private static void MergeAndFinalizeBam(List <string> perChromPaths, ISamtoolsWrapper samtoolsWrapper, string outFolder, string outputBamName)
{
var finalBams = perChromPaths;
var mergedBam = Path.Combine(outFolder, outputBamName);
Logger.WriteToLog($"Calling samtools cat on {perChromPaths.Count} files to create {mergedBam}.");
samtoolsWrapper.SamtoolsCat(mergedBam, finalBams);
Logger.WriteToLog($"Calling samtools index on {mergedBam}.");
samtoolsWrapper.SamtoolsIndex(mergedBam);
Logger.WriteToLog("Done finalizing bam.");
}
public GeminiWorkflow(IGeminiDataSourceFactory dataSourceFactory,
IGeminiDataOutputFactory dataOutputFactory, GeminiOptions geminiOptions,
GeminiSampleOptions geminiSampleOptions, RealignmentOptions realignmentOptions, StitcherOptions stitcherOptions, string outputDirectory, RealignmentAssessmentOptions realignmentAssessmentOptions, IndelFilteringOptions indelFilteringOptions, ISamtoolsWrapper samtoolsWrapper)
{
_dataSourceFactory = dataSourceFactory;
_dataOutputFactory = dataOutputFactory;
_geminiOptions = geminiOptions;
_geminiSampleOptions = geminiSampleOptions;
_realignmentOptions = realignmentOptions;
_samtoolsWrapper = samtoolsWrapper;
_stitcherOptions = stitcherOptions ?? new StitcherOptions();
_geminiFactory = new GeminiFactory(geminiOptions, indelFilteringOptions);
var bamRealignmentFactory = new BamRealignmentFactory(geminiOptions,
realignmentAssessmentOptions, stitcherOptions, realignmentOptions,
outputDirectory);
_bamRealignmentFactory = bamRealignmentFactory;
}
private static string MergeAndFinalizeBam(List <string> finalBams, ISamtoolsWrapper samtoolsWrapper, string outFolder, string outFileName, bool keepUnmerged = false, bool doIndex = false, bool doSort = true)
{
var mergedBam = Path.Combine(outFolder, outFileName);
var sortedMerged = mergedBam + ".sorted";
var sortedMergedFinal = sortedMerged + ".bam";
Logger.WriteToLog($"Calling cat on {finalBams.Count} with output at {mergedBam}.");
foreach (var finalBam in finalBams)
{
Logger.WriteToLog($"Intermediate bam: {finalBam}\t{new FileInfo(finalBam).Length}B");
}
samtoolsWrapper.SamtoolsCat(mergedBam, finalBams);
if (doSort)
{
Logger.WriteToLog($"Calling samtools sort on {mergedBam}.");
samtoolsWrapper.SamtoolsSort(mergedBam, sortedMerged, 3, 500);
}
if (doIndex)
{
Logger.WriteToLog($"Calling samtools index on {sortedMergedFinal}.");
samtoolsWrapper.SamtoolsIndex(sortedMergedFinal);
}
Logger.WriteToLog("Done finalizing bam.");
if (File.Exists(sortedMergedFinal) && !keepUnmerged)
{
Logger.WriteToLog("Deleting intermediate bams.");
foreach (var finalBam in finalBams)
{
File.Delete(finalBam);
}
File.Delete(mergedBam);
Logger.WriteToLog("Finished deleting intermediate bams.");
}
return(sortedMergedFinal);
}
public void Execute(ICliTaskManager cliTaskManager, Dictionary <string, int> chromRefIds, List <string> cmdLineList, ISamtoolsWrapper samtoolsWrapper)
{
var bamFile = _options.InputBam;
var outMultiPath = _options.OutputDirectory;
var exePath = _options.ExePath;
List <string> paths;
var taskDirectories = new List <string>();
var taskLogDir = Path.Combine(outMultiPath, "TaskLogs");
if (_options.MultiProcess)
{
paths = ExecuteChromosomeJobs(cliTaskManager, chromRefIds, cmdLineList, outMultiPath, taskLogDir, exePath, taskDirectories);
}
else
{
paths = ExecuteChromosomeJobs(chromRefIds, outMultiPath,
taskDirectories);
}
var outBamName = Path.GetFileNameWithoutExtension(bamFile) + ".PairRealigned.bam";
MergeAndFinalizeBam(paths, samtoolsWrapper, outMultiPath,
outBamName);
try
{
CleanUp(outMultiPath, taskDirectories, taskLogDir);
}
catch (Exception e)
{
Logger.WriteExceptionToLog(new Exception($"Error encountered during cleanup step (after analysis completion).", e));
}
}
private static void MergeAndFinalizeBam(Dictionary <string, Dictionary <PairClassification, List <string> > > categorizedAlignments, ISamtoolsWrapper samtoolsWrapper, string outFolder, bool keepUnmerged = false, bool indexFinalBam = false, bool doSort = true)
{
var finalBams = new List <string>();
if (categorizedAlignments.Keys.Count > 1)
{
foreach (var chrom in categorizedAlignments.Keys)
{
var chromFinalBams = categorizedAlignments[chrom].Values;
if (!chromFinalBams.Any())
{
Logger.WriteToLog($"No bams to merge for chromosome {chrom}.");
continue;
}
var sortedMergedChromBam = MergeAndFinalizeBam(chromFinalBams.SelectMany(x => x).ToList(), samtoolsWrapper, outFolder, chrom + "_merged.bam", keepUnmerged: keepUnmerged, doIndex: false, doSort: doSort);
finalBams.Add(sortedMergedChromBam);
}
}
else
{
if (categorizedAlignments.Keys.Any())
{
var chrom = categorizedAlignments.Keys.First();
finalBams = categorizedAlignments[chrom].Values.SelectMany(x => x).ToList();
}
else
{
Logger.WriteToLog("No chromosomes with bams to merge.");
}
}
if (finalBams.Any())
{
MergeAndFinalizeBam(finalBams, samtoolsWrapper, outFolder, "merged.bam", keepUnmerged, indexFinalBam, doSort: doSort);
}
}
