 bonds on it. Modified Probe2 to allow multiple bonds for a hydrogen, which can happen when we have screwed up alternate conformations.

russell-taylor commented 4 months ago

Added in the undowser2 branch, which has been pull requested into master.