public void TestEnumeratingPeptides()
{
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);
}
var db = new FastaDatabase(dbFile);
var indexedDb = new IndexedDatabase(db);
foreach (var annotationAndOffset in indexedDb.AnnotationsAndOffsetsNoEnzyme(10, 13))
{
Console.WriteLine(annotationAndOffset.Annotation);
}
sw.Stop();
Console.WriteLine(@"{0:f4} sec", sw.Elapsed.TotalSeconds);
}
[TestCase(3, @"\\proto-2\UnitTest_Files\InformedProteomics_TestFiles\MSPathFinderT\ID_005133_8491EFA2.fasta", 323719193)] // 3MB
//[TestCase(6, @"\\proto-2\UnitTest_Files\InformedProteomics_TestFiles\MSPathFinderT\ID_004530_B63BD900.fasta", 595227563)] // 6MB
//[TestCase(15, @"\\proto-2\UnitTest_Files\InformedProteomics_TestFiles\MSPathFinderT\ID_004208_295531A4.fasta", 1882434687)] // 15MB
public void TestSequenceEnumerationSerial(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
var db = new FastaDatabase(dbFile);
var indexedDb = new IndexedDatabase(db);
indexedDb.Read();
var numSequences = 0L;
var timeDB = sw.Elapsed;
Console.WriteLine("Read DB in " + timeDB.TotalSeconds + " Seconds");
var estimatedAnnOff = indexedDb.EstimateTotalPeptides(0, 30, 250);
var timeEstimate = sw.Elapsed;
Console.WriteLine("Read Estimate in " + (timeEstimate - timeDB).TotalSeconds + " Seconds");
Console.WriteLine("Estimated results: " + estimatedAnnOff);
var annotationsAndOffsets = indexedDb.AnnotationsAndOffsetsNoEnzyme(30, 250);
var timeGetAnn = sw.Elapsed;
Console.WriteLine("Read Annotations in " + (timeGetAnn - timeEstimate).TotalSeconds + " Seconds");
//foreach (var annotationsAndOffset in annotationsAndOffsets)
//{
// //Interlocked.Increment(ref numSequences);
// ++numSequences;
//}
//using (
// var ofstream =
// new FileStream(
// Path.Combine(@"F:\InformedProteomicsTestFiles",
// Path.GetFileNameWithoutExtension(dbFile) + "_old.txt"), FileMode.Create))
//using (var fout = new StreamWriter(ofstream))
//{
// foreach (var annOff in annotationsAndOffsets)
// {
// numSequences++;
// fout.WriteLine(annOff.Annotation);
// }
//}
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);
}
private IEnumerable<AnnotationAndOffset> GetAnnotationsAndOffsets(FastaDatabase database)
{
var indexedDbTarget = new IndexedDatabase(database);
IEnumerable<AnnotationAndOffset> annotationsAndOffsets;
if (NumTolerableTermini == 0)
{
annotationsAndOffsets = indexedDbTarget.AnnotationsAndOffsetsNoEnzyme(MinSequenceLength, MaxSequenceLength);
}
else
{
annotationsAndOffsets = indexedDbTarget.AnnotationsAndOffsets(MinSequenceLength, MaxSequenceLength,
NumTolerableTermini, 2, Enzyme);
}
return annotationsAndOffsets;
}
public void TestNominalMassErrors()
{
const int minLength = 300;
const int maxLength = 400;
var sw = new System.Diagnostics.Stopwatch();
// const string dbFile = @"\\proto-2\UnitTest_Files\InformedProteomics_TestFiles\H_sapiens_Uniprot_SPROT_2013-05-01_withContam.fasta";
const string dbFile = @"C:\cygwin\home\kims336\Data\TopDownJia\database\ID_003962_71E1A1D4.fasta";
//const string dbFile = @"C:\cygwin\home\kims336\Data\TopDownJia\database\TargetProteins.fasta";
var db = new FastaDatabase(dbFile);
db.Read();
var indexedDb = new IndexedDatabase(db);
var numSequences = 0L;
sw.Start();
var hist = new long[11];
var aaSet = new AminoAcidSet();
foreach (var peptideAnnotationAndOffset in indexedDb.AnnotationsAndOffsetsNoEnzyme(minLength, maxLength))
{
++numSequences;
var annotation = peptideAnnotationAndOffset.Annotation;
var sequenceStr = annotation.Substring(2, annotation.Length - 4);
var sequenceComp = aaSet.GetComposition(sequenceStr);
var mass = sequenceComp.Mass;
var nominalMass = sequenceComp.NominalMass;
var error = (int) Math.Round(mass*Constants.RescalingConstant) - nominalMass;
var errorBin = error + hist.Length/2;
if (errorBin < 0) errorBin = 0;
if (errorBin >= hist.Length) errorBin = hist.Length - 1;
hist[errorBin]++;
}
Console.WriteLine("NumSequences: {0}", numSequences);
for (var i = 0; i < hist.Length; i++)
{
Console.WriteLine("{0}\t{1}\t{2}", i - hist.Length/2, hist[i], hist[i]/(double)numSequences);
}
sw.Stop();
Console.WriteLine(@"Elapsed Time: {0:f4} sec", sw.Elapsed.TotalSeconds);
}