public void parseInputFile(File inputFile) throws IOException {
geneFeatures.clear();
otherRecords.clear();
try {
GFFEntrySet gffEntries = GFFTools.readGFF(inputFile);
Iterator itr = gffEntries.lineIterator();
int count = 0;
int intronFeatures = 0;
LinkedList<GFFRecord> cdsRecs = new LinkedList<GFFRecord>();
while (itr.hasNext()) {
Object val = itr.next();
if (val instanceof GFFRecord) {
GFFRecord rec = (GFFRecord) val;
count += 1;
if (rec.getFeature().endsWith("gene")) {
GeneFeatures gf = new GeneFeatures(rec);
geneFeatures.put(gf.id, gf);
} else if (rec.getFeature().equals("CDS")) {
cdsRecs.addLast(rec);
} else {
otherRecords.add(rec);
}
}
}
for (GFFRecord rec : cdsRecs) {
Map<String, List<String>> attrs = decodeAttrMap(rec);
if (geneFeatures.containsKey(attrs.get("Parent").get(0))) {
geneFeatures.get(attrs.get("Parent").get(0)).addCDS(rec, attrs);
} else {
System.err.println("Unknown CDS Parent: " + attrs.get("Parent").get(0));
}
}
for (String k : geneFeatures.keySet()) {
GeneFeatures gf = geneFeatures.get(k);
if (gf.cds != null && gf.cds.size() > 1) {
intronFeatures++;
}
}
System.err.println("# GFF Records: " + count);
System.err.println("# Gene Feature Sets: " + geneFeatures.size());
System.err.println("\t# Intron-Features: " + intronFeatures);
} catch (ParserException e) {
e.printStackTrace();
} catch (BioException e) {
e.printStackTrace();
}
}
public static Pair<String, List<String>> decodeKeyValues(String str) {
int index = str.indexOf("=");
if (index == -1) {
return null;
}
String k = str.substring(0, index);
LinkedList<String> vlist = new LinkedList<String>();
try {
String vstr = URLDecoder.decode(str.substring(index + 1, str.length()), "UTF-8");
String[] array = vstr.split(",");
for (int i = 0; i < array.length; i++) {
vlist.addLast(array[i]);
}
return new Pair<String, List<String>>(k, vlist);
} catch (UnsupportedEncodingException e) {
System.err.println("BAD STRING " + str);
e.printStackTrace();
return null;
} catch (Exception e) {
System.err.println("BAD STRING " + str);
return null;
}
}
public void addCDS(GFFRecord rec, Map<String, List<String>> attrs) {
if (!rec.getFeature().equals("CDS")) {
throw new IllegalArgumentException();
}
if (attrs == null) {
attrs = decodeAttrMap(rec);
}
if (!attrs.containsKey("Parent")) {
throw new IllegalArgumentException();
}
if (!attrs.get("Parent").contains(id)) {
throw new IllegalArgumentException();
}
cds.addLast(rec);
}