private IEnumerable<AnnotationAndOffset> GetAnnotationsAndOffsets(FastaDatabase database, out long estimatedProteins, CancellationToken? cancellationToken = null)
{
var indexedDb = new IndexedDatabase(database);
indexedDb.Read();
estimatedProteins = indexedDb.EstimateTotalPeptides(SearchMode, MinSequenceLength, MaxSequenceLength, MaxNumNTermCleavages, MaxNumCTermCleavages);
IEnumerable<AnnotationAndOffset> annotationsAndOffsets;
if (SearchMode == InternalCleavageType.MultipleInternalCleavages)
{
//annotationsAndOffsets = indexedDb.AnnotationsAndOffsetsNoEnzyme(MinSequenceLength, MaxSequenceLength);
annotationsAndOffsets = indexedDb.AnnotationsAndOffsetsNoEnzymeParallel(MinSequenceLength, MaxSequenceLength, MaxNumThreads, cancellationToken);
}
else if (SearchMode == InternalCleavageType.NoInternalCleavage)
{
annotationsAndOffsets = indexedDb.IntactSequenceAnnotationsAndOffsets(MinSequenceLength, MaxSequenceLength, MaxNumCTermCleavages);
}
else
{
annotationsAndOffsets = indexedDb
.SequenceAnnotationsAndOffsetsWithNtermOrCtermCleavageNoLargerThan(
MinSequenceLength, MaxSequenceLength, MaxNumNTermCleavages, MaxNumCTermCleavages);
}
return annotationsAndOffsets;
}
[TestCase(15, @"\\proto-2\UnitTest_Files\InformedProteomics_TestFiles\MSPathFinderT\ID_004208_295531A4.fasta", 14862126)] // 15MB
public void TestSequenceEnumerationNCTerm(double size, string dbFile, int expected)
{
var methodName = MethodBase.GetCurrentMethod().Name;
TestUtils.ShowStarting(methodName, dbFile);
var sw = new System.Diagnostics.Stopwatch();
sw.Start();
//const string dbFile = @"\\proto-2\UnitTest_Files\InformedProteomics_TestFiles\MSPathFinderT\ID_002216_235ACCEA.fasta"; // 1.5MB
//const string dbFile = @"\\proto-2\UnitTest_Files\InformedProteomics_TestFiles\MSPathFinderT\ID_005133_8491EFA2.fasta"; // 3MB
//const string dbFile = @"\\proto-2\UnitTest_Files\InformedProteomics_TestFiles\MSPathFinderT\ID_004530_B63BD900.fasta"; // 6MB
//const string dbFile = @"\\proto-2\UnitTest_Files\InformedProteomics_TestFiles\MSPathFinderT\ID_004208_295531A4.fasta"; // 15MB
const int numNTermCleavages = 1;
const int numCTermCleavages = 0;
var db = new FastaDatabase(dbFile);
var indexedDb = new IndexedDatabase(db);
var numSequences = 0L;
var timeDB = sw.Elapsed;
Console.WriteLine("Read DB in " + timeDB.TotalSeconds + " Seconds");
var estimatedAnnOff = indexedDb.EstimateTotalPeptides(1, 21, 300, numNTermCleavages, numCTermCleavages);
var timeEstimate = sw.Elapsed;
Console.WriteLine("Read Estimate in " + (timeEstimate - timeDB).TotalSeconds + " Seconds");
Console.WriteLine("Estimated results: " + estimatedAnnOff);
var annotationsAndOffsets = indexedDb.SequenceAnnotationsAndOffsetsWithNtermOrCtermCleavageNoLargerThan(21, 300, numNTermCleavages, numCTermCleavages);
var timeGetAnn = sw.Elapsed;
Console.WriteLine("Read Annotations in " + (timeGetAnn - timeEstimate).TotalSeconds + " Seconds");
/*/Parallel.ForEach(
annotationsAndOffsets,
// new ParallelOptions { MaxDegreeOfParallelism = 2},
annotationAndOffset =>
{
Interlocked.Increment(ref numSequences);
//++numSequences;
}
);/**/
//using (var ofstream = new FileStream(Path.Combine(@"F:\InformedProteomicsTestFiles", Path.GetFileNameWithoutExtension(dbFile) + "_par.txt"), FileMode.Create))
//using (var fout = new StreamWriter(ofstream))
//{
// foreach (var annOff in annotationsAndOffsets)
// {
// numSequences++;
// fout.WriteLine(annOff.Annotation);
// }
//}
//foreach (var sao in annotationsAndOffsets)
//{
// numSequences++;
//}
numSequences = annotationsAndOffsets.Count();
var timeParForEach = sw.Elapsed;
Console.WriteLine("Parallel ForEach in " + (timeParForEach - timeGetAnn).TotalSeconds + " Seconds");
Console.WriteLine("NumPeptides: {0}", numSequences);
sw.Stop();
Console.WriteLine(@"{0:f4} sec", sw.Elapsed.TotalSeconds);
//Assert.AreEqual(188961836, numSequences);
Assert.AreEqual(expected, numSequences);
}
public void TestCountingPeptides()
{
var methodName = MethodBase.GetCurrentMethod().Name;
TestUtils.ShowStarting(methodName);
var sw = new System.Diagnostics.Stopwatch();
sw.Start();
const string dbFile = @"\\proto-2\UnitTest_Files\InformedProteomics_TestFiles\MSPathFinderT\Short.fasta";
if (!File.Exists(dbFile))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, dbFile);
}
// const string dbFile = @"C:\cygwin\home\kims336\Data\QCShew\ID_003456_9B916A8B.fasta";
// const string dbFile = @"H:\Research\DDAPlus\database\Yeast_SGD_withContam.fasta";
// const string dbFile = @"H:\Research\CPTAC_Phospho\database\ID_004208_295531A4.fasta";
var db = new FastaDatabase(dbFile);
var indexedDb = new IndexedDatabase(db);
//var numPeptides = indexedDb.IntactSequenceAnnotationsAndOffsets(21, 300, 0).LongCount()*31;
var peptides = indexedDb
.SequenceAnnotationsAndOffsetsWithNtermOrCtermCleavageNoLargerThan(
100, 300, 1, 0);
var numPeptides = 0;
foreach (var peptide in peptides)
{
Console.WriteLine("{0}\t{1}",peptide.Annotation, peptide.Offset);
numPeptides++;
}
//var numPeptides = indexedDb.AnnotationsAndOffsetsNoEnzyme(7, 150).LongCount();
//var numPeptides =
// indexedDb.AnnotationsAndOffsets(7, 40, 2, 2, Enzyme.Trypsin).LongCount();
//var numPeptides = indexedDb.AnnotationsAndOffsets(6, 40, 2, 2, Enzyme.Trypsin).LongCount();
//var numPeptides = indexedDb.IntactSequenceAnnotationsAndOffsets(30, 250, 0).LongCount();
// .Select(annotationAndSequence => annotationAndSequence.Annotation.Length - 4)
// .Aggregate(0L, (current, length) => current + Math.Min(length - 29, 30));
Console.WriteLine("NumPeptides: {0}", numPeptides);
sw.Stop();
Console.WriteLine(@"{0:f4} sec", sw.Elapsed.TotalSeconds);
}