Idp go #593

csbrasnett · 2024-04-29T08:14:59Z

Adding -idp-tune to allow for addition of extra bonded potentials in IDRs

with the extra flag -idp-tune (probably a better name) martinize will add additional potentials as per the Go paper. This commit: 1) finds cross Go interactions between folded and idrs and removes them 2) identifies BB and SC1 beads in idrs

# Conflicts: # vermouth/data/force_fields/martini3001/citations.bib # vermouth/processors/water_bias.py # vermouth/tests/data/integration_tests/tier-1/hst5/martinize2/citation # vermouth/tests/data/integration_tests/tier-1/hst5/martinize2/command # vermouth/tests/data/integration_tests/tier-1/hst5/martinize2/molecule_0.itp # vermouth/tests/data/integration_tests/tier-1/hst5/martinize2/virtual_sites_nonbond_params.itp # vermouth/tests/integration_tests/test_integration.py

csbrasnett · 2024-04-29T09:27:05Z

bin/martinize2

+    # apply idr specific bonded potentials if required
+    if args.idr_tune:
+        if args.to_ff == 'martini3001':


I would ideally like a better way of ensuring these parameters get used when needed and ignored when they shouldn't be used, but this is my best idea for now

pckroon

Some small code comments, and a conceptual suggestion.
You want to add some interactions to specific disordered regions of a protein, right? My suggestion would be to have a simple processor annotate nodes/residues that are in an IDR, and add the interactions using links

vermouth/processors/tune_idp_bonds.py

pckroon · 2024-05-01T13:40:09Z

vermouth/processors/tune_idp_bonds.py

+                                                          "version": 1}
+                                                    )
+
+    def run_molecule(self, molecule):


Don't forget to attach the appropriate citation reference to the molecule

vermouth/processors/tune_idp_bonds.py

- processors/tune_idp_bonds.py now annotates residues which are disordered so that they can have their links added afterwards. Uses existing function from dssp for this purpose. - martinize2 edited to reflect new pipeline - removed now not needed function from rcsu/go_utils.py - addressed other comments

csbrasnett · 2024-05-02T11:26:47Z

This is a much nicer solution, thanks for suggesting!

pckroon

Much cleaner! <3
Still a few small comments

pckroon · 2024-05-06T08:27:32Z

bin/martinize2

-            raise argparse.ArgumentError(idr_tuning,
-                                         message=("You are trying to apply IDR biasing to a force field which "
-                                                  "isn't martini3001, which this is not designed for."))
+            vermouth.IDRBonds(idr_regions = [(int(start), int(stop)) for start, stop in args.water_idrs]).run_system(system)


Suggested change

vermouth.IDRBonds(idr_regions = [(int(start), int(stop)) for start, stop in args.water_idrs]).run_system(system)

vermouth.IDRBonds(idr_regions=[(int(start), int(stop)) for start, stop in args.water_idrs]).run_system(system)

pckroon · 2024-05-06T08:28:13Z

bin/martinize2

-                                         message=("You are trying to apply IDR biasing to a force field which "
-                                                  "isn't martini3001, which this is not designed for."))
+            vermouth.IDRBonds(idr_regions = [(int(start), int(stop)) for start, stop in args.water_idrs]).run_system(system)
+            vermouth.DoLinks().run_system(system)


I don't think you actually want to run DoLinks a second time. Instead, move the IDRBonds annotator to just after the mapping

pckroon · 2024-05-06T08:30:27Z

vermouth/processors/tune_idp_bonds.py

+    order_disorder_seq = []
+
+    for key, node in molecule.nodes.items():
+        if select_backbone(node):


I wouldn't tag just the backbone (or SC1), but rather just all nodes with the appropriate resid

pckroon · 2024-05-06T08:31:41Z

vermouth/processors/tune_idp_bonds.py

+        if select_backbone(node):
+            _old_resid = node['_old_resid']
+            if _in_resid_region(_old_resid, idr_regions):
+                order_disorder_seq.append("D")  # D for disordered


Why first build the list/string, and not just do molecule.nodes[key]['is_idr'] = True here?

- changed processor function to annotate IDRs first before applying links in martinize. - moved cross go potential removal to water_bias.py - changed aminoacids.ff to reflect new naming - removed no longer needed import in dssp/__init__.py - changed name of import in processors/__init__.py

pckroon · 2024-05-07T08:49:53Z

bin/martinize2

@@ -170,7 +170,7 @@ def pdb_to_universal(
    return canonicalized


-def martinize(system, mappings, to_ff, delete_unknown=False):
+def martinize(system, mappings, to_ff, delete_unknown=False, idrs=False, disordered_regions = []):


Suggested change

def martinize(system, mappings, to_ff, delete_unknown=False, idrs=False, disordered_regions = []):

def martinize(system, mappings, to_ff, delete_unknown=False, idrs=False, disordered_regions=(,)):

As before, no mutable types as default arguments

pckroon · 2024-05-07T08:51:25Z

bin/martinize2

+    #
+    # # apply idr specific bonded potentials if required
+    # if args.idr_tune:
+    #     if not target_ff.has_feature("idr"):
+    #         LOGGER.warning('Improved IDR potentials are not implemented '
+    #                        'for this force field',
+    #                        type="missing-feature")
+    #     else:
+    #         vermouth.IDRBonds(idr_regions=[(int(start), int(stop)) for start, stop in args.water_idrs]).run_system(system)
+    #         vermouth.DoLinks().run_system(system)


Suggested change

#

# # apply idr specific bonded potentials if required

# if args.idr_tune:

# if not target_ff.has_feature("idr"):

# LOGGER.warning('Improved IDR potentials are not implemented '

# 'for this force field',

# type="missing-feature")

# else:

# vermouth.IDRBonds(idr_regions=[(int(start), int(stop)) for start, stop in args.water_idrs]).run_system(system)

# vermouth.DoLinks().run_system(system)

pckroon · 2024-05-07T08:52:35Z

vermouth/processors/annotate_idrs.py

+def annotate_disorder(molecule, idr_regions):
+    """
+    annotate the disordered regions of the molecule
+    """


Docstring is incomplete.
Also, is it worth making the attribute used an argument (with default "cgidr")?

pckroon · 2024-05-07T08:52:54Z

vermouth/processors/annotate_idrs.py

+
+    """
+
+    def __init__(self, idr_regions = None):


Suggested change

def __init__(self, idr_regions = None):

def __init__(self, idr_regions=None):

pckroon · 2024-05-07T08:53:56Z

vermouth/processors/annotate_idrs.py

+        """
+        if not self.idr_regions:
+            return system
+        self.system = system


Suggested change

self.system = system

Why do you need this?

- changed default argument in martinize - renamed 'idr_regions' to 'id_regions' throughout - exposed the name of the extra annotation in annotate_disorder with default argument "cgidr"

csbrasnett · 2024-05-22T09:30:34Z

btw, in trying to write some tests to address the missing coverage I've found an error in the test here:

vermouth-martinize/vermouth/tests/test_water_bias.py

Lines 89 to 98 in e9bf1c9

    
           for nb_params in system.gmx_topology_params['nonbond_params']: 
        
               assert nb_params.atoms[0] == "W" 
        
               vs_node = nb_params.atoms[1] 
        
               for vs in system.molecules[0].interactions['virtual_sitesn']: 
        
                   if vs.atoms[0] == vs_node: 
        
                       bb_node = vs.atoms[1] 
        
                       water_eps = water_bias[expected[bb_node]] 
        
                       water_sig = sizes[bb_node] 
        
                       assert water_eps == nb_params.epsilon 
        
                       assert water_sig == nb_params.sigma

in that the if statement is testing an atomtype against an atom index. Easy enough to fix, but it does now cause this test to fail once the the eps and sig assertions are actually reached! The failure is unrelated to the new removal functionality in water_bias as well. I'll fix here and you can have a look

pckroon

LGTM! Was there anything still missing in this PR?

csbrasnett · 2024-05-22T11:53:22Z

Thanks! Making a test for the folded/disordered go bond removal, should be done soon!

pckroon

How much work would it be to add a test that removes some but not all go-bonds?

csbrasnett · 2024-05-22T12:24:06Z

you mean in addition to what I've done here now?

pckroon · 2024-05-22T12:31:22Z

Yes :)

csbrasnett · 2024-05-22T14:17:44Z

This is proving more difficult than anticipated... I'm guessing in general you have something in mind like this. Ie. you have two folded domains connected by some linker, and you want to 1) keep intra-domain go bonds, 2) remove domain-linker bonds, and 3) remove inter-domain bonds? I this should be possible but I need to make some big edits to the removal functionality

pckroon · 2024-05-22T14:20:49Z

What I mean is that in test_water_bias.py:test_cross_go_bond_removal you either remove all or none. I'm not sure that's going to be the same case as you describe tbh

- remove all bonds - remove some

csbrasnett · 2024-05-22T14:39:08Z

ok, I think that addresses your point?

pckroon

Yes, perfect

csbrasnett and others added 8 commits April 23, 2024 15:14

fix missing bias application, add preprint citation

7ce1254

Merge branch 'master' into idp-bugfix

7349c14

added integration test

51c284d

added more detail to -ss help

0aaf858

add idr interactions

359cd77

cleaned up and fixed

46e8473

csbrasnett requested a review from fgrunewald April 29, 2024 08:14

Merge branch 'master' into idp-go

67884de

csbrasnett commented Apr 29, 2024

View reviewed changes

Merge branch 'marrink-lab:master' into idp-go

d2a528e

pckroon requested changes May 1, 2024

View reviewed changes

csbrasnett added 3 commits May 2, 2024 13:14

added IDR links to force field

4399d7e

updated tests with new bonded potentials

d629b7b

improved error handling if no idp-tune

735accb

pckroon requested changes May 6, 2024

View reviewed changes

csbrasnett added 3 commits May 6, 2024 17:06

removed old commented code

9b646d1

removed unnecessary imports, added test for idr annotation

b62c3b2

pckroon requested changes May 7, 2024

View reviewed changes

csbrasnett added 2 commits May 7, 2024 11:26

addressed comments

8d75c3c

- changed default argument in martinize - renamed 'idr_regions' to 'id_regions' throughout - exposed the name of the extra annotation in annotate_disorder with default argument "cgidr"

corrected to new argument name in test_annotate_idrs.py

b3179b4

fixed test for assigning residue water bias

72b9d4e

pckroon previously approved these changes May 22, 2024

View reviewed changes

added test to check folded-disordered go bond removal

a801775

csbrasnett dismissed pckroon’s stale review via a801775 May 22, 2024 12:15

pckroon reviewed May 22, 2024

View reviewed changes

cleaned up test

635c4a2

changed test_annotate_idrs.py to test system not molecule

cd0d5bd

added more tests to test_w

5fd9e10

- remove all bonds - remove some

pckroon approved these changes May 22, 2024

View reviewed changes

pckroon merged commit 7e59d59 into marrink-lab:master May 22, 2024
8 checks passed

csbrasnett deleted the idp-go branch May 22, 2024 15:05

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Idp go #593

Idp go #593

csbrasnett commented Apr 29, 2024

csbrasnett Apr 29, 2024

pckroon left a comment

pckroon May 1, 2024

csbrasnett commented May 2, 2024

pckroon left a comment

pckroon May 6, 2024

pckroon May 6, 2024

pckroon May 6, 2024

pckroon May 6, 2024

pckroon May 7, 2024

pckroon May 7, 2024

pckroon May 7, 2024

pckroon May 7, 2024

pckroon May 7, 2024

csbrasnett commented May 22, 2024 •

edited

Loading

pckroon left a comment

csbrasnett commented May 22, 2024

pckroon left a comment

csbrasnett commented May 22, 2024

pckroon commented May 22, 2024

csbrasnett commented May 22, 2024

pckroon commented May 22, 2024

csbrasnett commented May 22, 2024

pckroon left a comment

	vermouth.IDRBonds(idr_regions = [(int(start), int(stop)) for start, stop in args.water_idrs]).run_system(system)
	vermouth.IDRBonds(idr_regions=[(int(start), int(stop)) for start, stop in args.water_idrs]).run_system(system)

	def martinize(system, mappings, to_ff, delete_unknown=False, idrs=False, disordered_regions = []):
	def martinize(system, mappings, to_ff, delete_unknown=False, idrs=False, disordered_regions=(,)):

	def __init__(self, idr_regions = None):
	def __init__(self, idr_regions=None):

Idp go #593

Idp go #593

Conversation

csbrasnett commented Apr 29, 2024

Choose a reason for hiding this comment

pckroon left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

csbrasnett commented May 2, 2024

pckroon left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

csbrasnett commented May 22, 2024 • edited Loading

pckroon left a comment

Choose a reason for hiding this comment

csbrasnett commented May 22, 2024

pckroon left a comment

Choose a reason for hiding this comment

csbrasnett commented May 22, 2024

pckroon commented May 22, 2024

csbrasnett commented May 22, 2024

pckroon commented May 22, 2024

csbrasnett commented May 22, 2024

pckroon left a comment

Choose a reason for hiding this comment

csbrasnett commented May 22, 2024 •

edited

Loading