Merge pull request #975 from haddocking/flexseg-onthefly-nfle

Simplify definition of counted parameters by computing them on the fly from definition.

examples/docking-protein-glycan/docking-flexref-protein-glycan-full.cfg

@@ -42,8 +42,7 @@ mdsteps_cool1 = 5000
 mdsteps_cool2 = 10000
 mdsteps_cool3 = 10000
 ambig_fname = "data/ambig.tbl"
-# give full flexibilit to the glycan
-nfle = 1
+# give full flexibility to the glycan
 fle_sta_1 = 1
 fle_end_1 = 4
 fle_seg_1 = "B"

examples/docking-protein-glycan/docking-flexref-protein-glycan-test.cfg

@@ -38,8 +38,7 @@ mdsteps_cool1 = 5000
 mdsteps_cool2 = 10000
 mdsteps_cool3 = 10000
 ambig_fname = "data/ambig.tbl"
-# give full flexibilit to the glycan
-nfle = 1
+# give full flexibility to the glycan
 fle_sta_1 = 1
 fle_end_1 = 4
 fle_seg_1 = "B"

examples/docking-protein-homotrimer/docking-protein-homotrimer-full.cfg

@@ -41,7 +41,6 @@ ambig_fname = "data/1qu9_whiscy_air.tbl"
 sampling = 2000
 # Define CNS restraints between molecules
 ncs_on = true
-numncs = 2
 ncs_sta1_1=2
 ncs_end1_1=128
 ncs_seg1_1="A"
@@ -56,7 +55,6 @@ ncs_end2_2=128
 ncs_seg2_2="C"
 # Define C3 symmetry restraints
 sym_on = true
-numc3sym = 1
 c3sym_sta1_1=2
 c3sym_end1_1=128
 c3sym_seg1_1="A"
@@ -79,7 +77,6 @@ tolerance = 5
 ambig_fname = "data/1qu9_whiscy_air.tbl"
 # Define NCS restraints between molecules
 ncs_on = true
-numncs = 2
 ncs_sta1_1=2
 ncs_end1_1=128
 ncs_seg1_1="A"
@@ -94,7 +91,6 @@ ncs_end2_2=128
 ncs_seg2_2="C"
 # Define C3 symmetry restraints
 sym_on = true
-numc3sym = 1
 c3sym_sta1_1=2
 c3sym_end1_1=128
 c3sym_seg1_1="A"
@@ -113,7 +109,6 @@ tolerance = 5
 ambig_fname = "data/1qu9_whiscy_air.tbl"
 # Define NCS restraints between molecules
 ncs_on = true
-numncs = 2
 ncs_sta1_1=2
 ncs_end1_1=128
 ncs_seg1_1="A"
@@ -128,7 +123,6 @@ ncs_end2_2=128
 ncs_seg2_2="C"
 # Define C3 symmetry restraints
 sym_on = true
-numc3sym = 1
 c3sym_sta1_1=2
 c3sym_end1_1=128
 c3sym_seg1_1="A"

examples/docking-protein-homotrimer/docking-protein-homotrimer-test.cfg

@@ -36,7 +36,6 @@ ambig_fname = "data/1qu9_whiscy_air.tbl"
 sampling = 20
 # Define NCS restraints between molecules
 ncs_on = true
-numncs = 2
 ncs_sta1_1=2
 ncs_end1_1=128
 ncs_seg1_1="A"
@@ -51,7 +50,6 @@ ncs_end2_2=128
 ncs_seg2_2="C"
 # Define C3 symmetry restraints
 sym_on = true
-numc3sym = 1
 c3sym_sta1_1=2
 c3sym_end1_1=128
 c3sym_seg1_1="A"
@@ -74,7 +72,6 @@ tolerance = 20
 ambig_fname = "data/1qu9_whiscy_air.tbl"
 # Define NCS restraints between molecules
 ncs_on = true
-numncs = 2
 ncs_sta1_1=2
 ncs_end1_1=128
 ncs_seg1_1="A"
@@ -89,7 +86,6 @@ ncs_end2_2=128
 ncs_seg2_2="C"
 # Define C3 symmetry restraints
 sym_on = true
-numc3sym = 1
 c3sym_sta1_1=2
 c3sym_end1_1=128
 c3sym_seg1_1="A"
@@ -108,7 +104,6 @@ tolerance = 20
 ambig_fname = "data/1qu9_whiscy_air.tbl"
 # Define NCS restraints between molecules
 ncs_on = true
-numncs = 2
 ncs_sta1_1=2
 ncs_end1_1=128
 ncs_seg1_1="A"
@@ -123,7 +118,6 @@ ncs_end2_2=128
 ncs_seg2_2="C"
 # Define C3 symmetry restraints
 sym_on = true
-numc3sym = 1
 c3sym_sta1_1=2
 c3sym_end1_1=128
 c3sym_seg1_1="A"

examples/docking-protein-peptide/docking-protein-peptide-cltsel-full.cfg

@@ -51,7 +51,6 @@ reference_fname = "data/1nx1_refe.pdb"
 tolerance = 5
 ambig_fname = "data/ambig.tbl"
 # Define peptide as fully flexible
-nfle = 1
 fle_sta_1 = 1
 fle_end_1 = 11
 fle_seg_1 = "B"
@@ -72,7 +71,6 @@ reference_fname = "data/1nx1_refe.pdb"
 tolerance = 5
 ambig_fname = "data/ambig.tbl"
 # Define peptide as fully flexible
-nfle = 1
 fle_sta_1 = 1
 fle_end_1 = 11
 fle_seg_1 = "B"

examples/docking-protein-peptide/docking-protein-peptide-full.cfg

@@ -45,7 +45,6 @@ select = 400
 tolerance = 5
 ambig_fname = "data/ambig.tbl"
 # Define peptide as fully flexible
-nfle = 1
 fle_sta_1 = 1
 fle_end_1 = 11
 fle_seg_1 = "B"
@@ -65,7 +64,6 @@ reference_fname = "data/1nx1_refe.pdb"
 tolerance = 5
 ambig_fname = "data/ambig.tbl"
 # Define peptide as fully flexible
-nfle = 1
 fle_sta_1 = 1
 fle_end_1 = 11
 fle_seg_1 = "B"

examples/docking-protein-peptide/docking-protein-peptide-mdref-full.cfg

@@ -45,7 +45,6 @@ select = 400
 tolerance = 5
 ambig_fname = "data/ambig.tbl"
 # Define peptide as fully flexible
-nfle = 1
 fle_sta_1 = 1
 fle_end_1 = 11
 fle_seg_1 = "B"
@@ -65,7 +64,6 @@ reference_fname = "data/1nx1_refe.pdb"
 tolerance = 5
 ambig_fname = "data/ambig.tbl"
 # Define peptide as fully flexible
-nfle = 1
 fle_sta_1 = 1
 fle_end_1 = 11
 fle_seg_1 = "B"

examples/docking-protein-peptide/docking-protein-peptide-mdref-test.cfg

@@ -39,7 +39,6 @@ select = 5
 tolerance = 20
 ambig_fname = "data/ambig.tbl"
 # Define peptide as fully flexible
-nfle = 1
 fle_sta_1 = 1
 fle_end_1 = 11
 fle_seg_1 = "B"
@@ -60,7 +59,6 @@ reference_fname = "data/1nx1_refe.pdb"
 tolerance = 20
 ambig_fname = "data/ambig.tbl"
 # Define peptide as fully flexible
-nfle = 1
 fle_sta_1 = 1
 fle_end_1 = 11
 fle_seg_1 = "B"

examples/docking-protein-peptide/docking-protein-peptide-test.cfg

@@ -39,7 +39,6 @@ select = 5
 tolerance = 20
 ambig_fname = "data/ambig.tbl"
 # Define peptide as fully flexible
-nfle = 2
 fle_sta_1 = 1
 fle_end_1 = 5
 fle_seg_1 = "B"
@@ -63,7 +62,6 @@ reference_fname = "data/1nx1_refe.pdb"
 tolerance = 20
 ambig_fname = "data/ambig.tbl"
 # Define peptide as fully flexible
-nfle = 1
 fle_sta_1 = 1
 fle_end_1 = 11
 fle_seg_1 = "B"

src/haddock/gear/expandable_parameters.py

@@ -109,8 +109,9 @@
 # common error messages
 _emsg_num_differ = (
     "The parameter block {!r} expects "
-    "{} parameters, but {} are present "
-    "in the user configuration file: {}."
+    "{} parameters ({}), but {} are present "
+    "in the user configuration file: {}. "
+    "Parameter(s) {} are missing."
 )
 
 
@@ -249,7 +250,7 @@ def _read_groups_in_user_config(
     extract_params: Callable[..., set[str]],
     _emsg_no_group: str = "Parameter block is not a valid expandable group",
     _emsg_unexpected_params: str = "Unexpected params for group",
-) -> set[str]:
+) -> tuple[set[str], dict[str, int]]:
     """
     Read groups in user config.
 
@@ -279,15 +280,18 @@ def _read_groups_in_user_config(
 
     Returns
     -------
-    set
+    new : set[str]
         A set with the new parameters in the user configuration file
         that are acceptable according to the expandable rules.
+    param_name_counts : dict[str, int]
+        Count of expendable parameter parameter names
     """
     # minimum=1 is used to capture groups with missing parameters
     user_groups = get_user_groups(user_config, minimum=1, reference=False)
     default_group_names = set(group[0] for group in default_groups)
 
     new: set[str] = set()
+    param_name_counts: dict[str, int] = {}
     for (param_name, group_idx), params in user_groups.items():
         if param_name not in default_group_names:
             emsg = _emsg_no_group.format(param_name, group_idx)
@@ -311,35 +315,40 @@ def _read_groups_in_user_config(
         # when num_found > num_expected the error about in `diff` is
         # triggered instead
         if num_found != num_expected:
+            expected_missings = set(expected_params).difference(params)
             emsg = _emsg_num_differ.format(
                 param_name,
                 num_expected,
+                ", ".join(expected_params),
                 num_found,
                 ", ".join(params),
+                ", ".join([f"`{p}`" for p in expected_missings]),
             )
             raise ConfigurationError(emsg)
 
         related_params = extract_params(user_config, param_name, group_idx)
+        # Increment counts for this parameter
+        if related_params:
+            param_name_counts.setdefault(param_name, 0)
+            param_name_counts[param_name] += 1
         new.update(related_params)
-
-    return new
+    return new, param_name_counts
 
 
 def read_simplest_expandable(
-    expparams: Iterable[str], config: Iterable[str]
+    config: Iterable[str], expparams: Iterable[str], 
 ) -> set[str]:
     """
     Read expandable parameters from config file of the type `param_1`.
 
     Parameters
     ----------
+    config : dict, dict.keys, set, or alike
+        The user configuration file.
     expparams : dict, dict.keys, set, or alike
         The parameter names that should be considered as expandable.
         Usually, this is a module subdictionary of `type_simplest_ep`.
 
-    config : dict, dict.keys, set, or alike
-        The user configuration file.
-
     Returns
     -------
     set of str

src/haddock/gear/prepare_run.py

@@ -799,18 +799,18 @@ def validate_ncs_params(params: dict) -> None:
     # Validate value of `numncs`
     ncs_suffixes = list(groupped_ncs.keys())
     # Case when number of definition do not match
-    if params["numncs"] != len(ncs_suffixes):
+    if params["nncs"] != len(ncs_suffixes):
         msg = (
-            f'Number of NCS restraints (`numncs = {params["numncs"]}`) '
+            f'Number of NCS restraints (`nncs = {params["nncs"]}`) '
             " do not match with the number of defined NCS restraints "
             f"({len(ncs_suffixes)})"
             )
         error_list.append(msg)
     else:
         # Case when numbers do not match
-        if max(ncs_suffixes) != params["numncs"]:
+        if max(ncs_suffixes) != params["nncs"]:
             msg = (
-                f'Number of NCS restraints (`numncs = {params["numncs"]}`) '
+                f'Number of NCS restraints (`nncs = {params["nncs"]}`) '
                 " do not match with the number of defined NCS restraints "
                 f"({', '.join([str(s) for s in ncs_suffixes])})"
                 )
@@ -1162,25 +1162,44 @@ def get_expandable_parameters(
 def _get_expandable(
     user_config: ParamMap, defaults: ParamMap, module_name: str, max_mols: int
 ) -> set[str]:
-    type_1 = get_single_index_groups(defaults)
-    type_2 = get_multiple_index_groups(defaults)
-    type_4 = get_mol_parameters(defaults)
-
+    # Set parsing vars
     allowed_params: set[str] = set()
-    allowed_params.update(
-        read_single_idx_groups_user_config(user_config, type_1)
-    )  # noqa: E501
-    allowed_params.update(
-        read_multiple_idx_groups_user_config(user_config, type_2)
-    )  # noqa: E501
+    all_counts: dict[str, int] = {}
+    # Read single indexed groups (terminating by `_X`)
+    news_t1, counts_t1 = read_single_idx_groups_user_config(
+        user_config,
+        get_single_index_groups(defaults),
+        )
+    allowed_params.update(news_t1)
+    all_counts.update(counts_t1)
+    # Read multiple indexed groups (terminating by `_X_Y`)
+    news_t2, counts_t2 = read_multiple_idx_groups_user_config(
+        user_config,
+        get_multiple_index_groups(defaults),
+        )
+    allowed_params.update(news_t2)
+    all_counts.update(counts_t2)
 
     with suppress(KeyError):
-        type_3 = type_simplest_ep[get_module_name(module_name)]
-        allowed_params.update(read_simplest_expandable(type_3, user_config))
-
-    _ = read_mol_parameters(user_config, type_4, max_mols=max_mols)
+        news_t3 = read_simplest_expandable(
+            user_config,
+            type_simplest_ep[get_module_name(module_name)],
+            )
+        allowed_params.update(news_t3)
+    # Read molecule paramters (starting by `mol_`)
+    _ = read_mol_parameters(
+        user_config,
+        get_mol_parameters(defaults),
+        max_mols=max_mols,
+        )
     allowed_params.update(_)
 
+    # Add counted parameters to hidden user parameters
+    for param, count in all_counts.items():
+        count_param_name = f"n{param}"
+        if count_param_name in defaults.keys():
+            user_config[count_param_name] = count
+    # Return new set of allowed parameters
     return allowed_params
 
 

src/haddock/modules/refinement/emref/cns/emref.cns

@@ -59,17 +59,17 @@ evaluate ($data.kcont=$kcont)
 ! NCS restraints
 evaluate ($data.kncs = $kncs)
 evaluate ($Data.flags.ncs = $ncs_on)
-evaluate ($data.numncs = $numncs)
+evaluate ($data.nncs = $nncs)
 
 ! Symmetry restraints
 evaluate ($data.ksym = $ksym)
 evaluate ($Data.flags.sym = $sym_on)
-evaluate ($data.numc2sym = $numc2sym)
-evaluate ($data.numc3sym = $numc3sym)
-evaluate ($data.nums3sym = $nums3sym)
-evaluate ($data.numc4sym = $numc4sym)
-evaluate ($data.numc5sym = $numc5sym)
-evaluate ($data.numc6sym = $numc6sym)
+evaluate ($data.nc2sym = $nc2sym)
+evaluate ($data.nc3sym = $nc3sym)
+evaluate ($data.ns3sym = $ns3sym)
+evaluate ($data.nc4sym = $nc4sym)
+evaluate ($data.nc5sym = $nc5sym)
+evaluate ($data.nc6sym = $nc6sym)
 
 !Dihedral angle energy term:
 evaluate ($data.flags.dihed = true)