/// <summary>
/// Create an index from a sorted BAM file
/// </summary>
/// <param name="bamFilePath">Path to BAM file</param>
public BamIndex(string bamFilePath)
{
_numUnalignedWithoutCoordinates = 0;
// allocate space for the reference index
using (var reader = new BamReader(bamFilePath))
{
List<GenomeMetadata.SequenceMetadata> references = reader.GetReferences();
Initialize(references.Count, reader.Tell());
BamAlignment alignment = new BamAlignment();
while (reader.GetNextAlignment(ref alignment, true))
{
if (!UpdateReferenceIndex(ref alignment, reader.Tell())) break;
}
// perform some post-processing on the index
PostProcessing(reader.Tell());
if (_hasUnalignedReads)
{
while (reader.GetNextAlignment(ref alignment, true)) ++_numUnalignedWithoutCoordinates;
}
}
}
/// <summary>
/// Step 2: Get the ref and variant allele frequencies for the variants of interest, in the tumor bam file.
/// </summary>
protected void ProcessBamFile(string bamPath)
{
Console.WriteLine("{0} Looping over bam records from {1}", DateTime.Now, bamPath);
int overallCount = 0;
int nextVariantIndex = 0;
using (BamReader reader = new BamReader(bamPath))
{
BamAlignment read = new BamAlignment();
int refID = reader.GetReferenceIndex(this.Chromosome);
if (refID < 0)
{
throw new ArgumentException(string.Format("Error: Chromosome name '{0}' does not match bam file at '{1}'", this.Chromosome, bamPath));
}
Console.WriteLine("Jump to refid {0} {1}", refID, this.Chromosome);
reader.Jump(refID, 0);
while (true)
{
bool result = reader.GetNextAlignment(ref read, false);
if (!result) break;
if (!read.HasPosition() || read.RefID > refID) break; // We're past our chromosome of interest.
if (read.RefID < refID) continue; // We're not yet on our chromosome of interest.
overallCount++;
if (overallCount % 1000000 == 0)
{
Console.WriteLine("Record {0} at {1}...", overallCount, read.Position);
}
// Skip over unaligned or other non-count-worthy reads:
if (!read.IsPrimaryAlignment()) continue;
if (!read.IsMapped()) continue;
if (read.IsDuplicate()) continue;
if (read.MapQuality <= MinimumMapQ) continue;
// Scan forward through the variants list, to keep up with our reads:
while (nextVariantIndex < this.Variants.Count && this.Variants[nextVariantIndex].ReferencePosition < read.Position)
{
nextVariantIndex++;
}
if (nextVariantIndex >= this.Variants.Count) break;
// If the read doesn't look like it has a reasonable chance of touching the next variant, continue:
if (read.Position + 1000 < this.Variants[nextVariantIndex].ReferencePosition) continue;
// This read potentially overlaps next variant (and further variants). Count bases!
ProcessReadBases(read, nextVariantIndex);
}
}
Console.WriteLine("Looped over {0} bam records in all", overallCount);
}
/// <summary>
/// Reads in a bam file and marks within the BitArrays which genomic mers are present.
/// </summary>
/// <param name="bamFile">bam file read alignments from.</param>
/// <param name="observedAlignments">Dictioanry of BitArrays, one for each chromosome, to store the alignments in.</param>
static void LoadObservedAlignmentsBAM(string bamFile, bool isPairedEnd, string chromosome, CanvasCoverageMode coverageMode, HitArray observed, Int16[] fragmentLengths)
{
// Sanity check: The .bai file must exist, in order for us to seek to our target chromosome!
string indexPath = bamFile + ".bai";
if (!File.Exists(indexPath))
{
throw new Exception(string.Format("Fatal error: Bam index not found at {0}", indexPath));
}
using (BamReader reader = new BamReader(bamFile))
{
int desiredRefIndex = -1;
desiredRefIndex = reader.GetReferenceIndex(chromosome);
if (desiredRefIndex == -1)
{
throw new ApplicationException(
string.Format("Unable to retrieve the reference sequence index for {0} in {1}.", chromosome,
bamFile));
}
bool result = reader.Jump(desiredRefIndex, 0);
if (!result)
{
// Note: This is not necessarily an error, it just means that there *are* no reads for this chromosome in this
// .bam file. That is not uncommon e.g. for truseq amplicon.
return;
}
int readCount = 0;
int keptReadCount = 0;
string header = reader.GetHeader();
BamAlignment alignment = new BamAlignment();
while (reader.GetNextAlignment(ref alignment, true))
{
readCount++;
// Flag check - Require reads to be aligned, passing filter, non-duplicate:
if (!alignment.IsMapped()) continue;
if (alignment.IsFailedQC()) continue;
if (alignment.IsDuplicate()) continue;
if (alignment.IsReverseStrand()) continue;
if (!alignment.IsMainAlignment()) continue;
// Require the alignment to start with 35 bases of non-indel:
if (alignment.CigarData[0].Type != 'M' || alignment.CigarData[0].Length < 35) continue;
if (isPairedEnd && !alignment.IsProperPair()) continue;
int refID = alignment.RefID;
// quit if the current reference index is different from the desired reference index
if (refID != desiredRefIndex)
break;
if (refID == -1)
continue;
keptReadCount++;
if (coverageMode == CanvasCoverageMode.Binary)
{
observed.Data[alignment.Position] = 1;
}
else
{
observed.Set(alignment.Position);
}
// store fragment size, make sure it's within Int16 range and is positive (simplification for now)
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
fragmentLengths[alignment.Position] = Convert.ToInt16(Math.Max(Math.Min(Int16.MaxValue, alignment.FragmentLength), 0));
}
Console.WriteLine("Kept {0} of {1} total reads", keptReadCount, readCount);
}
}
/// <summary>
/// Bins fragments.
/// </summary>
private void binFragments()
{
// Sanity check: The BAM index file must exist, in order for us to seek to our target chromosome!
if (!Bam.Index.Exists)
{
throw new Exception(string.Format("Fatal error: Bam index not found at {0}", Bam.Index.FullName));
}
long pairedAlignmentCount = 0; // keep track of paired alignments
usableFragmentCount = 0;
using (BamReader reader = new BamReader(Bam.BamFile.FullName))
{
int desiredRefIndex = -1;
desiredRefIndex = reader.GetReferenceIndex(Chromosome);
if (desiredRefIndex == -1)
{
throw new ApplicationException(
string.Format("Unable to retrieve the reference sequence index for {0} in {1}.", Chromosome, Bam.BamFile.FullName));
}
bool result = reader.Jump(desiredRefIndex, 0);
if (!result)
{
// Note: This is not necessarily an error, it just means that there *are* no reads for this chromosome in this
// .bam file. That is not uncommon e.g. for truseq amplicon.
return;
}
Dictionary<string, int> readNameToBinIndex = new Dictionary<string, int>();
HashSet<string> samePositionReadNames = new HashSet<string>();
int binIndexStart = 0;
int prevPosition = -1;
BamAlignment alignment = new BamAlignment();
while (reader.GetNextAlignment(ref alignment, true))
{
int refID = alignment.RefID;
// quit if the current reference index is different from the desired reference index
if (refID != desiredRefIndex)
break;
if (refID == -1)
continue;
if (alignment.Position < prevPosition) // Make sure the BAM is properly sorted
{
throw new ApplicationException(
string.Format("The alignment on {0} are not properly sorted in {1}: {2}", Chromosome, Bam.BamFile.FullName, alignment.Name));
}
prevPosition = alignment.Position;
if (alignment.IsPaired()) { pairedAlignmentCount++; }
BinOneAlignment(alignment, FragmentBinnerConstants.MappingQualityThreshold, readNameToBinIndex,
samePositionReadNames, ref usableFragmentCount, Bins, ref binIndexStart);
}
}
if (pairedAlignmentCount == 0)
{
throw new ApplicationException(string.Format("No paired alignments found for {0} in {1}", Chromosome, Bam.BamFile.FullName));
}
}