public void TestGetScanInfo(string mzXmlFileName, int expectedScanCount, int scanStart, int scanEnd, int expectedMS1, int expectedMS2)
{
// Keys in this dictionary are the scan number whose metadata is being retrieved
var expectedScanInfo = MSDataFileReaderTests.GetExpectedMzXmlScanInfo(mzXmlFileName);
var dataFile = FindInputFile(mzXmlFileName);
var reader = new MzXMLFileAccessor();
reader.OpenFile(dataFile.FullName);
MSDataFileReaderTests.ValidateScanInfo(reader, expectedScanCount, scanStart, scanEnd, expectedMS1, expectedMS2, expectedScanInfo);
}
public void TestGetNumScans(string mzXmlFileName, int expectedResult)
{
var dataFile = FindInputFile(mzXmlFileName);
var reader = new MzXMLFileAccessor();
reader.OpenFile(dataFile.FullName);
var scanCount = reader.ScanCount;
Console.WriteLine("Scan count for {0}: {1}", dataFile.Name, scanCount);
if (expectedResult >= 0)
{
Assert.AreEqual(expectedResult, scanCount, "Scan count mismatch");
}
}
public void TestGetParentScan(string mzXmlFileName, int startScan, int endScan, params int[] expectedParents)
{
var dataFile = FindInputFile(mzXmlFileName);
var reader = new MzXMLFileAccessor();
reader.OpenFile(dataFile.FullName);
var i = 0;
var validScanCount = 0;
for (var scanNumber = startScan; scanNumber <= endScan; scanNumber++)
{
var validScan = reader.GetSpectrumByScanNumber(scanNumber, out var spectrumInfo);
if (!validScan)
{
ConsoleMsgUtils.ShowWarning("Invalid scan number: {0}", scanNumber);
i++;
continue;
}
validScanCount++;
if (spectrumInfo is not SpectrumInfoMzXML mzXmlSpectrum)
{
Assert.Fail("Input file is not a .mzXML file; cannot validate precursor scan numbers");
return;
}
Console.WriteLine("MS{0} scan {1,-4} has parent {2,-4}", spectrumInfo.MSLevel, scanNumber, mzXmlSpectrum.PrecursorScanNum);
if (i < expectedParents.Length && expectedParents[i] != 0)
{
Assert.AreEqual(
expectedParents[i], mzXmlSpectrum.PrecursorScanNum,
"Parent scan does not match expected value: {0}",
expectedParents[i]);
}
i++;
}
var percentValid = validScanCount / (double)(endScan - startScan + 1) * 100;
Assert.Greater(percentValid, 90, "Over 10% of the spectra had invalid scan numbers");
}
public void TestMzXmlAccessor(string mzXmlFileName, int expectedScanCount, int scanStart, int scanEnd, int expectedMS1, int expectedMS2, bool cacheIfNoIndex = false)
{
var expectedScanInfo = GetExpectedMzXmlScanInfo(mzXmlFileName);
var dataFile = FindInputFile(mzXmlFileName);
var reader = new MzXMLFileAccessor();
reader.OpenFile(dataFile.FullName);
if (reader.IndexedSpectrumCount == 0 && cacheIfNoIndex)
{
// Cache the entire file in memory
reader.ReadAndCacheEntireFileNonIndexed();
}
ValidateScanInfo(reader, expectedScanCount, scanStart, scanEnd, expectedMS1, expectedMS2, expectedScanInfo);
}
public void TestGetCollisionEnergy(string mzXmlFileName)
{
// Keys in this Dictionary are filename, values are Collision Energies by scan
var expectedData = new Dictionary <string, Dictionary <int, List <double> > >();
var ce30 = new List <double> {
30.00
};
var ce45 = new List <double> {
45.00
};
var ce120 = new List <double> {
120.55119
};
var ms1Scan = new List <double>();
var ms2ScanWithoutCollisionMode = new List <double>();
// Keys in this dictionary are scan number and values are collision energies
var file1Data = new Dictionary <int, List <double> >
{
{ 2250, ce45 },
{ 2251, ce45 },
{ 2252, ce45 },
{ 2253, ms1Scan },
{ 2254, ce45 },
{ 2255, ce45 },
{ 2256, ce45 },
{ 2257, ms1Scan },
{ 2258, ce45 },
{ 2259, ce45 },
{ 2260, ce45 }
};
expectedData.Add("Shew_246a_LCQa_15Oct04_Andro_0904-2_4-20", file1Data);
var file2Data = new Dictionary <int, List <double> >
{
{ 19000, ms2ScanWithoutCollisionMode },
{ 19001, ce120 },
{ 19002, ce120 },
{ 19003, ms1Scan },
{ 19004, ce30 },
{ 19005, ce30 },
{ 19006, ce30 },
{ 19007, ms2ScanWithoutCollisionMode },
{ 19008, ce120 },
{ 19009, ce120 },
{ 19010, ce30 }
};
expectedData.Add("HCC-38_ETciD_EThcD_07Jan16_Pippin_15-08-53", file2Data);
var file3Data = new Dictionary <int, List <double> >
{
{ 1, ms2ScanWithoutCollisionMode },
{ 2, ms2ScanWithoutCollisionMode }
};
expectedData.Add("MZ0210MnxEF889ETD", file3Data);
var file4Data = new Dictionary <int, List <double> >
{
{ 27799, ms1Scan },
{ 27800, ce30 },
{ 27801, ce30 },
{ 27802, ce30 },
{ 27803, ce30 },
{ 27804, ce30 },
{ 27805, ms1Scan },
{ 27806, ce30 },
{ 27807, ce30 }
};
expectedData.Add("QC_Mam_16_01_125ng_2pt0-IT22_Run-A_16Oct17_Pippin_AQ_17-10-01", file4Data);
var dataFile = FindInputFile(mzXmlFileName);
if (!expectedData.TryGetValue(Path.GetFileNameWithoutExtension(dataFile.Name), out var collisionEnergiesThisFile))
{
Assert.Fail("Dataset {0} not found in dictionary expectedData", dataFile.Name);
}
// Keys are scan number, values are the collision energy
var collisionEnergiesActual = new Dictionary <int, float>();
// Keys are scan number, values are msLevel
var msLevelsActual = new Dictionary <int, int>();
// Keys are scan number, values are the ActivationType, for example cid, etd, hcd
var activationTypesActual = new Dictionary <int, string>();
var reader = new MzXMLFileAccessor();
reader.OpenFile(dataFile.FullName);
foreach (var scanNumber in collisionEnergiesThisFile.Keys)
{
var validScan = reader.GetSpectrumByScanNumber(scanNumber, out var spectrumInfo);
Assert.IsTrue(validScan, "GetSpectrumByScanNumber returned false for scan {0}", scanNumber);
if (spectrumInfo is not SpectrumInfoMzXML mzXmlSpectrum)
{
Assert.Fail("Input file is not a .mzXML file; cannot validate collision energy");
return;
}
collisionEnergiesActual.Add(scanNumber, mzXmlSpectrum.CollisionEnergy);
msLevelsActual.Add(scanNumber, mzXmlSpectrum.MSLevel);
activationTypesActual.Add(scanNumber, mzXmlSpectrum.ActivationMethod);
}
Console.WriteLine("{0,-5} {1,-5} {2}", "Valid", "Scan", "Collision Energy");
foreach (var actualEnergyOneScan in (from item in collisionEnergiesActual orderby item.Key select item))
{
var scanNumber = actualEnergyOneScan.Key;
var expectedEnergies = collisionEnergiesThisFile[scanNumber];
var activationTypes = string.Join(", ", activationTypesActual[scanNumber]);
if (Math.Abs(actualEnergyOneScan.Value) < float.Epsilon)
{
var msLevel = msLevelsActual[scanNumber];
if (msLevel != 1)
{
if (expectedEnergies.Count == 0)
{
var warning = string.Format(
"Scan {0} has msLevel={1} and activationType={2}, but no collision energy. This isn't typically true, but can happen in .mzXML files",
scanNumber, msLevel, activationTypes);
Console.WriteLine(warning);
continue;
}
var msg = string.Format(
"Scan {0} has no collision energy, which should only be true for spectra with msLevel=1. This scan has msLevel={1} and activationType={2}",
scanNumber, msLevel, activationTypes);
Console.WriteLine(msg);
Assert.Fail(msg);
}
else
{
Console.WriteLine("{0,-5} {1,-5} {2}", true, scanNumber, "MS1 scan");
}
}
else
{
var actualEnergy = actualEnergyOneScan.Value;
var isValid = expectedEnergies.Any(expectedEnergy => Math.Abs(actualEnergy - expectedEnergy) < 0.00001);
Console.WriteLine("{0,-5} {1,-5} {2:F2}", isValid, scanNumber, actualEnergy);
Assert.IsTrue(isValid, "Unexpected collision energy {0:F2} for scan {1}", actualEnergy, scanNumber);
}
}
}
public void TestDataIsSortedByMz(string mzXmlFileName, int scanStart, int scanEnd, int scanStep)
{
var dataFile = FindInputFile(mzXmlFileName);
var reader = new MzXMLFileAccessor();
reader.OpenFile(dataFile.FullName);
for (var iteration = 1; iteration <= 4; iteration++)
{
int maxNumberOfPeaks;
bool centroidData;
switch (iteration)
{
case 1:
maxNumberOfPeaks = 0;
centroidData = false;
break;
case 2:
maxNumberOfPeaks = 0;
centroidData = true;
break;
case 3:
maxNumberOfPeaks = 50;
centroidData = false;
break;
default:
maxNumberOfPeaks = 50;
centroidData = true;
break;
}
if (iteration == 1)
{
Console.WriteLine("Scan data for {0}", dataFile.Name);
Console.WriteLine(
"{0,5} {1,-5} {2,-10} {3,-8} {4,-8} {5,-10} {6,-8} {7,-10} {8,-8} {9}",
"Scan", "Max#", "Centroid", "MzCount", "IntCount",
"FirstMz", "FirstInt", "MidMz", "MidInt", "ScanFilter");
}
if (scanEnd > reader.CachedSpectraScanNumberMaximum)
{
scanEnd = reader.CachedSpectraScanNumberMaximum;
}
if (scanStep < 1)
{
scanStep = 1;
}
var statsInterval = (int)Math.Floor((scanEnd - scanStart) / (double)scanStep / 10);
var scansProcessed = 0;
for (var scanNumber = scanStart; scanNumber <= scanEnd; scanNumber += scanStep)
{
var validScan = reader.GetSpectrumByScanNumber(scanNumber, out var spectrumInfo);
if (!validScan)
{
Assert.Fail("Invalid scan number: {0}", scanNumber);
}
var mzList = spectrumInfo.MzList;
var intensityList = spectrumInfo.IntensityList;
var unsortedMzValues = 0;
for (var i = 0; i < mzList.Count - 1; i++)
{
if (mzList[i] > mzList[i + 1])
{
unsortedMzValues++;
}
}
Assert.AreEqual(0, unsortedMzValues, "Scan {0} has {1} m/z values not sorted properly", scanNumber, unsortedMzValues);
scansProcessed++;
if (scansProcessed % statsInterval == 0)
{
var scanInfo = GenerateThermoScanFilter(spectrumInfo);
if (mzList.Count > 0)
{
var midIndex = (int)Math.Floor(mzList.Count / 2.0);
Console.WriteLine(
"{0,5} {1,-5} {2,-10} {3,-8} {4,-8} {5,-10:0.0000} {6,-8:0.0} {7,-10:0.0000} {8,-8:0.0} {9}",
scanNumber, maxNumberOfPeaks, centroidData, mzList.Count, intensityList.Count,
mzList[0], intensityList[0], mzList[midIndex], intensityList[midIndex], scanInfo);
}
else
{
Console.WriteLine(
"{0,5} {1,-5} {2,-10} {3,-8} {4,-8} {5,-10} {6,-8} {7,-10} {8,-8} {9}",
scanNumber, maxNumberOfPeaks, centroidData, mzList.Count, intensityList.Count,
"n/a", "n/a", "n/a", "n/a", scanInfo);
}
}
}
}
}