cctbx_project icon indicating copy to clipboard operation
cctbx_project copied to clipboard

Update __init__.py

Open tomburnley opened this issue 3 years ago • 7 comments

Update ensemble_refinement to 2020 version. Added DEN restraints (now used by default). Fixed refinement flag manager. Added support for importing external TLS parameters from additional PDB.

tomburnley avatar Sep 16 '20 16:09 tomburnley

Tom

Can you tell me why DEN is default?

Do you have any tests for these changes? Cheers

Nigel


Nigel W. Moriarty Building 33R0349, Molecular Biophysics and Integrated Bioimaging Lawrence Berkeley National Laboratory Berkeley, CA 94720-8235 Phone : 510-486-5709 Email : [email protected] Fax : 510-486-5909 Web : CCI.LBL.gov

On Wed, Sep 16, 2020 at 9:09 AM Tom Burnley [email protected] wrote:

@tomburnley https://github.com/tomburnley requested your review on: #528 https://github.com/cctbx/cctbx_project/pull/528 Update init.py.

— You are receiving this because your review was requested. Reply to this email directly, view it on GitHub https://github.com/cctbx/cctbx_project/pull/528#event-3773994548, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAHYQPDIXHD5S4EZ74YCT3DSGDPLDANCNFSM4RPB7OWQ .

nwmoriarty avatar Sep 16 '20 16:09 nwmoriarty

Hi Nigel,

DEN -> we found in testing that it makes the simulations more stable without negatively affecting sampling. W/out DEN on some occasions parts of the structure can unfold out of density. The attractive terms in Amber may alleviate this but for now w/ the repel potential it's better. The DEN pairs are regenerated multiple times during the simulation so sampling does not seem to be compromised.

Testing -> yes... see separate email.

Cheers,

Tom

On Wed, 16 Sep 2020 at 17:15, Nigel W. Moriarty [email protected] wrote:

Tom

Can you tell me why DEN is default?

Do you have any tests for these changes? Cheers

Nigel


Nigel W. Moriarty Building 33R0349, Molecular Biophysics and Integrated Bioimaging Lawrence Berkeley National Laboratory Berkeley, CA 94720-8235 Phone : 510-486-5709 Email : [email protected] Fax : 510-486-5909 Web : CCI.LBL.gov

On Wed, Sep 16, 2020 at 9:09 AM Tom Burnley [email protected] wrote:

@tomburnley https://github.com/tomburnley requested your review on: #528 https://github.com/cctbx/cctbx_project/pull/528 Update init.py.

— You are receiving this because your review was requested. Reply to this email directly, view it on GitHub https://github.com/cctbx/cctbx_project/pull/528#event-3773994548, or unsubscribe < https://github.com/notifications/unsubscribe-auth/AAHYQPDIXHD5S4EZ74YCT3DSGDPLDANCNFSM4RPB7OWQ

.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/cctbx/cctbx_project/pull/528#issuecomment-693511586, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACSF77V6VBGGQ2ZLVTQSFHLSGDQBXANCNFSM4RPB7OWQ .

tomburnley avatar Sep 16 '20 16:09 tomburnley

There are unused imports causing the Checks tests to fail.

In file __init__.py:
time imported at line 34
miller imported at line 19
geometry_restraints imported at line 17
cdl_utils imported at line 636
xray imported at line 21

An update will cause the CI to run again and the error for Full and Base should go away since I checked in some updates to the master branch.

bkpoon avatar Sep 20 '20 00:09 bkpoon

Thanks! The Ci branch failures are fine. That pipeline does not run on the final, merged code with the updates to master.

bkpoon avatar Sep 22 '20 16:09 bkpoon

This requested change seems to be missed:

please use iotbx.pdb.input() instead of iotbx.pdb.hierarchy.input(), preferably checking if the file exists.

olegsobolev avatar Sep 22 '20 16:09 olegsobolev

This requested change seems to be missed:

please use iotbx.pdb.input() instead of iotbx.pdb.hierarchy.input(), preferably checking if the file exists.

Can you explain the need for this and point to an example of use I can copy from please.

tomburnley avatar Sep 23 '20 12:09 tomburnley

@tomburnley , iotbx.pdb.hierarchy.input() seems to be a very thin wrapper that pretty much constructs hierarchy and keep input object: https://github.com/cctbx/cctbx_project/blob/1be58fdfe34797a7126eb9ab22c5805326c4a154/iotbx/pdb/hierarchy.py#L2493-L2536

The original reason to create an extra layer for such a simple task is not clear to me in the first place. I could very well miss something, so clarifications from anyone are welcomed. As of now, I believe this whole layer better be removed, therefore advocate against using it anywhere.

In practical terms regarding this pull request I suggest to change this:

pdb_tls_inp = iotbx.pdb.hierarchy.input(file_name=pdb_import_tls)
tls_params = pdb_tls_inp.input.extract_tls_params(pdb_tls_inp.hierarchy)

to this:

pdb_tls_inp = iotbx.pdb.input(file_name=pdb_import_tls)
tls_params = pdb_tls_inp.extract_tls_params(pdb_tls_inp.construct_hierarchy())

Other examples of iotbx.pdb.input() usage can be found e.g. in iotbx/command_line/pdb_labels_comparison.py and many other files across the repository.

olegsobolev avatar Sep 23 '20 17:09 olegsobolev