Added UI for adding missing residues. Added option to delete chains.

pdbfixer.py

@@ -69,7 +69,7 @@ def _overlayPoints(points1, points2):
     if len(points1) == 0:
     if len(points1) == 0:
         return (Vec3(0, 0, 0), np.identity(3), Vec3(0, 0, 0))
         return (Vec3(0, 0, 0), np.identity(3), Vec3(0, 0, 0))
     if len(points1) == 1:
     if len(points1) == 1:
-        return (points1[0], np.identity(3), -points2[0])
+        return (points1[0], np.identity(3), -1*points2[0])
     
     
     # Compute centroids.
     # Compute centroids.
     
     
@@ -98,6 +98,7 @@ class PDBFixer(object):
         self.topology = self.pdb.topology
         self.topology = self.pdb.topology
         self.positions = self.pdb.positions
         self.positions = self.pdb.positions
         self.centroid = unit.sum(self.positions)/len(self.positions)
         self.centroid = unit.sum(self.positions)/len(self.positions)
+        self._structureChains = list(self.structure.iter_chains())
         
         
         # Load the templates.
         # Load the templates.
         
         
@@ -126,8 +127,8 @@ class PDBFixer(object):
                 
                 
                 # Insert missing residues here.
                 # Insert missing residues here.
                 
                 
-                insertHere = [r[2] for r in self.missingResidues if r[0] == chain.index and r[1] == indexInChain]
-                if len(insertHere) > 0:
+                if (chain.index, indexInChain) in self.missingResidues:
+                    insertHere = self.missingResidues[(chain.index, indexInChain)]
                     endPosition = self._computeResidueCenter(residue)
                     endPosition = self._computeResidueCenter(residue)
                     if indexInChain > 0:
                     if indexInChain > 0:
                         startPosition = self._computeResidueCenter(chainResidues[indexInChain-1])
                         startPosition = self._computeResidueCenter(chainResidues[indexInChain-1])
@@ -180,8 +181,8 @@ class PDBFixer(object):
 
 
                 # If this is the end of the chain, add any missing residues that come after it.
                 # If this is the end of the chain, add any missing residues that come after it.
                 
                 
-                if residue == chainResidues[-1]:
-                    insertHere = [r[2] for r in self.missingResidues if r[0] == chain.index and r[1] > indexInChain]
+                if residue == chainResidues[-1] and (chain.index, indexInChain+1) in self.missingResidues:
+                    insertHere = self.missingResidues[(chain.index, indexInChain+1)]
                     if len(insertHere) > 0:
                     if len(insertHere) > 0:
                         startPosition = self._computeResidueCenter(residue)
                         startPosition = self._computeResidueCenter(residue)
                         outward = startPosition-self.centroid
                         outward = startPosition-self.centroid
@@ -241,6 +242,14 @@ class PDBFixer(object):
                 templatePosition = template.positions[atom.index].value_in_unit(unit.nanometer)
                 templatePosition = template.positions[atom.index].value_in_unit(unit.nanometer)
                 newPositions.append(mm.Vec3(*np.dot(rotate, templatePosition))*unit.nanometer+translate)
                 newPositions.append(mm.Vec3(*np.dot(rotate, templatePosition))*unit.nanometer+translate)
     
     
+    def removeChains(self, chainIndices):
+        modeller = app.Modeller(self.topology, self.positions)
+        allChains = list(self.topology.chains())
+        modeller.delete(allChains[i] for i in chainIndices)
+        self.topology = modeller.topology
+        self.positions = modeller.positions
+        self._structureChains = [self._structureChains[i] for i in range(len(self._structureChains)) if i not in chainIndices]
+    
     def findNonstandardResidues(self):
     def findNonstandardResidues(self):
         self.nonstandardResidues = [r for r in self.topology.residues() if r.name in substitutions]
         self.nonstandardResidues = [r for r in self.topology.residues() if r.name in substitutions]
     
     
@@ -266,7 +275,7 @@ class PDBFixer(object):
             self.positions = modeller.positions
             self.positions = modeller.positions
     
     
     def findMissingResidues(self):
     def findMissingResidues(self):
-        chains = [c for c in self.structure.iter_chains() if any(atom.record_name == 'ATOM' for atom in c.iter_atoms())]
+        chains = [c for c in self._structureChains if any(atom.record_name == 'ATOM' for atom in c.iter_atoms())]
         chainWithGaps = {}
         chainWithGaps = {}
         
         
         # Find the sequence of each chain, with gaps for missing residues.
         # Find the sequence of each chain, with gaps for missing residues.
@@ -299,8 +308,8 @@ class PDBFixer(object):
         
         
         # Now build the list of residues to add.
         # Now build the list of residues to add.
         
         
-        self.missingResidues = []
-        for structChain, topChain in zip(self.structure.iter_chains(), self.pdb.topology.chains()):
+        self.missingResidues = {}
+        for structChain, topChain in zip(self._structureChains, self.pdb.topology.chains()):
             if structChain in chainSequence:
             if structChain in chainSequence:
                 offset = chainOffset[structChain]
                 offset = chainOffset[structChain]
                 sequence = chainSequence[structChain].residues
                 sequence = chainSequence[structChain].residues
@@ -308,7 +317,10 @@ class PDBFixer(object):
                 index = 0
                 index = 0
                 for i in range(len(sequence)):
                 for i in range(len(sequence)):
                     if i < offset or i >= len(gappedSequence)+offset or gappedSequence[i-offset] is None:
                     if i < offset or i >= len(gappedSequence)+offset or gappedSequence[i-offset] is None:
-                        self.missingResidues.append((topChain.index, index, sequence[i]))
+                        key = (topChain.index, index)
+                        if key not in self.missingResidues:
+                            self.missingResidues[key] = []
+                        self.missingResidues[key].append(sequence[i])
                     else:
                     else:
                         index += 1
                         index += 1
     
     
@@ -337,7 +349,7 @@ class PDBFixer(object):
                     # Add missing terminal atoms.
                     # Add missing terminal atoms.
                     
                     
                     terminals = []
                     terminals = []
-                    if residue == chainResidues[-1] and not any(r[0] == chain.index and r[1] >= len(chainResidues) for r in self.missingResidues):
+                    if residue == chainResidues[-1] and (chain.index, len(chainResidues)) not in self.missingResidues:
                         templateNames = set(atom.name for atom in template.topology.atoms())
                         templateNames = set(atom.name for atom in template.topology.atoms())
                         if 'OXT' not in atomNames and all(name in templateNames for name in ['C', 'O', 'CA']):
                         if 'OXT' not in atomNames and all(name in templateNames for name in ['C', 'O', 'CA']):
                             terminals.append('OXT')
                             terminals.append('OXT')

ui.py

@@ -10,7 +10,20 @@ def startPageCallback(parameters, handler):
         global fixer
         global fixer
         pdb = PdbStructure(parameters['pdbfile'].value.splitlines())
         pdb = PdbStructure(parameters['pdbfile'].value.splitlines())
         fixer = PDBFixer(pdb)
         fixer = PDBFixer(pdb)
-        displayConvertResiduesPage()
+        displayDeleteChainsPage()
+
+def deleteChainsPageCallback(parameters, handler):
+    numChains = len(list(fixer.topology.chains()))
+    deleteIndices = [i for i in range(numChains) if 'include'+str(i) not in parameters]
+    fixer.removeChains(deleteIndices)
+    displayAddResiduesPage()
+
+def addResiduesPageCallback(parameters, handler):
+    keys = [key for key in sorted(fixer.missingResidues)]
+    for i, key in enumerate(keys):
+        if 'add'+str(i) not in parameters:
+            del fixer.missingResidues[key]
+    displayMissingAtomsPage()
 
 
 def convertResiduesPageCallback(parameters, handler):
 def convertResiduesPageCallback(parameters, handler):
     fixer.nonstandardResidues = [residue for i, residue in enumerate(fixer.nonstandardResidues) if 'convert'+str(i) in parameters]
     fixer.nonstandardResidues = [residue for i, residue in enumerate(fixer.nonstandardResidues) if 'convert'+str(i) in parameters]
@@ -47,9 +60,75 @@ PDB File: <input type="file" name="pdbfile"/>
 </html>
 </html>
 """)
 """)
 
 
+def displayDeleteChainsPage():
+    uiserver.setCallback(deleteChainsPageCallback)
+    numChains = len(list(fixer.topology.chains()))
+    if numChains < 2:
+        displayAddResiduesPage()
+        return
+    table = ""
+    proteinResidues = ['ALA', 'ASN', 'CYS', 'GLU', 'HIS', 'LEU', 'MET', 'PRO', 'THR', 'TYR.pdb ARG', 'ASP', 'GLN', 'GLY', 'ILE', 'LYS', 'PHE', 'SER', 'TRP', 'VAL']
+    rnaResidues = ['A', 'G', 'C', 'U']
+    dnaResidues = ['DA', 'DG', 'DC', 'DT']
+    for i, chain in enumerate(fixer.topology.chains()):
+        residues = list(r.name for r in chain.residues())
+        if any(r in proteinResidues for r in residues):
+            content = "Protein"
+        elif any(r in rnaResidues for r in residues):
+            content = "RNA"
+        elif any(r in dnaResidues for r in residues):
+            content = "DNA"
+        else:
+            content = ', '.join(set(residues))
+        table += '    <tr><td>%d</td><td>%d</td><td>%s</td><td><input type="checkbox" name="include%d" checked></td></tr>\n' % (chain.index+1, len(residues), content, i)
+    uiserver.setContent("""
+<html>
+<head><title>PDB Fixer</title></head>
+<body>
+This PDB file contains %d chains.  Select which ones to include.
+<p>
+<form method="post" action="/">
+<table border="1">
+    <tr><th>Chain</th><th># Residues</th><th>Content</th><th>Include?</th></tr>
+%s
+</table>
+<p>
+<input type="submit" value="Continue"/>
+</form>
+</body>
+<html>
+""" % (numChains, table))
+
+def displayAddResiduesPage():
+    uiserver.setCallback(addResiduesPageCallback)
+    fixer.findMissingResidues()
+    if len(fixer.missingResidues) == 0:
+        displayConvertResiduesPage()
+        return
+    table = ""
+    for i, key in enumerate(sorted(fixer.missingResidues)):
+        residues = fixer.missingResidues[key]
+        table += '    <tr><td>%d</td><td>%d to %d</td><td>%s</td><td><input type="checkbox" name="add%d" checked></td></tr>\n' % (key[0]+1, key[1]+1, key[1]+len(residues), ', '.join(residues), i)
+    uiserver.setContent("""
+<html>
+<head><title>PDB Fixer</title></head>
+<body>
+The SEQRES records in this PDB file include residues that are missing from the atom data section.  Do you want to add the missing residues?
+<p>
+<form method="post" action="/">
+<table border="1">
+    <tr><th>Chain</th><th>Residue Positions</th><th>Sequence</th><th>Add?</th></tr>
+%s
+</table>
+<p>
+<input type="submit" value="Continue"/>
+</form>
+</body>
+<html>
+""" % table)
+
 def displayConvertResiduesPage():
 def displayConvertResiduesPage():
     uiserver.setCallback(convertResiduesPageCallback)
     uiserver.setCallback(convertResiduesPageCallback)
-    fixer.findMissingResidues()
     fixer.findNonstandardResidues()
     fixer.findNonstandardResidues()
     if len(fixer.nonstandardResidues) == 0:
     if len(fixer.nonstandardResidues) == 0:
         displayMissingAtomsPage()
         displayMissingAtomsPage()
@@ -105,7 +184,7 @@ def displayMissingAtomsPage():
 <body>
 <body>
 The following residues are missing heavy atoms, which will be added.
 The following residues are missing heavy atoms, which will be added.
 <p>
 <p>
-<form method="get" action="/">
+<form method="post" action="/">
 <table border="1">
 <table border="1">
     <tr><th>Chain</th><th>Residue</th><th>Missing Atoms</th></tr>
     <tr><th>Chain</th><th>Residue</th><th>Missing Atoms</th></tr>
 %s
 %s
@@ -125,7 +204,7 @@ def displayDownloadPage():
 <body>
 <body>
 The fixed PDB file is ready to download.
 The fixed PDB file is ready to download.
 <p>
 <p>
-<form method="get" action="/">
+<form method="post" action="/">
 <input type="submit" name="download" value="Download"/>
 <input type="submit" name="download" value="Download"/>
 <input type="submit" name="newfile" value="Process Another File"/>
 <input type="submit" name="newfile" value="Process Another File"/>
 </form>
 </form>