Fixing shape support in cgtoaa module

CHANGELOG.md

@@ -1,5 +1,6 @@
 # Changelog
 
+- 2026-04-13: Fixed issue with shape in cgtoa module and added shape docking example
 - 2026-03-30: Added nucleosome-PCR1 CG docking example
 - 2026-03-09: Automated type casting for optional argument seed in haddock3-restraints random_removal - Issue #1485
 - 2026-02-28: Switched to ilRMSD clustering from protein-ligand examples - Issue #1481

end-to-end_tests/test_protein-protein-shape.py

@@ -0,0 +1,54 @@
+"""End-to-end test for the protein-protein-shape docking example."""
+
+import shutil
+import tempfile
+import gzip
+from pathlib import Path
+
+from haddock.clis.cli import main as cli_main
+
+import warnings
+from Bio.PDB.PDBExceptions import PDBConstructionWarning
+
+
+EXAMPLE_DIR = (
+    Path(__file__).resolve().parents[1] / "examples" / "docking-protein-protein-shape"
+)
+
+def test_protein_protein_shape(monkeypatch):
+    """Test protein-protein-shape CG docking example.
+
+    Uses docking-protein-protein-shape-CG-test.cfg to run the full pipeline
+    (topoaa, topocg rigidbody, caprieval, seletop, flexref, caprieval, cgtoaa, emref, caprieval,
+    clustfcc, seletopclusts, caprieva) 
+    """
+    warnings.simplefilter('ignore', PDBConstructionWarning)
+
+    with tempfile.TemporaryDirectory() as tmpdir:
+        shutil.copytree(Path(EXAMPLE_DIR, "data"), Path(tmpdir, "data"))
+        cfg = Path(tmpdir, "docking-protein-protein-shape-CG-test.cfg")
+        shutil.copy(
+            Path(EXAMPLE_DIR, "docking-protein-protein-shape-CG-test.cfg"), cfg
+        )
+
+        monkeypatch.chdir(tmpdir)
+        cli_main(cfg)
+
+        run_dir = Path("run1-CG-test")
+
+        # Verify all workflow steps produced output directories
+        assert Path(run_dir, "00_topoaa").exists(), f"00_topoaa not created"
+        assert Path(run_dir, "01_topocg").exists(), f"01_topocg not created"
+        assert Path(run_dir, "02_rigidbody").exists(), f"02_rigidbody not created"
+        assert Path(run_dir, "03_caprieval").exists(), f"03_caprieval not created"
+        assert Path(run_dir, "04_seletop").exists(), f"04_seletop created"
+        assert Path(run_dir, "05_flexref").exists(), f"05_flexref not created"
+        assert Path(run_dir, "06_caprieval").exists(), f"06_caprieval not created"
+        assert Path(run_dir, "07_cgtoaa").exists(), f"07_cgtoaa not created"
+        assert Path(run_dir, "08_emref").exists(), f"08_emref not created"
+        assert Path(run_dir, "09_caprieval").exists(), f"09_caprieval not created"
+        assert Path(run_dir, "10_clustfcc").exists(), f"10_clustfcc not created"
+        assert Path(run_dir, "11_seletopclusts").exists(), f"10_seletopclusts not created"
+        assert Path(run_dir, "12_caprieval").exists(), f"12_caprieval not created"
+        with gzip.open(Path(run_dir, "08_emref/emref_1.pdb.gz"), 'rt') as f:
+            assert any('SHA SHA S' in line for line in f), f"Shape atoms not found in emref PDB file"

examples/docking-protein-protein-shape/docking-protein-protein-shape-CG-full.cfg

@@ -0,0 +1,67 @@
+# ====================================================================
+# Protein-protein docking example with shape restraints
+
+# directory in which the scoring will be done
+run_dir = "run1-CG-full"
+
+# execution mode
+mode = "local"
+ncores = 50
+
+# molecules to be docked
+molecules =  [
+    "data/2r15_A.pdb",
+    "data/2r15_B.pdb",
+    "data/shape.pdb",
+    ]
+
+# ====================================================================
+# Parameters for each stage are defined below, prefer full paths
+# ====================================================================
+[topoaa]
+
+[topocg]
+
+[rigidbody]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+fnat_cutoff = 7.0 # cutoff for coarse-grained structures
+
+[seletop]
+
+[flexref]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+fnat_cutoff = 7.0 # cutoff for coarse-grained structures
+
+[cgtoaa]
+mol_shape_3 = true
+log_level = 'verbose'
+
+[emref]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+[clustfcc]
+min_population = 4
+
+[seletopclusts]
+top_models = 4
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+# ====================================================================
+

examples/docking-protein-protein-shape/docking-protein-protein-shape-CG-test.cfg

@@ -0,0 +1,69 @@
+# ====================================================================
+# Protein-protein docking example with shape restraints
+
+# directory in which the scoring will be done
+run_dir = "run1-CG-test"
+
+# execution mode
+mode = "local"
+ncores = 10
+debug = true
+
+# molecules to be docked
+molecules =  [
+    "data/2r15_A.pdb",
+    "data/2r15_B.pdb",
+    "data/shape.pdb",
+    ]
+
+# ====================================================================
+# Parameters for each stage are defined below, prefer full paths
+# ====================================================================
+[topoaa]
+
+[topocg]
+
+[rigidbody]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+sampling = 5
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+fnat_cutoff = 7.0 # cutoff for coarse-grained structures
+
+[seletop]
+select = 2
+
+[flexref]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+fnat_cutoff = 7.0 # cutoff for coarse-grained structures
+
+[cgtoaa]
+mol_shape_3 = true
+
+[emref]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+[clustfcc]
+min_population = 1
+
+[seletopclusts]
+top_models = 4
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+# ====================================================================
+

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

@@ -0,0 +1,59 @@
+# ====================================================================
+# Protein-protein docking example with shape restraints
+
+# directory in which the scoring will be done
+run_dir = "run1-full"
+
+# execution mode
+mode = "local"
+ncores = 50
+
+# molecules to be docked
+molecules =  [
+    "data/2r15_A.pdb",
+    "data/2r15_B.pdb",
+    "data/shape.pdb",
+    ]
+
+# ====================================================================
+# Parameters for each stage are defined below, prefer full paths
+# ====================================================================
+[topoaa]
+
+[rigidbody]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+[seletop]
+
+[flexref]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+[emref]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+[clustfcc]
+min_population = 4
+
+[seletopclusts]
+top_models = 4
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+# ====================================================================
+

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

@@ -0,0 +1,61 @@
+# ====================================================================
+# Protein-protein docking example with shape restraints
+
+# directory in which the scoring will be done
+run_dir = "run1-test"
+
+# execution mode
+mode = "local"
+ncores = 10
+
+# molecules to be docked
+molecules =  [
+    "data/2r15_A.pdb",
+    "data/2r15_B.pdb",
+    "data/shape.pdb",
+    ]
+
+# ====================================================================
+# Parameters for each stage are defined below, prefer full paths
+# ====================================================================
+[topoaa]
+
+[rigidbody]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+sampling = 5
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+[seletop]
+select = 2
+
+[flexref]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+[emref]
+tolerance = 20
+ambig_fname = "data/ambig-shape.tbl"
+mol_shape_3 = true
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+[clustfcc]
+min_population = 1
+
+[seletopclusts]
+top_models = 4
+
+[caprieval]
+reference_fname = "data/2r15_reference.pdb"
+
+# ====================================================================
+

src/haddock/libs/libaa2cg.py

@@ -850,6 +850,7 @@ def martinize(input_pdb, output_path, skipss):
     Returns:
         cg_pdb_name: str
     """
+    shape = False
 
     if not input_pdb:
         emsg = "No input file detected"
@@ -906,6 +907,8 @@ def martinize(input_pdb, output_path, skipss):
             for residue in mapping_dic:
                 if residue.id[0] != " ":  # filter HETATMS
                     continue
+                if residue.resname == "SHA":
+                    shape = True
 
                 structure_builder.init_residue(residue.resname, residue.id[0],
                                                residue.id[1], residue.id[2])
@@ -964,5 +967,5 @@ def martinize(input_pdb, output_path, skipss):
     tbl_file.write(f"\n{tbl_str}")
     tbl_file.close()
 
-    return cg_pdb_name
+    return cg_pdb_name, shape
 

src/haddock/modules/analysis/caprieval/__init__.py

@@ -217,7 +217,7 @@ def _run(self) -> None:
 
         if ff == "martini2":
             references = [
-                Path(martinize(ref_aa, self.path, False))
+                Path(martinize(ref_aa, self.path, False)[0])
                 for ref_aa in self.get_reference(models)
                 ]
         else:

src/haddock/modules/refinement/cgtoaa/cns/cgtoaa.cns

@@ -98,8 +98,10 @@ do (segid = segid + "CG") (not segid S)
 evaluate ($nchain=0)
 while ($nchain < $data.ncomponents) loop nloop1
     evaluate ($nchain = $nchain + 1)
-    structure @@$input_aa_psf_filename_$nchain end
-    coor @@$input_aa_pdb_filename_$nchain
+    if ($mol_shape_$nchain eq false) then
+        structure @@$input_aa_psf_filename_$nchain end
+        coor @@$input_aa_pdb_filename_$nchain
+    end if
 end loop nloop1
 
 {* Read paramters and set various force field related variables *}
@@ -161,17 +163,19 @@ evaluate ($kinter = 0.001)
 
 flag excl * incl noe end
 
-! Fix the CG model
-fix sele=(segid *CG) end
+! Fix the CG model and any shape
+fix sele=(segid *CG or segid S) end
 
 ! Rigid body EM to position AA model onto CG model
 !
 minimize rigid
     eval ($nchain = 0)
     while ($nchain < $data.ncomponents) loop nloop1
         eval($nchain = $nchain + 1)
-        group (segid $prot_segid_$nchain)
-        translation=true
+        if ($mol_shape_$nchain eq false) then
+            group (segid $prot_segid_$nchain)
+            translation=true
+        end if
     end loop nloop1
     nstep 200
     nprint 10