public void TestSearching()
{
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 searchableDb = new SearchableDatabase(db);
//const string pattern = "NSGSHFCGGSLINSQWVVSAAH";
const string pattern = "FPTDDDDK";
var position = searchableDb.Search(pattern);
Assert.True(position >= 0);
Console.WriteLine("Position: {0}", position);
Console.WriteLine("Matched indices: {0}", string.Join(",", searchableDb.FindAllMatchedSequenceIndices(pattern)));
Console.WriteLine("Protein indices: {0}", string.Join(",", searchableDb.FindAllMatchedSequenceIndices(pattern).Select(i => db.GetOneBasedPositionInProtein(i))));
sw.Stop();
Console.WriteLine(@"{0:f4} sec", sw.Elapsed.TotalSeconds);
}
public void TestTagMatching()
{
var methodName = MethodBase.GetCurrentMethod().Name;
TestUtils.ShowStarting(methodName);
// Parse sequence tags
const string dataSet = @"H:\Research\QCShew_TopDown\Production\QC_Shew_Intact_26Sep14_Bane_C2Column3";
const int minTagLength = 8;
var tagFileName = MassSpecDataReaderFactory.ChangeExtension(dataSet, ".seqtag");
if (!File.Exists(tagFileName))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, tagFileName);
}
var tagParser = new SequenceTagParser(tagFileName, minTagLength);
// Parse raw file
var rawFileName = MassSpecDataReaderFactory.ChangeExtension(dataSet, ".raw");
if (!File.Exists(rawFileName))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, rawFileName);
}
var run = PbfLcMsRun.GetLcMsRun(rawFileName);
// Parse ID file
const string resultFilePath = @"H:\Research\QCShew_TopDown\Production\M1_V092\QC_Shew_Intact_26Sep14_Bane_C2Column3_IcTda.tsv";
if (!File.Exists(resultFilePath))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, resultFilePath);
}
var resultParser = new MsPathFinderParser(resultFilePath);
const double qValueThreshold = 0.01;
var idList = resultParser.GetIdWithQValuesNoLargerThan(qValueThreshold);
var idFlag = new bool[run.MaxLcScan + 1];
foreach (var id in idList) idFlag[id.Scan] = true;
const string fastaFilePath = @"H:\Research\QCShew_TopDown\Production\ID_002216_235ACCEA.fasta";
if (!File.Exists(fastaFilePath))
{
Assert.Ignore(@"Skipping test {0} since file not found: {1}", methodName, fastaFilePath);
}
// const string fastaFilePath = @"H:\Research\QCShew_TopDown\Production\ID_002216_235ACCEA.icsfldecoy.fasta";
var fastaDb = new FastaDatabase(fastaFilePath);
var searchableDb = new SearchableDatabase(fastaDb);
var numMs2Spectra = 0;
var numSpectraWithTag = 0;
var numSpectraWithMatchingTag = 0;
var numSpectraWithMatchedTagNoId = 0;
foreach (var ms2ScanNum in run.GetScanNumbers(2))
{
++numMs2Spectra;
var tags = tagParser.GetSequenceTags(ms2ScanNum);
if (tags != null)
{
++numSpectraWithTag;
foreach (var tag in tags)
{
if (searchableDb.Search(tag.Sequence) >= 0)
{
//Console.WriteLine(tag.Sequence);
++numSpectraWithMatchingTag;
if (!idFlag[ms2ScanNum]) ++numSpectraWithMatchedTagNoId;
break;
}
}
}
}
Console.WriteLine("Tag length: {0}", minTagLength);
Console.WriteLine("NumMs2Spectra: {0}", numMs2Spectra);
Console.WriteLine("NumMs2SpectraWithTags: {0} ({1})", numSpectraWithTag, numSpectraWithTag/(float)numMs2Spectra);
Console.WriteLine("NumMs2SpectraWithMatchingTags: {0} ({1})", numSpectraWithMatchingTag, numSpectraWithMatchingTag/(float)numMs2Spectra);
Console.WriteLine("NumMs2SpectraWithMatchingTagsWithNoId: {0} ({1})", numSpectraWithMatchedTagNoId, numSpectraWithMatchedTagNoId / (float)numMs2Spectra);
}
