/// <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> /// 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); } }