// Refactored to remove this block from GetAlignmentMap()
private SAMAlignedSequence BamIndexing(SAMAlignedSequence alignedSeq, BAMReferenceIndexes refIndices, BAMIndex index,
ulong lastcOffset, ushort lastuOffset, ref Chunk lastChunk)
{
int lastBin = int.MaxValue;
Bin bin;
Chunk chunk;
int lastRefSeqIndex = 0;
int curRefSeqIndex;
#region BAM indexing
if (createBamIndex)
{
curRefSeqIndex = refSeqNames.IndexOf(alignedSeq.RName);
if (lastRefSeqIndex != curRefSeqIndex)
{
refIndices = index.RefIndexes[curRefSeqIndex];
lastBin = int.MaxValue;
lastRefSeqIndex = curRefSeqIndex;
}
if (lastBin != alignedSeq.Bin)
{
bin = refIndices.Bins.FirstOrDefault(B => B.BinNumber == alignedSeq.Bin);
if (bin == null)
{
bin = new Bin();
bin.BinNumber = (uint)alignedSeq.Bin;
refIndices.Bins.Add(bin);
}
if (lastChunk != null)
{
lastChunk.ChunkEnd.CompressedBlockOffset = lastcOffset;
lastChunk.ChunkEnd.UncompressedBlockOffset = lastuOffset;
}
chunk = new Chunk();
chunk.ChunkStart = new FileOffset();
chunk.ChunkEnd = new FileOffset();
chunk.ChunkStart.CompressedBlockOffset = lastcOffset;
chunk.ChunkStart.UncompressedBlockOffset = lastuOffset;
bin.Chunks.Add(chunk);
lastChunk = chunk;
lastBin = alignedSeq.Bin;
}
// store linear index other than 16k bins, that is bin number less than 4681.
if (alignedSeq.Bin < 4681)
{
int pos = alignedSeq.Pos > 0 ? alignedSeq.Pos - 1 : 0;
int end = alignedSeq.RefEndPos > 0 ? alignedSeq.RefEndPos - 1 : 0;
pos = pos >> 14;
end = end >> 14;
if (refIndices.LinearOffsets.Count == 0)
{
refIndices.LinearOffsets.Add(new FileOffset());
}
if (refIndices.LinearOffsets.Count <= end)
{
for (int i = refIndices.LinearOffsets.Count; i <= end; i++)
{
refIndices.LinearOffsets.Add(new FileOffset());
}
}
for (int i = pos + 1; i <= end; i++)
{
FileOffset offset = refIndices.LinearOffsets[i];
if (offset.CompressedBlockOffset == 0 && offset.UncompressedBlockOffset == 0)
{
offset.CompressedBlockOffset = lastcOffset;
offset.UncompressedBlockOffset = lastuOffset;
}
}
}
}
#endregion
return alignedSeq;
}
/// <summary>
/// Returns BAMIndex instance by parsing BAM index source.
/// </summary>
public BAMIndex Read()
{
if (Source == null)
{
throw new InvalidOperationException(Properties.Resource.BAM_CantUseBAMIndexStreamDisposed);
}
BAMIndex bamIndex = new BAMIndex();
byte[] arrays = new byte[20];
Read(arrays, 0, 4);
if (arrays[0] != 66 || arrays[1] != 65 || arrays[2] != 73 || arrays[3] != 1)
{
throw new FormatException(Properties.Resource.BAM_InvalidIndexFile);
}
Read(arrays, 0, 4);
int n_ref = Helper.GetInt32(arrays, 0);
for (Int32 refindex = 0; refindex < n_ref; refindex++)
{
BAMReferenceIndexes bamindices = new BAMReferenceIndexes();
bamIndex.RefIndexes.Add(bamindices);
Read(arrays, 0, 4);
int n_bin = Helper.GetInt32(arrays, 0);
for (Int32 binIndex = 0; binIndex < n_bin; binIndex++)
{
Bin bin = new Bin();
Read(arrays, 0, 4);
bin.BinNumber = Helper.GetUInt32(arrays, 0);
Read(arrays, 0, 4);
int n_chunk = Helper.GetInt32(arrays, 0);
if (bin.BinNumber == MaxBins)//some groups use this to place meta-data, such as the picard toolkit and now SAMTools
{
//Meta data was later added in to the SAMTools specification
for (Int32 chunkIndex = 0; chunkIndex < n_chunk; chunkIndex++)
{
bamindices.HasMetaData = true;
Read(arrays, 0, 8);
bamindices.MappedReadsCount = Helper.GetUInt64(arrays, 0);
Read(arrays, 0, 8);
bamindices.UnMappedReadsCount = Helper.GetUInt64(arrays, 0);
}
}
else if (bin.BinNumber > MaxBins)
{
throw new Exception("BAM Index is incorrectly formatted. Bin number specified is higher than the maximum allowed.");
}
else
{
bamindices.Bins.Add(bin);
for (Int32 chunkIndex = 0; chunkIndex < n_chunk; chunkIndex++)
{
Chunk chunk = new Chunk();
bin.Chunks.Add(chunk);
Read(arrays, 0, 8);
chunk.ChunkStart = GetBAMOffset(arrays, 0);
Read(arrays, 0, 8);
chunk.ChunkEnd = GetBAMOffset(arrays, 0);
}
}
}
//Get number of linear bins
Read(arrays, 0, 4);
int n_intv = Helper.GetInt32(arrays, 0);
for (Int32 offsetIndex = 0; offsetIndex < n_intv; offsetIndex++)
{
FileOffset value;
Read(arrays, 0, 8);
value = GetBAMOffset(arrays, 0);
bamindices.LinearIndex.Add(value);
}
}
return bamIndex;
}
/// <summary>
/// Returns BAMIndex instance by parsing BAM index source.
/// </summary>
public BAMIndex Read()
{
if (sourceStream == null)
{
throw new InvalidOperationException(Properties.Resource.BAM_CantUseBAMIndexStreamDisposed);
}
BAMIndex bamIndex = new BAMIndex();
byte[] arrays = new byte[20];
Read(arrays, 0, 4);
if (arrays[0] != 66 || arrays[1] != 65 || arrays[2] != 73 || arrays[3] != 1)
{
throw new FormatException(Properties.Resource.BAM_InvalidIndexFile);
}
Read(arrays, 0, 4);
int n_ref = Helper.GetInt32(arrays, 0);
for (Int32 refindex = 0; refindex < n_ref; refindex++)
{
BAMReferenceIndexes bamindices = new BAMReferenceIndexes();
bamIndex.RefIndexes.Add(bamindices);
Read(arrays, 0, 4);
int n_bin = Helper.GetInt32(arrays, 0);
for (Int32 binIndex = 0; binIndex < n_bin; binIndex++)
{
Bin bin = new Bin();
bamindices.Bins.Add(bin);
Read(arrays, 0, 4);
bin.BinNumber = Helper.GetUInt32(arrays, 0);
Read(arrays, 0, 4);
int n_chunk = Helper.GetInt32(arrays, 0);
for (Int32 chunkIndex = 0; chunkIndex < n_chunk; chunkIndex++)
{
Chunk chunk = new Chunk();
bin.Chunks.Add(chunk);
Read(arrays, 0, 8);
chunk.ChunkStart = GetBAMOffset(arrays, 0);
Read(arrays, 0, 8);
chunk.ChunkEnd = GetBAMOffset(arrays, 0);
}
}
Read(arrays, 0, 4);
int n_intv = Helper.GetInt32(arrays, 0);
for (Int32 offsetIndex = 0; offsetIndex < n_intv; offsetIndex++)
{
FileOffset value;
Read(arrays, 0, 8);
value = GetBAMOffset(arrays, 0);
bamindices.LinearOffsets.Add(value);
}
}
return(bamIndex);
}
