public Bed2Bar(String[] args) {
try {
processArgs(args);
// load Window[]
for (int i = 0; i < bedFiles.length; i++) {
bedFile = bedFiles[i];
System.out.println("Parsing " + bedFile.getName());
bedLinesHash = Bed.parseBedFile(bedFile, true, false);
if (bedLinesHash == null || bedLinesHash.size() == 0) {
System.out.println("Problem parsing bed file, skipping!");
continue;
}
barDirectory = IO.makeDirectory(bedFile, "");
File bedOutFile =
new File(Misc.removeExtension(bedFile.toString()) + "_" + threshold + "_Filt.bed");
bedOut = new PrintWriter(new FileWriter(bedOutFile));
makeStairStepBarFiles();
}
bedOut.close();
System.out.println("\nDone!\n");
} catch (IOException e) {
e.printStackTrace();
}
}
public String fetchCleanupReviewDirLinkGenXml() {
String s = null;
try {
s = // Organize and Clean Up Section
"<ReviewDirGenerator class='operator.ReviewDirGenerator' destination.dir='"
+ outputDirectory.getCanonicalPath()
+ "' sample='"
+ outputDirectory.getName()
+ "' submitter='"
+ submitter
+ "' analysis.type='"
+ analysisType
+ "'> \n"
+ " <finalVariants /> \n"
+ " <ViewerCSV /> \n"
+ " <InstanceLog class='buffer.InstanceLogFile' /> \n"
+ " <finalBAM /> \n"
+ " <QCOutputFile /> \n"
+ " <bedForVars /> \n"
+ "</ReviewDirGenerator> \n\n";
} catch (IOException e) {
e.printStackTrace();
Misc.printErrAndExit("\nError fetching clean up and review dir links.\n");
}
return s;
}
public void checkPrintFiles() {
LinkedHashMap<String, File> nameFile = new LinkedHashMap<String, File>();
nameFile.put("Reference directory", referenceDir);
nameFile.put("Pipeline.jar", pJar);
nameFile.put("Pipeline properties", truncPipePropFile);
nameFile.put("Coverage QC bed", bedForCoverageQC);
nameFile.put("Variant calling bed", bedForVarCalling);
nameFile.put("Fasta genome reference", fastaReference);
nameFile.put("Unfiltered bam", unfilteredBam);
nameFile.put("Final bam", finalBam);
nameFile.put("Final vcf", finalVcf);
if (webRootForLinks != null) nameFile.put("Web links dir", webRootForLinks);
nameFile.put("Final output dir", outputDirectory);
boolean missingFile = false;
System.out.println("\nResources (name exists path):");
for (String name : nameFile.keySet()) {
File f = nameFile.get(name);
boolean fExists = true;
if (f == null) {
fExists = false;
missingFile = true;
} else {
fExists = f.exists();
if (fExists == false) missingFile = true;
}
System.out.println(name + "\t" + fExists + "\t" + f);
}
if (missingFile) Misc.printErrAndExit("\nMissing resources! See above.");
}
/** This method will process each argument and assign new varibles */
public void processArgs(String[] args) {
File file = null;
Pattern pat = Pattern.compile("-[a-z]");
for (int i = 0; i < args.length; i++) {
String lcArg = args[i].toLowerCase();
Matcher mat = pat.matcher(lcArg);
if (mat.matches()) {
char test = args[i].charAt(1);
try {
switch (test) {
case 'f':
file = new File(args[i + 1]);
i++;
break;
case 'v':
genomeVersion = args[i + 1];
i++;
break;
case 's':
sumScores = true;
break;
case 't':
threshold = Float.parseFloat(args[++i]);
break;
case 'h':
printDocs();
System.exit(0);
default:
Misc.printExit("\nError, unknown option! " + mat.group());
}
} catch (Exception e) {
System.out.print("\nSorry, something doesn't look right with this parameter: -" + test);
System.out.println();
System.exit(0);
}
}
}
if (file == null || file.exists() == false)
Misc.printErrAndExit("Problem finding your bed files!\n");
// pull files
File[][] tot = new File[3][];
tot[0] = IO.extractFiles(file, ".bed");
tot[1] = IO.extractFiles(file, ".bed.zip");
tot[2] = IO.extractFiles(file, ".bed.gz");
bedFiles = IO.collapseFileArray(tot);
if (bedFiles == null || bedFiles.length == 0)
Misc.printErrAndExit("Problem finding your xxx.bed(.zip/.gz OK) files!\n");
// genome version
if (genomeVersion == null)
Misc.printErrAndExit(
"Please enter a genome version (e.g. H_sapiens_Mar_2006, see http://genome.ucsc.edu/FAQ/FAQreleases\n");
}
private void checkForGzippedVcf() {
String vcfName = finalVcf.getName();
if (vcfName.endsWith(".gz")) {
File uncomp = new File(finalVcf.getParentFile(), vcfName.substring(0, vcfName.length() - 3));
if (uncomp.exists() == false) {
deleteTempVcf = true;
if (IO.uncompress(finalVcf, uncomp) == null)
Misc.printErrAndExit("ERROR: failed to uncompress -> " + finalVcf);
}
finalVcf = uncomp;
}
}
/** This method will process each argument and assign new varibles */
public void processArgs(String[] args) {
Pattern pat = Pattern.compile("-[a-z]");
File dir = null;
for (int i = 0; i < args.length; i++) {
String lcArg = args[i].toLowerCase();
Matcher mat = pat.matcher(lcArg);
if (mat.matches()) {
char test = args[i].charAt(1);
try {
switch (test) {
case 'f':
dir = new File(args[i + 1]);
i++;
break;
case 'v':
genomeVersion = args[i + 1];
i++;
break;
case 's':
strand = args[++i];
break;
case 't':
stairStep = true;
break;
case 'h':
printDocs();
System.exit(0);
default:
Misc.printExit("\nError: unknown option! " + mat.group());
}
} catch (Exception e) {
Misc.printExit(
"\nSorry, something doesn't look right with this parameter: -" + test + "\n");
}
}
}
if (dir == null || dir.canRead() == false)
Misc.printExit("\nError: cannot find or read your sgr file/ directory.\n");
File[][] tot = new File[3][];
tot[0] = IO.extractFiles(dir, ".sgr");
tot[1] = IO.extractFiles(dir, ".sgr.zip");
tot[2] = IO.extractFiles(dir, ".sgr.gz");
sgrFiles = IO.collapseFileArray(tot);
if (sgrFiles == null || sgrFiles.length == 0)
Misc.printExit("\nError: cannot find your xxx.sgr.zip file(s)");
if (genomeVersion == null)
Misc.printExit(
"\nError: you must supply a genome version. Goto http://genome.ucsc.edu/cgi-"
+ "bin/hgGateway load your organism to find the associated genome version.\n");
}
public String fetchVariantOutputXml() {
String s =
// Write variants to csv file no need for including bad.region, use the USeq VCFRegionMarker
// instead
// "<ViewerFile class='plugins.writers.varviewer.VarViewerWriter' anno.keys='bad.region'>
// \n"+
"<ViewerFile class='plugins.writers.varviewer.VarViewerWriter'> \n"
+ " <VariantPool /> \n"
+ " <Genes /> \n"
+ " <ViewerCSV class='buffer.CSVFile' filename='"
+ outputDirectory.getName()
+ "_annotated.csv' /> \n"
+ "</ViewerFile> \n\n"
+
// Create Final JsonQC Section (needs to happen after variant calls)
"<QCtoJSON class='operator.qc.QCtoJSON'> \n"
+ " <rawBamMetrics /> \n"
+ " <rawdocmetrics /> \n"
+ " <finalBamMetrics /> \n"
+ " <finaldocmetrics /> \n"
+ " <bedForCoverageQC /> \n"
+ " <filteredVars /> \n"
+ " <NoCallCSV /> \n"
+ " <QCOutputFile class='buffer.TextBuffer' filename='qc.json' /> \n"
+ "</QCtoJSON> \n"
+ "\n";
return s;
}
// constructor
public Sgr2Bar(String[] args) {
try {
// check for args
processArgs(args);
System.out.println("Genome version -> " + genomeVersion);
System.out.println("Strand -> " + strand);
System.out.println("Stair Step? -> " + stairStep);
System.out.println();
// load tagValues
if (stairStep) tagValues.put(BarParser.GRAPH_TYPE_TAG, BarParser.GRAPH_TYPE_STAIRSTEP);
else tagValues.put(BarParser.GRAPH_TYPE_TAG, BarParser.GRAPH_TYPE_BAR);
for (int x = 0; x < sgrFiles.length; x++) {
System.out.println("\tLoading -> " + sgrFiles[x]);
GrGraph[] grs = GrGraph.loadSgrFile(sgrFiles[x]);
// make save directory
String dirName;
dirName = Misc.removeExtension(sgrFiles[x].getCanonicalPath());
File dir = new File(dirName);
dir.mkdir();
// print bar files
System.out.println("\tSaving...");
for (int i = 0; i < grs.length; i++) {
File barFile = new File(dir, grs[i].getChromosome() + ".bar");
barParser.writeBarFile(
barFile,
grs[i].getChromosome(),
genomeVersion,
strand.charAt(0),
grs[i].getBasePositions(),
grs[i].getValues(),
tagValues);
}
}
System.out.println("\nDone!\n");
} catch (IOException e) {
e.printStackTrace();
Misc.printErrAndExit("\nFailed to parse file!");
}
}
public void splitIndex(String name, String seq) {
int start = 0;
boolean go = true;
while (go) {
int end = start + numberOfBases;
if (end >= seq.length()) {
end = seq.length();
go = false;
}
String subSeq = seq.substring(start, end);
File binarySeq = new File(indexDirectory, name + "_" + start + "-" + (end - 1));
if (binarySeq.exists()) {
System.out.println("WARNING, " + binarySeq + " already exists, skipping!");
return;
}
Seq.writeBinarySequence(subSeq, binarySeq);
start = end;
}
}
private void buildXmlPropertiesFile() {
System.out.println(
"\nBuilding and checking your pipeline properties file -> " + truncPipePropFile);
StringBuilder toPrint = new StringBuilder();
// walk through the prop file
String[] prop = IO.loadFileIntoStringArray(truncPipePropFile);
Pattern val = Pattern.compile("(<entry key.+>)([D|A|B].*)</entry>");
boolean missingFile = false;
for (String s : prop) {
// does it match a file needing prepending? Data/, Apps/, Bed/
Matcher mat = val.matcher(s);
if (mat.matches()) {
File test = new File(referenceDir, mat.group(2));
if (test.exists()) {
System.out.println("Found\t" + test);
toPrint.append(mat.group(1));
toPrint.append(test.toString());
toPrint.append("</entry>");
} else {
System.out.println("Misssing\t" + test);
missingFile = true;
}
}
// threads?
else if (s.contains("threads"))
toPrint.append("<entry key=\"threads\">" + threads + "</entry>");
// nope just save it and add a line return
else toPrint.append(s);
toPrint.append("\n");
}
// anything missing? if so exit
if (missingFile)
Misc.printErrAndExit(
"\nFailed to find all of the files in your properties file, see above.\n");
// OK, write it out
completePipelinePropFile = new File(outputDirectory, "pipelineProperties.xml");
if (IO.writeString(toPrint.toString(), completePipelinePropFile) == false)
Misc.printErrAndExit("Problem writing -> " + truncPipePropFile);
}
public String fetchWebRootLinksXml() {
String s =
"<CreateBAMLink class='operator.LinkCreator' sample='"
+ outputDirectory.getName()
+ "' web.root='"
+ webRootForLinks
+ "' result.dir='links/'> \n"
+ " <finalBAM /> \n"
+ " <bedForVars /> \n"
+ "</CreateBAMLink> \n\n";
return s;
}
private void executePipelineJob() {
String[] cmd = null;
try {
// write out tempTemplate
File template = new File(outputDirectory, "pipelineTemplate.xml");
if (IO.writeString(xmlTemplate, template) == false)
Misc.printErrAndExit("Problem writing -> " + template);
// build and execute cmd
cmd =
new String[] {
"java",
"-jar",
"-Xmx2G",
pJar.getCanonicalPath(),
"-props",
completePipelinePropFile.getCanonicalPath(),
template.getCanonicalPath()
};
String stringCmd = Misc.stringArrayToString(cmd, " ");
System.out.println("\nExecuting:\n" + stringCmd);
System.out.println("\nPipelineOutput:");
String[] out = IO.executeViaProcessBuilder(cmd, true);
// check output for possible errors
for (String line : out) {
String lcLine = line.toLowerCase();
// watch out for cases where error is mentioned in a warning output line.
if (lcLine.contains("error") && lcLine.startsWith("warning") == false)
Misc.printErrAndExit(
"\n\nERROR found in Pipeline.jar output, see above. Aborting!\n" + line);
}
} catch (Exception e) {
e.printStackTrace();
Misc.printErrAndExit("ERROR: executing " + Misc.stringArrayToString(cmd, " "));
}
}
/** This method will process each argument and assign new varibles */
public void processArgs(String[] args) {
Pattern pat = Pattern.compile("-[a-z]");
for (int i = 0; i < args.length; i++) {
String lcArg = args[i].toLowerCase();
Matcher mat = pat.matcher(lcArg);
if (mat.matches()) {
char test = args[i].charAt(1);
try {
switch (test) {
case 'f':
fastas = IO.extractFiles(args[i + 1], "fasta");
i++;
break;
case 'n':
numberOfBases = Integer.parseInt(args[i + 1]);
i++;
break;
case 'i':
indexDirectory = new File(args[i + 1]);
i++;
break;
case 'h':
printDocs();
System.exit(0);
default:
Misc.printExit("\nError: unknown option! " + mat.group());
}
} catch (Exception e) {
Misc.printExit(
"\nSorry, something doesn't look right with this parameter: -" + test + "\n");
}
}
}
// make index directory?
if (indexDirectory == null) {
indexDirectory = new File(fastas[0].getParentFile(), "IndexedSequences");
indexDirectory.mkdir();
}
}
/**
* Makes a stair step heat map from an array of windows in bar format. One per chromosome. Don't
* forget to set the barDirectory and score Index!!!!!!!
*/
public void makeStairStepBarFiles() {
// make bar parser
BarParser bp = new BarParser();
bp.setZipCompress(true);
HashMap<String, String> tagVals = new HashMap<String, String>();
tagVals.put(BarParser.GRAPH_TYPE_TAG, BarParser.GRAPH_TYPE_STAIRSTEP);
tagVals.put(BarParser.GRAPH_TYPE_COLOR_TAG, "#FF00FF"); // fusha
tagVals.put(BarParser.SOURCE_TAG, bedFile.toString());
// for each chromosome
System.out.print("Printing... ");
Iterator<String> it = bedLinesHash.keySet().iterator();
while (it.hasNext()) {
chromosome = it.next();
System.out.print(chromosome + " ");
windows = bedLinesHash.get(chromosome);
// add blocks
assembleBlocks();
// balance by adding max or min at zero base
balanceValues();
// write bar file
File barFile = new File(barDirectory, chromosome + ".bar");
bp.writeBarFile(
barFile,
chromosome,
genomeVersion,
'.',
Num.arrayListOfIntegerToInts(bases),
Num.arrayListOfFloatToArray(values),
tagVals);
// clear ArrayLists
bases.clear();
values.clear();
}
System.out.println();
}
// remove temp files, the pipeline.jar operator isn't working.
public void deleteTempFiles() {
System.out.println("\nRemoving these temp files:");
File workingDir = new File(System.getProperty("user.dir"));
File[] toExamine = workingDir.listFiles();
for (File f : toExamine) {
boolean d = false;
String n = f.getName();
if (n.startsWith("pipeinstancelog")) d = true;
else if (n.contains(".DOC.sample")) d = true;
else if (n.contains("allDepths.")) d = true;
else if (n.startsWith("nocalls.")) d = true;
else if (n.startsWith("snpeff.")) d = true;
else if (n.contains("plice")) d = true;
if (d) {
System.out.println("\t" + n);
f.deleteOnExit();
}
}
// delete the temp uncompressed vcf (required by Pipeline.jar)
if (deleteTempVcf) System.out.println("\t" + finalVcf.getName());
}
/** This method will process each argument and assign new variables */
public void processArgs(String[] args) {
Pattern pat = Pattern.compile("-[a-z]");
System.out.println(
"\n" + IO.fetchUSeqVersion() + " Arguments: " + Misc.stringArrayToString(args, " ") + "\n");
for (int i = 0; i < args.length; i++) {
String lcArg = args[i].toLowerCase();
Matcher mat = pat.matcher(lcArg);
if (mat.matches()) {
char test = args[i].charAt(1);
try {
switch (test) {
case 'o':
jobId = args[++i];
break;
case 's':
sampleId = args[++i];
break;
case 'm':
submitter = args[++i];
break;
case 'y':
analysisType = args[++i];
break;
case 'w':
webRootForLinks = new File(args[++i]);
break;
case 'e':
snpEffGenome = args[++i];
break;
case 'i':
minimumReadDepth = args[++i];
break;
case 't':
threads = args[++i];
break;
case 'l':
uploadVarsToNGSWeb = true;
break;
case 'j':
pJar = new File(args[++i]);
break;
case 'p':
truncPipePropFile = new File(args[++i]);
break;
case 'q':
bedForCoverageQC = new File(args[++i]);
break;
case 'b':
bedForVarCalling = new File(args[++i]);
break;
case 'r':
fastaReference = new File(args[++i]);
break;
case 'u':
unfilteredBam = new File(args[++i]);
break;
case 'f':
finalBam = new File(args[++i]);
break;
case 'v':
finalVcf = new File(args[++i]);
break;
case 'd':
outputDirectory = new File(args[++i]);
break;
case 'c':
referenceDir = new File(args[++i]);
break;
case 'h':
printDocs();
System.exit(0);
default:
Misc.printErrAndExit("\nProblem, unknown option! " + mat.group());
}
} catch (Exception e) {
Misc.printErrAndExit(
"\nSorry, something doesn't look right with this parameter: -" + test + "\n");
}
}
}
// check output dir and if needed set sampleId
if (outputDirectory != null) {
outputDirectory.mkdirs();
if (sampleId.length() == 0) sampleId = outputDirectory.getName();
}
// check root directory if needed
if (webRootForLinks != null) {
if (webRootForLinks.exists() == false) webRootForLinks.mkdirs();
// links dir?
File links = new File(webRootForLinks, "links");
if (links.exists() == false) links.mkdirs();
}
// look for required fields and files
checkPrintFields();
checkPrintFiles();
checkForGzippedVcf();
}