@Override
public ExampleSet read() throws OperatorException {
FileInputStream inStream = null;
try {
inStream = new FileInputStream(getParameterAsFile(PARAMETER_FASTA_FILE_NAME));
} catch (FileNotFoundException e) {
// TODO: "Fill"
}
FastaReader<DNASequence, NucleotideCompound> fastaReader =
new FastaReader<DNASequence, NucleotideCompound>(
inStream,
new GenericFastaHeaderParser<DNASequence, NucleotideCompound>(),
new DNASequenceCreator(DNACompoundSet.getDNACompoundSet()));
LinkedHashMap<String, DNASequence> b = null;
try {
b = fastaReader.process();
} catch (Exception e) {
// TODO: "Fill"
}
String data[][] = new String[0][];
if (b != null) {
data = new String[b.size()][2];
int i = 0;
for (Map.Entry<String, DNASequence> entry : b.entrySet()) {
data[i][0] = entry.getValue().getOriginalHeader();
data[i][1] = entry.getValue().getSequenceAsString();
i++;
}
}
ExampleSet outSet = ExampleSetFactory.createExampleSet(data);
outSet.getAttributes().get("att1").setName("DNA name");
outSet.getAttributes().get("att2").setName("Chain");
return outSet;
}
public static void main(String[] args) {
DNASequence dnaSequence = new DNASequence("ATCG");
System.out.println(dnaSequence.toString());
SequenceStringProxyLoader<NucleotideCompound> sequenceStringProxyLoader =
new SequenceStringProxyLoader("GCTA", DNACompoundSet.getDNACompoundSet());
DNASequence dnaSequenceFromProxy = new DNASequence(sequenceStringProxyLoader);
System.out.println(dnaSequenceFromProxy.toString());
}
public DNASequence(SequenceProxyLoader<NucleotideCompound> proxyLoader) {
super(proxyLoader, DNACompoundSet.getDNACompoundSet());
}
public DNASequence(String seqString) {
super(seqString, DNACompoundSet.getDNACompoundSet());
}
@Override
public ExampleSet read() throws OperatorException {
FileInputStream inStream = null;
try {
inStream = new FileInputStream(getParameterAsFile(PARAMETER_FASTA_NAME));
} catch (FileNotFoundException e) {
// TODO: "Fill"
}
FastaReader<DNASequence, NucleotideCompound> fastaReader =
new FastaReader<DNASequence, NucleotideCompound>(
inStream,
new GenericFastaHeaderParser<DNASequence, NucleotideCompound>(),
new DNASequenceCreator(DNACompoundSet.getDNACompoundSet()));
LinkedHashMap<String, DNASequence> genome = null;
try {
genome = fastaReader.process();
} catch (Exception e) {
// TODO: "Fill"
}
ORMFinder finder = new ORMFinder();
ArrayList<DNASequence> startTrip = new ArrayList<DNASequence>();
startTrip.add(new DNASequence("ATG")); // +
startTrip.add(new DNASequence("TTG")); // +
startTrip.add(new DNASequence("CTG")); // +
startTrip.add(new DNASequence("ATT")); // +
startTrip.add(new DNASequence("ATC")); // --
startTrip.add(new DNASequence("ATA")); // --
startTrip.add(new DNASequence("GTG")); // +
ArrayList<DNASequence> stopTrip = new ArrayList<DNASequence>();
stopTrip.add(new DNASequence("TAA"));
stopTrip.add(new DNASequence("TAG"));
stopTrip.add(new DNASequence("TGA"));
ArrayList<String> orfs = null;
ArrayList<String> starts = null;
ArrayList<String> stops = null;
for (Map.Entry<String, DNASequence> entry : genome.entrySet()) {
orfs =
finder.find(
entry.getValue(), getParameterAsInt(PARAMETER_ORF_MIN_LENGTH), startTrip, stopTrip);
starts = finder.getStarts();
stops = finder.getStops();
}
String data[][] = null;
if (orfs != null) {
data = new String[orfs.size()][4];
}
for (int i = 0; (i < orfs.size()) && (i < starts.size()) && (i < stops.size()); i++) {
int annotationEnd = orfs.get(i).indexOf(")");
String currentAnotation = orfs.get(i).substring(0, annotationEnd + 1);
String currentOrf = orfs.get(i).substring(annotationEnd + 2, orfs.get(i).length() - 1);
data[i][0] = currentAnotation;
data[i][1] = currentOrf;
data[i][2] = starts.get(i).toString();
data[i][3] = stops.get(i).toString();
}
ExampleSet outSet = ExampleSetFactory.createExampleSet(data);
outSet.getAttributes().get("att1").setName("Annotation");
outSet.getAttributes().get("att2").setName("Chain");
outSet.getAttributes().get("att3").setName("Start position");
outSet.getAttributes().get("att4").setName("Stop position");
return outSet;
}