public void TestForManyMods(string dbFile, int expectedMaxLen300, int expectedMaxLen400, int expectedMaxLen500)
{
var methodName = MethodBase.GetCurrentMethod().Name;
Utils.ShowStarting(methodName);
var fastaFile = Utils.GetTestFile(methodName, dbFile.Replace("TEST_FOLDER", Utils.DEFAULT_TEST_FILE_FOLDER));
var indexedDb = new IndexedDatabase(new FastaDatabase(fastaFile.FullName));
indexedDb.Read();
var totalPeptidesMaxLen300 = indexedDb.EstimateTotalPeptides(1, 21, 300, 1, 0);
Console.WriteLine("{0,12:N0} estimated peptides for max length 300", totalPeptidesMaxLen300);
var totalPeptidesMaxLen400 = indexedDb.EstimateTotalPeptides(1, 21, 400, 1, 0);
Console.WriteLine("{0,12:N0} estimated peptides for max length 400", totalPeptidesMaxLen400);
var totalPeptidesMaxLen500 = indexedDb.EstimateTotalPeptides(1, 21, 500, 1, 0);
Console.WriteLine("{0,12:N0} estimated peptides for max length 500", totalPeptidesMaxLen500);
Assert.AreEqual(expectedMaxLen300, totalPeptidesMaxLen300);
Assert.AreEqual(expectedMaxLen400, totalPeptidesMaxLen400);
Assert.AreEqual(expectedMaxLen500, totalPeptidesMaxLen500);
}
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(1.5, @"TEST_FOLDER\MSPathFinderT\ID_002216_235ACCEA.fasta", 188961836)] // 1.5MB
//[TestCase(3, @"TEST_FOLDER\MSPathFinderT\ID_005133_8491EFA2.fasta", 323719193)] // 3MB
//[TestCase(6, @"TEST_FOLDER\MSPathFinderT\ID_004530_B63BD900.fasta", 595227563)] // 6MB
//[TestCase(15, @"TEST_FOLDER\MSPathFinderT\ID_004208_295531A4.fasta", 1882434687)] // 15MB
public void TestSequenceEnumerationSerial(double size, string dbFile, int expected)
{
var methodName = MethodBase.GetCurrentMethod().Name;
Utils.ShowStarting(methodName, dbFile);
var fastaFile = Utils.GetTestFile(methodName, dbFile.Replace("TEST_FOLDER", Utils.DEFAULT_TEST_FILE_FOLDER));
var sw = new System.Diagnostics.Stopwatch();
sw.Start();
var db = new FastaDatabase(fastaFile.FullName);
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(fastaFile) + "_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);
}
[TestCase(@"TEST_FOLDER\MSPathFinderT\ID_004530_B63BD900.fasta", 8898)] // 6MB
//[TestCase(@"TEST_FOLDER\MSPathFinderT\ID_004208_295531A4.fasta", 6334)] // 15MB
public void TestSequenceEnumerationIntact(string dbFile, int expected)
{
var methodName = MethodBase.GetCurrentMethod().Name;
Utils.ShowStarting(methodName, dbFile);
var fastaFile = Utils.GetTestFile(methodName, dbFile.Replace("TEST_FOLDER", Utils.DEFAULT_TEST_FILE_FOLDER));
var sw = new Stopwatch();
sw.Start();
const int numCTermCleavages = 0;
var db = new FastaDatabase(fastaFile.FullName);
var indexedDb = new IndexedDatabase(db);
var timeDB = sw.Elapsed;
Console.WriteLine("Read DB in " + timeDB.TotalSeconds + " Seconds");
var estimatedAnnOff = indexedDb.EstimateTotalPeptides(2, 21, 300, 1, numCTermCleavages);
var timeEstimate = sw.Elapsed;
Console.WriteLine("Read Estimate in " + (timeEstimate - timeDB).TotalSeconds + " Seconds");
Console.WriteLine("Estimated results: " + estimatedAnnOff);
var annotationsAndOffsets = indexedDb.IntactSequenceAnnotationsAndOffsets(21, 300, 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(fastaFile) + "_par.txt"), FileMode.Create))
//using (var fout = new StreamWriter(ofstream))
//{
// foreach (var annOff in annotationsAndOffsets)
// {
// numSequences++;
// fout.WriteLine(annOff.Annotation);
// }
//}
long 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 TestForManyMods()
{
var methodName = MethodBase.GetCurrentMethod().Name;
TestUtils.ShowStarting(methodName);
const string dbFilePath = @"\\protoapps\UserData\Jungkap\Lewy\db\ID_005140_7A170668.fasta";
var indexedDb = new IndexedDatabase(new FastaDatabase(dbFilePath));
indexedDb.Read();
var nProt = indexedDb.EstimateTotalPeptides(1, 21, 300, 1, 0);
Console.WriteLine(nProt);
nProt = indexedDb.EstimateTotalPeptides(1, 21, 400, 1, 0);
Console.WriteLine(nProt);
nProt = indexedDb.EstimateTotalPeptides(1, 21, 500, 1, 0);
Console.WriteLine(nProt);
Console.WriteLine(@"Test not implemented: " + methodName);
}
[TestCase(3, @"TEST_FOLDER\MSPathFinderT\ID_005133_8491EFA2.fasta", 323719193)] // 3MB
//[TestCase(6, @"TEST_FOLDER\MSPathFinderT\ID_004530_B63BD900.fasta", 595227563)] // 6MB
//[TestCase(15, @"TEST_FOLDER\MSPathFinderT\ID_004208_295531A4.fasta", 1882434687)] // 15MB
public void TestSequenceEnumeration(double size, string dbFile, int expected)
{
var methodName = MethodBase.GetCurrentMethod().Name;
Utils.ShowStarting(methodName, dbFile);
var fastaFile = Utils.GetTestFile(methodName, dbFile.Replace("TEST_FOLDER", Utils.DEFAULT_TEST_FILE_FOLDER));
var sw = new System.Diagnostics.Stopwatch();
sw.Start();
var db = new FastaDatabase(fastaFile.FullName);
var indexedDb = new IndexedDatabase(db);
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");
//int coreCount = 0;
//foreach (var item in new System.Management.ManagementObjectSearcher("Select NumberOfCores from Win32_Processor").Get())
//{
// coreCount += int.Parse(item["NumberOfCores"].ToString());
//}
//Console.WriteLine("Number Of Cores: {0}", coreCount);
//Console.WriteLine("Processors: " + System.Environment.ProcessorCount);
Console.WriteLine("Estimated results: " + estimatedAnnOff);
var annotationsAndOffsets = indexedDb.AnnotationsAndOffsetsNoEnzymeParallel(30, 250);
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;
* }
* );/**/
//annotationsAndOffsets.Select(annotationsAndOffset => annotationsAndOffset.)
// Below, original: 110, 109(total) seconds
// Parallelizing AnnotationsAndOffsetsNoEnzyme: 86 seconds
// Parallelizing AnnotationsAndOffsetsNoEnzyme, yield returns: 79.6, 94, 60, 60 seconds
//
// 3MB
// serial:
// Parallel2: 107,
//
// 6MB
// serial:
// Parallel2:
//
// 15MB
// serial:
// Parallel2:
//using (var ofstream = new FileStream(Path.Combine(@"F:\InformedProteomicsTestFiles", Path.GetFileNameWithoutExtension(fastaFile) + "_par.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);
}
[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);
}
[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);
}
