Upload refine.py

scripts/refine.py

@@ -1,52 +1,60 @@
 import os
-from Bio.PDB import PDBParser, Superimposer
+from Bio.PDB import PDBParser, Superimposer, PDBIO, Select
 from openmm.app import PDBFile, ForceField, Simulation, Modeller
 from openmm import LangevinIntegrator
 from openmm.unit import kelvin, picosecond
 
+# A custom selector to strip out any 'weird' terminal atoms OpenMM hates
+class TerminalSanitizer(Select):
+    def accept_atom(self, atom):
+        # We strip OXT (terminal oxygen) and any hydrogen that shouldn't be there
+        return atom.get_name() not in ["OXT", "H1", "H2", "H3"]
+
 def polish_design(target_pdb_id, uploaded_file_path):
     # 1. Setup paths
     target_path = os.path.join("data", f"{target_pdb_id.lower()}.pdb")
+    sanitized_path = "sanitized_design.pdb"
     
-    # 2. ALIGNMENT (Biopython)
+    # 2. SANITIZE & ALIGN (The "Shut Up" Step)
     parser = PDBParser(QUIET=True)
     target_struct = parser.get_structure("target", target_path)
     design_struct = parser.get_structure("design", uploaded_file_path)
     
+    # Strip terminal junk before OpenMM sees it
+    io = PDBIO()
+    io.set_structure(design_struct)
+    io.save(sanitized_path, TerminalSanitizer())
+    
+    # Standard Alignment to hit that 0.75 Å RMSD
     sup = Superimposer()
     t_atoms = [a for a in target_struct.get_atoms() if a.get_name() == 'CA']
     d_atoms = [a for a in design_struct.get_atoms() if a.get_name() == 'CA']
-    
     sup.set_atoms(t_atoms[:len(d_atoms)], d_atoms)
     sup.apply(design_struct.get_atoms())
     rmsd = sup.rms
 
-    # 3. PHYSICS (OpenMM Modeller replaces PDBFixer)
-    # Load the "broken" PDB
-    pdb = PDBFile(uploaded_file_path)
-    
-    # Use Modeller to repair the C-terminus
-    # Modeller.addHydrogens and addExtraParticles are usually the fix, 
-    # but we can force it to ignore the missing bond during template matching.
-    forcefield = ForceField('amber14-all.xml', 'amber14/tip3p.xml')
+    # 3. PHYSICS (The Lobotomized Loader)
+    # We load the SANITIZED file, not the raw upload
+    pdb = PDBFile(sanitized_path)
     modeller = Modeller(pdb.topology, pdb.positions)
+    forcefield = ForceField('amber14-all.xml', 'amber14/tip3p.xml')
     
-    # THE FIX: This adds the missing OXT or Hydrogens that OpenMM is whining about
-    modeller.addHydrogens(forcefield)
+    # Re-add hydrogens properly according to the forcefield
+    modeller.addHydrogens(forcefield, pH=7.4)
     
-    # Setup the system using the REPAIRED modeller topology
-    # Note: ignoreExternalBonds=True kills the LYS 155 error
+    # Ignore bonds and force system creation
     system = forcefield.createSystem(modeller.topology, ignoreExternalBonds=True)
+    
+    # 4. MINIMIZE
     integrator = LangevinIntegrator(300*kelvin, 1/picosecond, 0.002*picosecond)
     simulation = Simulation(modeller.topology, system, integrator)
     simulation.context.setPositions(modeller.positions)
-    
     simulation.minimizeEnergy(maxIterations=100)
     
-    # 4. EXPORT
+    # 5. EXPORT FINAL LEAD
     output_name = "Broteinshake_Lead_Refined.pdb"
     with open(output_name, "w") as f:
         PDBFile.writeFile(simulation.topology, 
                          simulation.context.getState(getPositions=True).getPositions(), f)
     
-    return output_name, rmsd
+    return output_name, rmsd