Merge pull request #73 from peastman/ids

Eliminated most dependence on PdbStructure

pdbfixer/pdbfixer.py

@@ -6,7 +6,7 @@ Simbios, the NIH National Center for Physics-Based Simulation of
 Biological Structures at Stanford, funded under the NIH Roadmap for
 Biological Structures at Stanford, funded under the NIH Roadmap for
 Medical Research, grant U54 GM072970. See https://simtk.org.
 Medical Research, grant U54 GM072970. See https://simtk.org.
 
 
-Portions copyright (c) 2013-2014 Stanford University and the Authors.
+Portions copyright (c) 2013-2015 Stanford University and the Authors.
 Authors: Peter Eastman
 Authors: Peter Eastman
 Contributors:
 Contributors:
 
 
@@ -222,7 +222,6 @@ class PDBFixer(object):
         self.pdb = app.PDBFile(structure)
         self.pdb = app.PDBFile(structure)
         self.topology = self.pdb.topology
         self.topology = self.pdb.topology
         self.positions = self.pdb.positions
         self.positions = self.pdb.positions
-        self.structureChains = list(self.structure.iter_chains())
         
         
         # Load the templates.
         # Load the templates.
         
         
@@ -463,7 +462,6 @@ class PDBFixer(object):
         modeller.delete(allChains[i] for i in chainIndices)
         modeller.delete(allChains[i] for i in chainIndices)
         self.topology = modeller.topology
         self.topology = modeller.topology
         self.positions = modeller.positions
         self.positions = modeller.positions
-        self.structureChains = [self.structureChains[i] for i in range(len(self.structureChains)) if i not in chainIndices]
 
 
         return
         return
     
     
@@ -482,17 +480,17 @@ class PDBFixer(object):
         >>> missing_residues = fixer.missingResidues
         >>> missing_residues = fixer.missingResidues
 
 
         """
         """
-        chains = [c for c in self.structureChains if any(atom.record_name == 'ATOM' for atom in c.iter_atoms())]
+        chains = [c for c in self.topology.chains() if len(list(c.residues())) > 0]
         chainWithGaps = {}
         chainWithGaps = {}
         
         
         # Find the sequence of each chain, with gaps for missing residues.
         # Find the sequence of each chain, with gaps for missing residues.
         
         
         for chain in chains:
         for chain in chains:
-            minResidue = min(r.number for r in chain.iter_residues())
-            maxResidue = max(r.number for r in chain.iter_residues())
+            minResidue = min(int(r.id) for r in chain.residues())
+            maxResidue = max(int(r.id) for r in chain.residues())
             residues = [None]*(maxResidue-minResidue+1)
             residues = [None]*(maxResidue-minResidue+1)
-            for r in chain.iter_residues():
-                residues[r.number-minResidue] = r.get_name()
+            for r in chain.residues():
+                residues[int(r.id)-minResidue] = r.name
             chainWithGaps[chain] = residues
             chainWithGaps[chain] = residues
         
         
         # Try to find the chain that matches each sequence.
         # Try to find the chain that matches each sequence.
@@ -501,7 +499,7 @@ class PDBFixer(object):
         chainOffset = {}
         chainOffset = {}
         for sequence in self.structure.sequences:
         for sequence in self.structure.sequences:
             for chain in chains:
             for chain in chains:
-                if chain.chain_id != sequence.chain_id:
+                if chain.id != sequence.chain_id:
                     continue
                     continue
                 if chain in chainSequence:
                 if chain in chainSequence:
                     continue
                     continue
@@ -516,15 +514,15 @@ class PDBFixer(object):
         # Now build the list of residues to add.
         # Now build the list of residues to add.
         
         
         self.missingResidues = {}
         self.missingResidues = {}
-        for structChain, topChain in zip(self.structureChains, self.topology.chains()):
-            if structChain in chainSequence:
-                offset = chainOffset[structChain]
-                sequence = chainSequence[structChain].residues
-                gappedSequence = chainWithGaps[structChain]
+        for chain in self.topology.chains():
+            if chain in chainSequence:
+                offset = chainOffset[chain]
+                sequence = chainSequence[chain].residues
+                gappedSequence = chainWithGaps[chain]
                 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:
-                        key = (topChain.index, index)
+                        key = (chain.index, index)
                         if key not in self.missingResidues:
                         if key not in self.missingResidues:
                             self.missingResidues[key] = []
                             self.missingResidues[key] = []
                         residueName = sequence[i]
                         residueName = sequence[i]
@@ -557,16 +555,16 @@ class PDBFixer(object):
         
         
         # Now add ones based on MODRES records.
         # Now add ones based on MODRES records.
         
         
-        modres = dict(((m.chain_id, m.number, m.residue_name), m.standard_name) for m in self.structure.modified_residues)
-        for structChain, topChain in zip(self.structureChains, self.topology.chains()):
-            for structResidue, topResidue in zip(structChain.iter_residues(), topChain.residues()):
-                key = (structChain.chain_id, structResidue.number, structResidue.name)
+        modres = dict(((m.chain_id, str(m.number), m.residue_name), m.standard_name) for m in self.structure.modified_residues)
+        for chain in self.topology.chains():
+            for residue in chain.residues():
+                key = (chain.id, residue.id, residue.name)
                 if key in modres:
                 if key in modres:
                     replacement = modres[key]
                     replacement = modres[key]
                     if replacement == 'DU':
                     if replacement == 'DU':
                         replacement = 'DT'
                         replacement = 'DT'
                     if replacement in self.templates:
                     if replacement in self.templates:
-                        nonstandard[topResidue] = replacement
+                        nonstandard[residue] = replacement
         self.nonstandardResidues = [(r, nonstandard[r]) for r in sorted(nonstandard, key=lambda r: r.index)]
         self.nonstandardResidues = [(r, nonstandard[r]) for r in sorted(nonstandard, key=lambda r: r.index)]
     
     
     def replaceNonstandardResidues(self):
     def replaceNonstandardResidues(self):

pdbfixer/ui.py

@@ -83,7 +83,7 @@ def addResiduesPageCallback(parameters, handler):
     for i, key in enumerate(keys):
     for i, key in enumerate(keys):
         if 'add'+str(i) not in parameters:
         if 'add'+str(i) not in parameters:
             del fixer.missingResidues[key]
             del fixer.missingResidues[key]
-    displayMissingAtomsPage()
+    displayConvertResiduesPage()
 
 
 def convertResiduesPageCallback(parameters, handler):
 def convertResiduesPageCallback(parameters, handler):
     for i in range(len(fixer.nonstandardResidues)):
     for i in range(len(fixer.nonstandardResidues)):
@@ -144,7 +144,7 @@ def displayDeleteChainsPage():
             content = "DNA"
             content = "DNA"
         else:
         else:
             content = ', '.join(set(residues))
             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)
+        table += '    <tr><td>%s</td><td>%d</td><td>%s</td><td><input type="checkbox" name="include%d" checked></td></tr>\n' % (chain.id, len(residues), content, i)
     uiserver.setContent(header+loadHtmlFile("removeChains.html") % (numChains, table))
     uiserver.setContent(header+loadHtmlFile("removeChains.html") % (numChains, table))
 
 
 def displayAddResiduesPage():
 def displayAddResiduesPage():
@@ -154,14 +154,16 @@ def displayAddResiduesPage():
         displayConvertResiduesPage()
         displayConvertResiduesPage()
         return
         return
     table = ""
     table = ""
+    chains = list(fixer.topology.chains())
     for i, key in enumerate(sorted(fixer.missingResidues)):
     for i, key in enumerate(sorted(fixer.missingResidues)):
         residues = fixer.missingResidues[key]
         residues = fixer.missingResidues[key]
-        chain = fixer.structureChains[key[0]]
-        if key[1] < len(chain.residues):
-            offset = chain.residues[key[1]].number-len(residues)-1
+        chain = chains[key[0]]
+        chainResidues = list(chain.residues())
+        if key[1] < len(chainResidues):
+            offset = int(chainResidues[key[1]].id)-len(residues)-1
         else:
         else:
-            offset = chain.residues[-1].number
-        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, offset+1, offset+len(residues), ', '.join(residues), i)
+            offset = int(chainResidues[-1].id)
+        table += '    <tr><td>%s</td><td>%d to %d</td><td>%s</td><td><input type="checkbox" name="add%d" checked></td></tr>\n' % (chain.id, offset+1, offset+len(residues), ', '.join(residues), i)
     uiserver.setContent(header+loadHtmlFile("addResidues.html") % table)
     uiserver.setContent(header+loadHtmlFile("addResidues.html") % table)
 
 
 def displayConvertResiduesPage():
 def displayConvertResiduesPage():
@@ -170,10 +172,6 @@ def displayConvertResiduesPage():
     if len(fixer.nonstandardResidues) == 0:
     if len(fixer.nonstandardResidues) == 0:
         displayMissingAtomsPage()
         displayMissingAtomsPage()
         return
         return
-    indexInChain = {}
-    for structChain, topChain in zip(fixer.structureChains, fixer.topology.chains()):
-        for structResidue, topResidue in zip(structChain.iter_residues(), topChain.residues()):
-            indexInChain[topResidue] = structResidue.number
     table = ''
     table = ''
     nucleotides = ['DA', 'DC', 'DG', 'DT', 'A', 'C', 'G', 'T']
     nucleotides = ['DA', 'DC', 'DG', 'DT', 'A', 'C', 'G', 'T']
     for i in range(len(fixer.nonstandardResidues)):
     for i in range(len(fixer.nonstandardResidues)):
@@ -188,7 +186,7 @@ def displayConvertResiduesPage():
             if res == replaceWith:
             if res == replaceWith:
                 selected = ' selected'
                 selected = ' selected'
             options += '<option value="%s"%s>%s</option>' % (res, selected, res)
             options += '<option value="%s"%s>%s</option>' % (res, selected, res)
-        table += '    <tr><td>%d</td><td>%s %d</td><td><select name="residue%d">%s</select></td><td><input type="checkbox" name="convert%d" checked></td></tr>\n' % (residue.chain.index+1, residue.name, indexInChain[residue], i, options, i)
+        table += '    <tr><td>%s</td><td>%s %s</td><td><select name="residue%d">%s</select></td><td><input type="checkbox" name="convert%d" checked></td></tr>\n' % (residue.chain.id, residue.name, residue.id, i, options, i)
     uiserver.setContent(header+loadHtmlFile("convertResidues.html") % table)
     uiserver.setContent(header+loadHtmlFile("convertResidues.html") % table)
 
 
 def displayMissingAtomsPage():
 def displayMissingAtomsPage():
@@ -200,10 +198,6 @@ def displayMissingAtomsPage():
         fixer.addMissingAtoms()
         fixer.addMissingAtoms()
         displayAddHydrogensPage()
         displayAddHydrogensPage()
         return
         return
-    indexInChain = {}
-    for structChain, topChain in zip(fixer.structureChains, fixer.topology.chains()):
-        for structResidue, topResidue in zip(structChain.iter_residues(), topChain.residues()):
-            indexInChain[topResidue] = structResidue.number
     table = ""
     table = ""
     for residue in allResidues:
     for residue in allResidues:
         atoms = []
         atoms = []
@@ -211,7 +205,7 @@ def displayMissingAtomsPage():
             atoms.extend(atom.name for atom in fixer.missingAtoms[residue])
             atoms.extend(atom.name for atom in fixer.missingAtoms[residue])
         if residue in fixer.missingTerminals:
         if residue in fixer.missingTerminals:
             atoms.extend(atom for atom in fixer.missingTerminals[residue])
             atoms.extend(atom for atom in fixer.missingTerminals[residue])
-        table += '    <tr><td>%d</td><td>%s %d</td><td>%s</td></tr>\n' % (residue.chain.index+1, residue.name, indexInChain[residue], ', '.join(atoms))
+        table += '    <tr><td>%s</td><td>%s %s</td><td>%s</td></tr>\n' % (residue.chain.id, residue.name, residue.id, ', '.join(atoms))
     uiserver.setContent(header+loadHtmlFile("addHeavyAtoms.html") % table)
     uiserver.setContent(header+loadHtmlFile("addHeavyAtoms.html") % table)
 
 
 def displayAddHydrogensPage():
 def displayAddHydrogensPage():

tests/test_removechains.py

@@ -9,7 +9,7 @@ def remove_chain_ids_and_verify(pdbid, chain_ids_to_remove, expected_chain_ids_r
     # Remove specified chains.
     # Remove specified chains.
     fixer.removeChains(chainIds=chain_ids_to_remove)
     fixer.removeChains(chainIds=chain_ids_to_remove)
     # Check to make sure asserted chains remain.
     # Check to make sure asserted chains remain.
-    chain_ids_remaining = [c.chain_id for c in fixer.structureChains]
+    chain_ids_remaining = [c.id for c in fixer.topology.chains()]
     assert_items_equal(chain_ids_remaining, expected_chain_ids_remaining)
     assert_items_equal(chain_ids_remaining, expected_chain_ids_remaining)
 
 
 def test_removechain_ids():
 def test_removechain_ids():
@@ -25,7 +25,7 @@ def remove_chain_indices_and_verify(pdbid, chain_indices_to_remove, expected_cha
     # Remove specified chains.
     # Remove specified chains.
     fixer.removeChains(chainIndices=chain_indices_to_remove)
     fixer.removeChains(chainIndices=chain_indices_to_remove)
     # Check to make sure asserted chains remain.
     # Check to make sure asserted chains remain.
-    chain_ids_remaining = [c.chain_id for c in fixer.structureChains]
+    chain_ids_remaining = [c.id for c in fixer.topology.chains()]
     assert_items_equal(chain_ids_remaining, expected_chain_ids_remaining)
     assert_items_equal(chain_ids_remaining, expected_chain_ids_remaining)
 
 
 def test_removechain_indices():
 def test_removechain_indices():