/// <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>
/// 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 (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>
/// Bin alignments.
/// </summary>
/// <param name="referenceFile">Reference fasta file.</param>
/// <param name="binSize">Desired number of alignments per bin.</param>
/// <param name="possibleAlignments">BitArrays of possible alignments.</param>
/// <param name="observedAlignments">BitArrays of observed alignments.</param>
/// <param name="predefinedBins">Pre-defined bins. null if not available.</param>
/// <returns>A list of bins.</returns>
static List <GenomicBin> BinCounts(string referenceFile, int binSize, CanvasCoverageMode coverageMode, NexteraManifest manifest,
Dictionary <string, BitArray> possibleAlignments,
Dictionary <string, HitArray> observedAlignments,
Dictionary <string, Int16[]> fragmentLengths,
Dictionary <string, List <GenomicBin> > predefinedBins,
string outFile)
{
bool debugGCCorrection = false; // write value of GC bins and correction factor
Dictionary <string, GenericRead> fastaEntries = new Dictionary <string, GenericRead>();
List <string> chromosomes = new List <string>();
Int16 meanFragmentSize = 0;
Int16 meanFragmentCutoff = 3;
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
{
meanFragmentSize = MeanFragmentSize(fragmentLengths);
}
using (FastaReader reader = new FastaReader(referenceFile))
{
GenericRead fastaEntry = new GenericRead();
// Loop through each chromosome in the reference.
while (reader.GetNextEntry(ref fastaEntry))
{
chromosomes.Add(fastaEntry.Name);
fastaEntries[fastaEntry.Name] = fastaEntry;
fastaEntry = new GenericRead();
}
}
// calculate GC content of the forward read at every position along the genome
Dictionary <string, byte[]> readGCContent = new Dictionary <string, byte[]>();
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
{
byte gcCap = (byte)numberOfGCbins;
List <ThreadStart> normalizationTasks = new List <ThreadStart>();
foreach (KeyValuePair <string, Int16[]> fragmentLengthsKVP in fragmentLengths)
{
string chr = fragmentLengthsKVP.Key;
GenericRead fastaEntry = fastaEntries[chr];
normalizationTasks.Add(new ThreadStart(() =>
{
// contains GC content of the forward read at every position for current chr
byte[] gcContent = new byte[fastaEntry.Bases.Length];
uint gcCounter = 0;
// Iteratively calculate GC content of "reads" using fasta genome reference
for (int pos = 0; pos < fastaEntry.Bases.Length - meanFragmentSize * meanFragmentCutoff - 1; pos++)
{
Int16 currentFragment = 0;
if (fragmentLengthsKVP.Value[pos] == 0)
{
currentFragment = meanFragmentSize;
}
else
{
currentFragment = Convert.ToInt16(Math.Min(fragmentLengthsKVP.Value[pos], meanFragmentSize * meanFragmentCutoff));
}
for (int i = pos; i < pos + currentFragment; i++)
{
switch (fastaEntry.Bases[i])
{
case 'C':
case 'c':
case 'G':
case 'g':
gcCounter++;
break;
default:
break;
}
}
gcContent[pos] = (byte)Math.Min(100 * gcCounter / currentFragment, gcCap);
gcCounter = 0;
}
lock (readGCContent)
{
readGCContent[chr] = gcContent;
}
}));
}
Console.WriteLine("{0} Launching normalization tasks.", DateTime.Now);
Console.Out.Flush();
Isas.Shared.Utilities.DoWorkParallelThreads(normalizationTasks);
Console.WriteLine("{0} Normalization tasks complete.", DateTime.Now);
Console.Out.Flush();
}
// populate observed and expected read GC bin vectors
float[] observedVsExpectedGC = new float[0];
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
{
observedVsExpectedGC = ComputeObservedVsExpectedGC(observedAlignments, readGCContent, manifest, debugGCCorrection, outFile);
}
Dictionary <string, List <GenomicBin> > perChromosomeBins = new Dictionary <string, List <GenomicBin> >();
List <ThreadStart> binningTasks = new List <ThreadStart>();
foreach (KeyValuePair <string, GenericRead> fastaEntryKVP in fastaEntries)
{
string chr = fastaEntryKVP.Key;
if (!possibleAlignments.ContainsKey(chr))
{
continue;
}
if (predefinedBins != null && !predefinedBins.ContainsKey(chr))
{
continue;
}
BinTaskArguments args = new BinTaskArguments();
args.FastaEntry = fastaEntryKVP.Value;
args.Chromosome = chr;
args.PossibleAlignments = possibleAlignments[chr];
args.ObservedAlignments = observedAlignments[chr];
args.CoverageMode = coverageMode;
perChromosomeBins[chr] = predefinedBins == null ? new List <GenomicBin>() : predefinedBins[chr];
args.Bins = perChromosomeBins[chr];
args.BinSize = binSize;
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
{
args.ReadGCContent = readGCContent[chr];
}
else
{
args.ReadGCContent = null;
}
args.ObservedVsExpectedGC = observedVsExpectedGC;
binningTasks.Add(new ThreadStart(() => { BinCountsForChromosome(args); }));
}
Console.WriteLine("{0} Launch BinCountsForChromosome jobs...", DateTime.Now);
Console.Out.WriteLine();
//Parallel.ForEach(binningTasks, t => { t.Invoke(); });
Isas.Shared.Utilities.DoWorkParallelThreads(binningTasks);
Console.WriteLine("{0} Completed BinCountsForChromosome jobs.", DateTime.Now);
Console.Out.WriteLine();
List <GenomicBin> finalBins = new List <GenomicBin>();
foreach (string chr in chromosomes)
{
if (!perChromosomeBins.ContainsKey(chr))
{
continue;
}
finalBins.AddRange(perChromosomeBins[chr]);
}
return(finalBins);
}
/// <summary>
/// Bin alignments.
/// </summary>
/// <param name="referenceFile">Reference fasta file.</param>
/// <param name="binSize">Desired number of alignments per bin.</param>
/// <param name="possibleAlignments">BitArrays of possible alignments.</param>
/// <param name="observedAlignments">BitArrays of observed alignments.</param>
/// <param name="predefinedBins">Pre-defined bins. null if not available.</param>
/// <returns>A list of bins.</returns>
static List<GenomicBin> BinCounts(string referenceFile, int binSize, CanvasCoverageMode coverageMode, NexteraManifest manifest,
Dictionary<string, BitArray> possibleAlignments,
Dictionary<string, HitArray> observedAlignments,
Dictionary<string, Int16[]> fragmentLengths,
Dictionary<string, List<GenomicBin>> predefinedBins,
string outFile)
{
bool debugGCCorrection = false; // write value of GC bins and correction factor
Dictionary<string, GenericRead> fastaEntries = new Dictionary<string, GenericRead>();
List<string> chromosomes = new List<string>();
Int16 meanFragmentSize = 0;
Int16 meanFragmentCutoff = 3;
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
meanFragmentSize = MeanFragmentSize(fragmentLengths);
using (FastaReader reader = new FastaReader(referenceFile))
{
GenericRead fastaEntry = new GenericRead();
// Loop through each chromosome in the reference.
while (reader.GetNextEntry(ref fastaEntry))
{
chromosomes.Add(fastaEntry.Name);
fastaEntries[fastaEntry.Name] = fastaEntry;
fastaEntry = new GenericRead();
}
}
// calculate GC content of the forward read at every position along the genome
Dictionary<string, byte[]> readGCContent = new Dictionary<string, byte[]>();
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
{
byte gcCap = (byte)numberOfGCbins;
List<ThreadStart> normalizationTasks = new List<ThreadStart>();
foreach (KeyValuePair<string, Int16[]> fragmentLengthsKVP in fragmentLengths)
{
string chr = fragmentLengthsKVP.Key;
GenericRead fastaEntry = fastaEntries[chr];
normalizationTasks.Add(new ThreadStart(() =>
{
// contains GC content of the forward read at every position for current chr
byte[] gcContent = new byte[fastaEntry.Bases.Length];
int gcCounter = 0;
// Iteratively calculate GC content of "reads" using fasta genome reference
for (int pos = 0; pos < fastaEntry.Bases.Length - meanFragmentSize * meanFragmentCutoff - 1; pos++)
{
Int16 currentFragment = 0;
if (fragmentLengthsKVP.Value[pos] == 0)
currentFragment = meanFragmentSize;
else
currentFragment = Convert.ToInt16(Math.Min(fragmentLengthsKVP.Value[pos], meanFragmentSize * meanFragmentCutoff));
for (int i = pos; i < pos + currentFragment; i++)
{
switch (fastaEntry.Bases[i])
{
case 'C':
case 'c':
case 'G':
case 'g':
gcCounter++;
break;
default:
break;
}
}
if (gcCounter < 0)
gcCounter = 0;
gcContent[pos] = (byte)Math.Min(100 * gcCounter / currentFragment, gcCap);
gcCounter = 0;
}
lock (readGCContent)
{
readGCContent[chr] = gcContent;
}
}));
}
Console.WriteLine("{0} Launching normalization tasks.", DateTime.Now);
Console.Out.Flush();
//Parallel.ForEach(normalizationTasks, t => { t.Invoke(); });
Illumina.SecondaryAnalysis.Utilities.DoWorkParallelThreads(normalizationTasks);
Console.WriteLine("{0} Normalization tasks complete.", DateTime.Now);
Console.Out.Flush();
}
// populate observed and expected read GC bin vectors
float[] observedVsExpectedGC = new float[0];
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
observedVsExpectedGC = ComputeObservedVsExpectedGC(observedAlignments, readGCContent, manifest, debugGCCorrection, outFile);
Dictionary<string, List<GenomicBin>> perChromosomeBins = new Dictionary<string, List<GenomicBin>>();
List<ThreadStart> binningTasks = new List<ThreadStart>();
foreach (KeyValuePair<string, GenericRead> fastaEntryKVP in fastaEntries)
{
string chr = fastaEntryKVP.Key;
if (!possibleAlignments.ContainsKey(chr)) continue;
if (predefinedBins != null && !predefinedBins.ContainsKey(chr)) continue;
BinTaskArguments args = new BinTaskArguments();
args.FastaEntry = fastaEntryKVP.Value;
args.Chromosome = chr;
args.PossibleAlignments = possibleAlignments[chr];
args.ObservedAlignments = observedAlignments[chr];
args.CoverageMode = coverageMode;
perChromosomeBins[chr] = predefinedBins == null ? new List<GenomicBin>() : predefinedBins[chr];
args.Bins = perChromosomeBins[chr];
args.BinSize = binSize;
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
args.ReadGCContent = readGCContent[chr];
else
args.ReadGCContent = null;
args.ObservedVsExpectedGC = observedVsExpectedGC;
binningTasks.Add(new ThreadStart(() => { BinCountsForChromosome(args); }));
}
Console.WriteLine("{0} Launch BinCountsForChromosome jobs...", DateTime.Now);
Console.Out.WriteLine();
//Parallel.ForEach(binningTasks, t => { t.Invoke(); });
Illumina.SecondaryAnalysis.Utilities.DoWorkParallelThreads(binningTasks);
Console.WriteLine("{0} Completed BinCountsForChromosome jobs.", DateTime.Now);
Console.Out.WriteLine();
List<GenomicBin> finalBins = new List<GenomicBin>();
foreach (string chr in chromosomes)
{
if (!perChromosomeBins.ContainsKey(chr)) continue;
finalBins.AddRange(perChromosomeBins[chr]);
}
return finalBins;
}