/// <summary>
/// Sets up two Dictionaries holding BitArrays, one BitArray for each chromosome in a fasta file. One bit for each nucleotide.
/// </summary>
/// <param name="fastaFile">Fasta file containing uniquemer-marked reference genome.</param>
/// <param name="possibleAlignments">Stores which alignments are possible (perfect and unique).</param>
/// <param name="observedAlignments">Stores observed alignments from a sample.</param>
/// <param name="fragmentLengths">Stores fragment length (Int16).</param>
static void InitializeAlignmentArrays(string fastaFile, string chromosome, CanvasCoverageMode coverageMode, IDictionary <string, BitArray> possibleAlignments, IDictionary <string, HitArray> observedAlignments, IDictionary <string, Int16[]> fragmentLengths)
{
string referenceBases = FastaLoader.LoadFastaSequence(fastaFile, chromosome);
BitArray possible = new BitArray(referenceBases.Length);
possibleAlignments[chromosome] = possible;
observedAlignments[chromosome] = new HitArray(referenceBases.Length);
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
{
fragmentLengths[chromosome] = new Int16[referenceBases.Length];
}
else
{
fragmentLengths[chromosome] = new Int16[0];
}
// Mark which k-mers in the fasta file are unique. These are indicated by upper-case letters.
for (int i = 0; i < referenceBases.Length; i++)
{
if (char.IsUpper(referenceBases[i]))
{
possible[i] = true;
}
}
}
/// <summary>
/// Calculates how many possible alignments corresponds to the desired number of observed alignments per bin.
/// </summary>
/// <param name="countsPerBin">Desired number of observed alignments per bin.</param>
/// <param name="possibleAlignments">BitArrays of possible alignments (unique mers).</param>
/// <param name="observedAlignments">BitArrays storing the observed alignments.</param>
/// <returns>Median alignment rate observed on the autosomes.</returns>
static int CalculateNumberOfPossibleAlignmentsPerBin(int countsPerBin, Dictionary <string, BitArray> possibleAlignments,
Dictionary <string, HitArray> observedAlignments, NexteraManifest manifest = null)
{
List <double> rates = new List <double>();
Dictionary <string, List <NexteraManifest.ManifestRegion> > manifestRegionsByChrom = null;
if (manifest != null)
{
manifestRegionsByChrom = manifest.GetManifestRegionsByChromosome();
}
List <ThreadStart> tasks = new List <ThreadStart>();
foreach (string chr in possibleAlignments.Keys)
{
// We don't want to include the sex chromosomes because they may not be copy number 2
if (!GenomeMetadata.SequenceMetadata.IsAutosome(chr))
{
continue;
}
HitArray observed = observedAlignments[chr];
BitArray possible = possibleAlignments[chr];
List <NexteraManifest.ManifestRegion> regions = null;
if (manifestRegionsByChrom != null)
{
if (!manifestRegionsByChrom.ContainsKey(chr))
{
continue;
}
regions = manifestRegionsByChrom[chr];
}
tasks.Add(new ThreadStart(() =>
{
int numberObserved = observed.CountSetBits(regions);
int numberPossible = CountSetBits(possible, regions);
double rate = numberObserved / (double)numberPossible;
lock (rates)
{
rates.Add(rate);
}
}));
}
Console.WriteLine("Launch CalculateNumberOfPossibleAlignmentsPerBin jobs...");
Console.Out.WriteLine();
//Parallel.ForEach(tasks, t => { t.Invoke(); }); //todo allow controling degree of parallelism
Isas.Shared.Utilities.DoWorkParallelThreads(tasks);
Console.WriteLine("CalculateNumberOfPossibleAlignmentsPerBin jobs complete.");
Console.Out.WriteLine();
double medianRate = CanvasCommon.Utilities.Median(rates);
return((int)(countsPerBin / medianRate));
}
public void GetData(out Dictionary <string, BitArray> possibleAlignments, out Dictionary <string, HitArray> observedAlignments, out Dictionary <string, Int16[]> fragmentLengths)
{
possibleAlignments = Convert(this.PossibleAlignments, this.BitsInLastBytePossibleAlignments);
observedAlignments = new Dictionary <string, HitArray>();
foreach (string key in this.ObservedAlignments.Keys)
{
observedAlignments[key] = new HitArray(this.ObservedAlignments[key]);
}
fragmentLengths = new Dictionary <string, Int16[]>();
foreach (string key in this.FragmentLengths.Keys)
{
fragmentLengths[key] = this.FragmentLengths[key];
}
}
/// <summary>
/// Remove any observed alignment if it wasn't 'possible'.
/// </summary>
/// <param name="observedAlignments">BitArrays of observed alignment positions.</param>
/// <param name="possibleAlignments">BitArrays of possible alignment positions.</param>
static void ScreenObservedTags(IDictionary <string, HitArray> observedAlignments, IDictionary <string, BitArray> possibleAlignments)
{
foreach (string chr in possibleAlignments.Keys)
{
if (!observedAlignments.ContainsKey(chr))
{
continue;
}
HitArray observed = observedAlignments[chr];
BitArray possible = possibleAlignments[chr];
for (int i = 0; i < possible.Length; i++)
{
if (!possible[i])
{
observed.Data[i] = 0;
}
}
}
}
/// <summary>
/// Populate the list of GenomicBin objects for this chromosome.
/// </summary>
static void BinCountsForChromosome(BinTaskArguments arguments)
{
List <GenomicBin> bins = arguments.Bins;
bool usePredefinedBins = bins.Any();
int predefinedBinIndex = 0;
GenericRead fastaEntry = arguments.FastaEntry; //fastaEntryKVP.Value;
BinState currentBin = new BinState();
string chr = arguments.Chromosome;
BitArray possibleAlignments = arguments.PossibleAlignments;
HitArray observedAlignments = arguments.ObservedAlignments;
CanvasCoverageMode coverageMode = arguments.CoverageMode;
int pos = usePredefinedBins ? bins[predefinedBinIndex].Start : 0;
// Skip past leading Ns
while (fastaEntry.Bases[pos].Equals('n'))
{
pos++;
}
List <float> binPositions = new List <float>();
List <int> binObservations = new List <int>();
for (; pos < fastaEntry.Bases.Length; pos++)
{
// Sets the start of the bin
if (currentBin.StartPosition == -1)
{
currentBin.StartPosition = pos;
}
if (!fastaEntry.Bases[pos].Equals("n"))
{
currentBin.NucleotideCount++;
}
//if (Utilities.IsGC(fastaEntry.Bases[pos]))
// currentBin.GCCount++;
switch (fastaEntry.Bases[pos])
{
case 'C':
case 'c':
case 'G':
case 'g':
currentBin.GCCount++;
break;
}
if (possibleAlignments[pos])
{
currentBin.PossibleCount++;
currentBin.ObservedCount += observedAlignments.Data[pos];
binObservations.Add(observedAlignments.Data[pos]);
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
{
binPositions.Add(arguments.ObservedVsExpectedGC[arguments.ReadGCContent[pos]]);
}
}
// We've seen the desired number of possible alignment positions.
if ((!usePredefinedBins && currentBin.PossibleCount == arguments.BinSize) ||
(usePredefinedBins && pos == bins[predefinedBinIndex].Stop - 1))
{
if (coverageMode == CanvasCoverageMode.TruncatedDynamicRange) // Truncated dynamic range
{
currentBin.ObservedCount = 0;
foreach (int Value in binObservations)
{
currentBin.ObservedCount += Math.Min(10, Value);
}
}
if (coverageMode == CanvasCoverageMode.GCContentWeighted) // read GC content weighted
{
currentBin.ObservedCount = 0;
float tmpObservedCount = 0;
for (int i = 0; i < binObservations.Count; i++)
{
tmpObservedCount += Math.Min(10, (float)binObservations[i] / binPositions[i]);
}
currentBin.ObservedCount = (int)Math.Round(tmpObservedCount);
}
int gc = (int)(100 * currentBin.GCCount / currentBin.NucleotideCount);
if (usePredefinedBins)
{
bins[predefinedBinIndex].GC = gc;
bins[predefinedBinIndex].Count = currentBin.ObservedCount;
predefinedBinIndex++;
if (predefinedBinIndex >= bins.Count)
{
break;
} // we have processed all the bins
pos = bins[predefinedBinIndex].Start - 1; // jump to right before the next predefined bin
}
else
{
// Note the pos + 1 to make the first three conform to bed specification
GenomicBin bin = new GenomicBin(chr, currentBin.StartPosition, pos + 1, gc, currentBin.ObservedCount);
bins.Add(bin);
}
// Reset all relevant variables
currentBin.Reset();
binObservations.Clear();
binPositions.Clear();
}
}
}
/// <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);
}
}
public void GetData(out Dictionary<string, BitArray> possibleAlignments, out Dictionary<string, HitArray> observedAlignments, out Dictionary<string, Int16[]> fragmentLengths)
{
possibleAlignments = Convert(this.PossibleAlignments, this.BitsInLastBytePossibleAlignments);
observedAlignments = new Dictionary<string, HitArray>();
foreach (string key in this.ObservedAlignments.Keys)
{
observedAlignments[key] = new HitArray(this.ObservedAlignments[key]);
}
fragmentLengths = new Dictionary<string, Int16[]>();
foreach (string key in this.FragmentLengths.Keys)
{
fragmentLengths[key] = this.FragmentLengths[key];
}
}
/// <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>
/// Sets up two Dictionaries holding BitArrays, one BitArray for each chromosome in a fasta file. One bit for each nucleotide.
/// </summary>
/// <param name="fastaFile">Fasta file containing uniquemer-marked reference genome.</param>
/// <param name="possibleAlignments">Stores which alignments are possible (perfect and unique).</param>
/// <param name="observedAlignments">Stores observed alignments from a sample.</param>
/// <param name="fragmentLengths">Stores fragment length (Int16).</param>
static void InitializeAlignmentArrays(string fastaFile, string chromosome, CanvasCoverageMode coverageMode, IDictionary<string, BitArray> possibleAlignments, IDictionary<string, HitArray> observedAlignments, IDictionary<string, Int16[]> fragmentLengths)
{
string referenceBases = FastaLoader.LoadFastaSequence(fastaFile, chromosome);
BitArray possible = new BitArray(referenceBases.Length);
possibleAlignments[chromosome] = possible;
observedAlignments[chromosome] = new HitArray(referenceBases.Length);
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
fragmentLengths[chromosome] = new Int16[referenceBases.Length];
else
fragmentLengths[chromosome] = new Int16[0];
// Mark which k-mers in the fasta file are unique. These are indicated by upper-case letters.
for (int i = 0; i < referenceBases.Length; i++)
{
if (char.IsUpper(referenceBases[i]))
possible[i] = true;
}
}