private Set<Component> unionComponentSet(Set<Component> set1, Set<Component> set2) {
if (set1 == null && set2 == null) return Collections.emptySet();
if (set1 == null) return set2;
if (set2 == null) return set1;
Set<Component> set = new HashSet<Component>(set1.size() + set2.size());
set.addAll(set1);
set.addAll(set2);
return set;
}
/**
* Utility method to group child nodes by their names.
*
* @param parent parent node.
* @return Map from name to child nodes.
*/
private static Map<String, List<Node>> getChildNodes(Node parent) {
if (parent == null) return Collections.emptyMap();
Map<String, List<Node>> children = new HashMap<String, List<Node>>();
NodeList nodes = parent.getChildNodes();
int nNodes = nodes.getLength();
for (int i = 0; i < nNodes; i++) {
Node node = nodes.item(i);
if (node.getNodeType() != Node.ELEMENT_NODE) continue;
String name = node.getNodeName();
List<Node> namesakes = children.get(name);
if (namesakes == null) {
namesakes = new ArrayList<Node>();
children.put(name, namesakes);
}
namesakes.add(node);
}
return children;
}
/** Merge identified modified compounds if linked. */
private void mergeModComps(List<ModifiedCompound> modComps) {
TreeSet<Integer> remove = new TreeSet<Integer>();
int n = modComps.size();
for (int icurr = 1; icurr < n; icurr++) {
ModifiedCompound curr = modComps.get(icurr);
String id = curr.getModification().getId();
if (ProteinModificationRegistry.getById(id).getCategory() != ModificationCategory.UNDEFINED)
continue;
// find linked compounds that before curr
// List<Integer> merging = new ArrayList<Integer>();
int ipre = 0;
for (; ipre < icurr; ipre++) {
if (remove.contains(ipre)) continue;
ModifiedCompound pre = modComps.get(ipre);
if (!Collections.disjoint(pre.getGroups(false), curr.getGroups(false))) {
break;
}
}
if (ipre < icurr) {
ModifiedCompound mcKeep = modComps.get(ipre);
// merge modifications of the same type
if (mcKeep.getModification().getId().equals(id)) {
// merging the current one to the previous one
mcKeep.addAtomLinkages(curr.getAtomLinkages());
remove.add(icurr);
}
}
}
Iterator<Integer> it = remove.descendingIterator();
while (it.hasNext()) {
modComps.remove(it.next().intValue());
}
}
/**
* @param modifications a set of {@link ProteinModification}s.
* @param residues
* @param ligands
* @param saveTo save result to
* @return map from component to list of corresponding residues in the chain.
*/
private void addModificationGroups(
final Set<ProteinModification> modifications,
final List<Group> residues,
final List<Group> ligands,
final Map<Component, Set<Group>> saveTo) {
if (residues == null || ligands == null || modifications == null) {
throw new IllegalArgumentException("Null argument(s).");
}
Map<Component, Set<Component>> mapSingleMultiComps = new HashMap<Component, Set<Component>>();
for (ProteinModification mod : modifications) {
ModificationCondition condition = mod.getCondition();
for (Component comp : condition.getComponents()) {
for (String pdbccId : comp.getPdbccIds()) {
Component single =
Component.of(Collections.singleton(pdbccId), comp.isNTerminal(), comp.isCTerminal());
Set<Component> mult = mapSingleMultiComps.get(single);
if (mult == null) {
mult = new HashSet<Component>();
mapSingleMultiComps.put(single, mult);
}
mult.add(comp);
}
}
}
{
// ligands
Set<Component> ligandsWildCard = mapSingleMultiComps.get(Component.of("*"));
for (Group group : ligands) {
String pdbccId = group.getPDBName().trim();
Set<Component> comps = mapSingleMultiComps.get(Component.of(pdbccId));
for (Component comp : unionComponentSet(ligandsWildCard, comps)) {
Set<Group> gs = saveTo.get(comp);
if (gs == null) {
gs = new LinkedHashSet<Group>();
saveTo.put(comp, gs);
}
gs.add(group);
}
}
}
{
// residues
if (residues.isEmpty()) {
return;
}
Set<Component> residuesWildCard = mapSingleMultiComps.get(Component.of("*"));
// for all residues
for (Group group : residues) {
String pdbccId = group.getPDBName().trim();
Set<Component> comps = mapSingleMultiComps.get(Component.of(pdbccId));
for (Component comp : unionComponentSet(residuesWildCard, comps)) {
Set<Group> gs = saveTo.get(comp);
if (gs == null) {
gs = new LinkedHashSet<Group>();
saveTo.put(comp, gs);
}
gs.add(group);
}
}
// for N-terminal
int nRes = residues.size();
int iRes = 0;
Group res;
do {
// for all ligands on N terminal and the first residue
res = residues.get(iRes++);
Set<Component> nTermWildCard = mapSingleMultiComps.get(Component.of("*", true, false));
Set<Component> comps = mapSingleMultiComps.get(Component.of(res.getPDBName(), true, false));
for (Component comp : unionComponentSet(nTermWildCard, comps)) {
Set<Group> gs = saveTo.get(comp);
if (gs == null) {
gs = new LinkedHashSet<Group>();
saveTo.put(comp, gs);
}
gs.add(res);
}
} while (iRes < nRes && ligands.contains(res));
// for C-terminal
iRes = residues.size() - 1;
do {
// for all ligands on C terminal and the last residue
res = residues.get(iRes--);
Set<Component> cTermWildCard = mapSingleMultiComps.get(Component.of("*", false, true));
Set<Component> comps = mapSingleMultiComps.get(Component.of(res.getPDBName(), false, true));
for (Component comp : unionComponentSet(cTermWildCard, comps)) {
Set<Group> gs = saveTo.get(comp);
if (gs == null) {
gs = new LinkedHashSet<Group>();
saveTo.put(comp, gs);
}
gs.add(res);
}
} while (iRes >= 0 && ligands.contains(res));
}
}
/**
* Identify a set of modifications in a a chains.
*
* @param chain query {@link Chain}.
* @param potentialModifications query {@link ProteinModification}s.
*/
public void identify(final Chain chain, final Set<ProteinModification> potentialModifications) {
identify(Collections.singletonList(chain), potentialModifications);
}
/**
* Identify all registered modifications in a chain.
*
* @param chain query {@link Chain}.
*/
public void identify(final Chain chain) {
identify(Collections.singletonList(chain));
}