/**
* Attempts to find an EC number for every domain in {@code census}, skipping those which are
* already listed in this ECFinder. Prints to {@code output} periodically.
*/
public void buildFromCensus(CensusResultList census, File output) {
ScopDatabase scop = ScopFactory.getSCOP();
CensusSignificance sig = CensusSignificanceFactory.forCeSymmOrd();
int i = 0;
for (CensusResult result : census.getEntries()) {
try {
String scopId = result.getId();
if (this.ecsByAsymmDomain.containsKey(scopId)
|| this.ecsBySymmDomain.containsKey(scopId)
|| ecsByUnknownDomain.contains(scopId)) {
continue;
}
ScopDomain domain = scop.getDomainByScopID(scopId);
if (domain == null) {
logger.error(result.getId() + " is null");
continue;
}
// got a result; what's its EC?
// we need to find the correct polymers corresponding to the domain
// note that this still isn't perfect, since we don't know what part of the polymer actually
// does the function
List<RCSBPolymer> polymers = new ArrayList<RCSBPolymer>();
Set<String> chains = domain.getChains();
RCSBDescription desc = RCSBDescriptionFactory.get(domain.getPdbId());
for (RCSBPolymer polymer : desc.getPolymers()) {
for (Character chain : polymer.getChains()) {
if (chains.contains(String.valueOf(chain))) {
polymers.add(polymer);
break;
}
}
}
// get the EC numbers
// use a set because we don't want > 1 just because we have duplicates
NavigableSet<String> ecs = new TreeSet<String>();
for (RCSBPolymer polymer : polymers) {
String ec = polymer.getEnzClass();
if (ec != null) ecs.add(ec);
}
if (ecs.size() == 1) {
String ec = ecs.first();
if (sig.isSignificant(result)) {
ecsBySymmDomain.put(scopId, ec);
} else {
ecsByAsymmDomain.put(scopId, ec);
}
} else if (ecs.size() > 1) {
logger.info(
"Found different EC numbers for "
+ domain.getScopId()); // technically, this doesn't mean anything's wrong
} else {
// logger.debug("Didn't find EC for " + scopId);
ecsByUnknownDomain.add(scopId);
}
if (i > 0 && i % 100 == 0) {
print(output);
logger.debug("Working on #" + i);
}
} catch (RuntimeException e) {
e.printStackTrace();
logger.error(e);
} finally {
i++;
}
}
}
public void rebuild(File file, File output) throws IOException {
buildFromCensus(CensusResultList.fromXML(file), output);
}