@Override
protected void finish() {
multiCollector.finish();
final MetricsFile<InsertSizeMetrics, Integer> file = getMetricsFile();
multiCollector.addAllLevelsToFile(file);
if (file.getNumHistograms() == 0) {
// can happen if user sets MINIMUM_PCT = 0.5, etc.
log.warn(
"All data categories were discarded because they contained < "
+ MINIMUM_PCT
+ " of the total aligned paired data.");
final InsertSizeMetricsCollector.PerUnitInsertSizeMetricsCollector allReadsCollector =
(InsertSizeMetricsCollector.PerUnitInsertSizeMetricsCollector)
multiCollector.getAllReadsCollector();
log.warn(
"Total mapped pairs in all categories: "
+ (allReadsCollector == null
? allReadsCollector
: allReadsCollector.getTotalInserts()));
} else {
file.write(OUTPUT);
final int rResult;
if (HISTOGRAM_WIDTH == null) {
rResult =
RExecutor.executeFromClasspath(
HISTOGRAM_R_SCRIPT,
OUTPUT.getAbsolutePath(),
HISTOGRAM_FILE.getAbsolutePath(),
INPUT.getName());
} else {
rResult =
RExecutor.executeFromClasspath(
HISTOGRAM_R_SCRIPT,
OUTPUT.getAbsolutePath(),
HISTOGRAM_FILE.getAbsolutePath(),
INPUT.getName(),
String.valueOf(
HISTOGRAM_WIDTH)); // HISTOGRAM_WIDTH is passed because R automatically sets
// histogram width to the last
// bin that has data, which may be less than HISTOGRAM_WIDTH and confuse the user.
}
if (rResult != 0) {
throw new PicardException(
"R script " + HISTOGRAM_R_SCRIPT + " failed with return code " + rResult);
}
}
}
/** @throws IOException */
private void readFileHeader() throws Exception {
// fisrt four bytes are empty
// it should be zero for new version of control file, backward compatibility
int emptyBytes = this.readFourBytes(inputStream);
if (emptyBytes != 0) {
log.warn(
"The first four bytes are not zero: "
+ emptyBytes
+ ". This is an old format control file.");
this.totalClusters = emptyBytes;
return;
}
// next four bytes should be version and greater or equal to the expected
int version = this.readFourBytes(inputStream);
if (version != this.EXPECTED_CONTROL_VERSION) {
log.error("Unexpected version byte: " + version);
throw new Exception("Unexpected version number in control file");
}
// next four bytes should be the total number of clusters
this.totalClusters = this.readFourBytes(inputStream);
log.info("The total number of clusters: " + this.getTotalClusters());
}
@Override
public Object next() {
try {
int nextByte = this.inputStream.readUnsignedShort();
if (nextByte == -1) {
log.warn(
"There is no more cluster in Control file after cluster "
+ this.getCurrentCluster()
+ " in file "
+ this.getFileName());
return null;
}
this.currentCluster++;
/*
Bit0: always empty (0)
Bit1: was the read identified as a control?
Bit2: was the match ambiguous?
Bit3: did the read match the phiX tag?
Bit4: did the read align to match the phiX tag?
Bit5: did the read match the control index sequence? (specified in controls.fata, TGTCACA)
Bits6,7: reserved for future use
Bits8..15: the report key for the matched record in the controls.fasta file (specified by the REPOControl FilesRT_ KEY metadata)
*/
nextByte = nextByte & 0x2;
if (nextByte != 0) {
this.currentControlClusters++;
}
return new Integer(nextByte);
} catch (IOException ex) {
log.error(ex, "Problem to read control file");
}
return null;
}
