/// <summary>
/// Read the next spectrum from a _dta.txt file
/// </summary>
/// <remarks>
/// If mCombineIdenticalSpectra is true, combines spectra that have the same scan number but different charge state
/// </remarks>
/// <param name="spectrumInfo"></param>
/// <returns>True if a spectrum is found, otherwise false</returns>
public override bool ReadNextSpectrum(out SpectrumInfo spectrumInfo)
{
var spectrumFound = false;
try
{
if (ReadingAndStoringSpectra || mCurrentSpectrum is null)
{
mCurrentSpectrum = new SpectrumInfoMsMsText();
}
else
{
mCurrentSpectrum.Clear();
}
if (mFileReader is null)
{
spectrumInfo = new SpectrumInfo();
mErrorMessage = "Data file not currently open";
return(false);
}
AddNewRecentFileText(string.Empty, true, false);
var lastProgressUpdateLine = mInFileLineNumber;
while (!spectrumFound && mFileReader.Peek() > -1 && !mAbortProcessing)
{
string lineIn;
if (mHeaderSaved.Length > 0)
{
lineIn = mHeaderSaved;
mHeaderSaved = string.Empty;
}
else
{
lineIn = mFileReader.ReadLine();
if (lineIn != null)
{
mTotalBytesRead += lineIn.Length + 2;
}
mInFileLineNumber++;
}
// See if lineIn is nothing or starts with the comment line character (equals sign)
if (lineIn != null && lineIn.Trim().StartsWith(CommentLineStartChar.ToString()))
{
AddNewRecentFileText(lineIn);
{
mCurrentSpectrum.SpectrumTitleWithCommentChars = lineIn;
mCurrentSpectrum.SpectrumTitle = CleanupComment(lineIn, CommentLineStartChar, true);
ExtractScanInfoFromDtaHeader(mCurrentSpectrum.SpectrumTitle, out var scanNumberStart, out var scanNumberEnd, out var scanCount);
mCurrentSpectrum.ScanNumber = scanNumberStart;
mCurrentSpectrum.ScanNumberEnd = scanNumberEnd;
mCurrentSpectrum.ScanCount = scanCount;
mCurrentSpectrum.MSLevel = 2;
mCurrentSpectrum.SpectrumID = mCurrentSpectrum.ScanNumber;
}
// Read the next line, which should have the parent ion MH value and charge
if (mFileReader.Peek() > -1)
{
lineIn = mFileReader.ReadLine();
}
else
{
lineIn = string.Empty;
}
if (lineIn != null)
{
mTotalBytesRead += lineIn.Length + 2;
}
mInFileLineNumber++;
if (string.IsNullOrWhiteSpace(lineIn))
{
// Spectrum header is not followed by a parent ion value and charge; ignore the line
}
else
{
AddNewRecentFileText(lineIn);
// Parse the parent ion info and read the MsMs Data
spectrumFound = ReadSingleSpectrum(
mFileReader, lineIn,
out mCurrentMsMsDataList,
mCurrentSpectrum,
ref mInFileLineNumber,
ref lastProgressUpdateLine,
out var mostRecentLineIn);
if (spectrumFound)
{
mCurrentSpectrum.ClearMzAndIntensityData();
if (ReadTextDataOnly)
{
// Do not parse the text data to populate .MzList and .IntensityList
mCurrentSpectrum.PeaksCount = 0;
}
else
{
try
{
ParseMsMsDataList(mCurrentMsMsDataList, out var mzList, out var intensityList);
mCurrentSpectrum.PeaksCount = mzList.Count;
mCurrentSpectrum.StoreIons(mzList, intensityList);
mCurrentSpectrum.Validate(true, true);
}
catch (Exception ex)
{
mCurrentSpectrum.PeaksCount = 0;
spectrumFound = false;
}
}
}
if (spectrumFound && CombineIdenticalSpectra && mCurrentSpectrum.ParentIonCharges[0] == 2)
{
// See if the next spectrum is the identical data, but the charge is 3 (this is a common situation with .dta files prepared by Lcq_Dta)
lineIn = mostRecentLineIn;
if (string.IsNullOrWhiteSpace(lineIn) && mFileReader.Peek() > -1)
{
// Read the next line
lineIn = mFileReader.ReadLine();
if (lineIn != null)
{
mTotalBytesRead += lineIn.Length + 2;
}
mInFileLineNumber++;
}
if (lineIn != null && lineIn.StartsWith(CommentLineStartChar.ToString()))
{
mHeaderSaved = lineIn;
var compareTitle = CleanupComment(mHeaderSaved, CommentLineStartChar, true);
if (compareTitle.EndsWith("3.dta", StringComparison.OrdinalIgnoreCase))
{
if (string.Equals(mCurrentSpectrum.SpectrumTitle.Substring(0, mCurrentSpectrum.SpectrumTitle.Length - 5), compareTitle.Substring(0, compareTitle.Length - 5), StringComparison.OrdinalIgnoreCase))
{
// Yes, the spectra match
mCurrentSpectrum.ParentIonChargeCount = 2;
mCurrentSpectrum.ParentIonCharges[1] = 3;
mCurrentSpectrum.ChargeIs2And3Plus = true;
mHeaderSaved = string.Empty;
// Read the next set of lines until the next blank line or comment line is found
while (mFileReader.Peek() > -1)
{
lineIn = mFileReader.ReadLine();
mInFileLineNumber++;
// See if lineIn is blank or starts with an equals sign
if (lineIn != null)
{
mTotalBytesRead += lineIn.Length + 2;
if (lineIn.Trim().Length == 0)
{
break;
}
if (lineIn.Trim().StartsWith(CommentLineStartChar.ToString()))
{
mHeaderSaved = lineIn;
break;
}
}
}
}
}
}
}
else if (mostRecentLineIn.StartsWith(CommentLineStartChar.ToString()))
{
mHeaderSaved = mostRecentLineIn;
}
} // EndIf for spectrumFound = True
} // EndIf for lineIn.Trim.StartsWith(mCommentLineStartChar)
if (mInFileLineNumber - lastProgressUpdateLine >= 250 || spectrumFound)
{
lastProgressUpdateLine = mInFileLineNumber;
UpdateStreamReaderProgress();
}
}
spectrumInfo = mCurrentSpectrum;
if (spectrumFound)
{
mScanCountRead++;
if (!ReadingAndStoringSpectra)
{
if (mInputFileStats.ScanCount < mScanCountRead)
{
mInputFileStats.ScanCount = mScanCountRead;
}
UpdateFileStats(mInputFileStats.ScanCount, spectrumInfo.ScanNumber, false);
}
}
}
catch (Exception ex)
{
OnErrorEvent("Error in ReadNextSpectrum", ex);
spectrumInfo = new SpectrumInfo();
}
return(spectrumFound);
}
/// <summary>
/// Read a single .dta file
/// </summary>
/// <param name="inputFilePath"></param>
/// <param name="msmsDataList"></param>
/// <param name="msmsDataCount"></param>
/// <param name="spectrumInfoMsMsText"></param>
/// <returns>True if the file was successfully opened and a spectrum was read</returns>
// ReSharper disable once UnusedMember.Global
public bool ReadSingleDtaFile(string inputFilePath, out string[] msmsDataList, out int msmsDataCount, out SpectrumInfoMsMsText spectrumInfoMsMsText)
{
var spectrumFound = false;
var msMsDataList = new List <string>();
msmsDataCount = 0;
spectrumInfoMsMsText = new SpectrumInfoMsMsText();
try
{
using var fileReader = new StreamReader(inputFilePath);
mTotalBytesRead = 0L;
ResetProgress("Parsing " + Path.GetFileName(inputFilePath));
mInFileLineNumber = 0;
var lastProgressUpdateLine = mInFileLineNumber;
while (!fileReader.EndOfStream && !mAbortProcessing)
{
var lineIn = fileReader.ReadLine();
mInFileLineNumber++;
if (lineIn != null)
{
mTotalBytesRead += lineIn.Length + 2;
}
if (!string.IsNullOrWhiteSpace(lineIn))
{
if (char.IsDigit(lineIn.Trim(), 0))
{
spectrumFound = ReadSingleSpectrum(
fileReader,
lineIn,
out msMsDataList,
spectrumInfoMsMsText,
ref mInFileLineNumber,
ref lastProgressUpdateLine,
out _);
break;
}
}
if (mInFileLineNumber - lastProgressUpdateLine >= 100)
{
lastProgressUpdateLine = mInFileLineNumber;
// MyBase.UpdateProgress(inFile.BaseStream.Position / inFile.BaseStream.Length * 100.0)
UpdateProgress(mTotalBytesRead / (double)fileReader.BaseStream.Length * 100.0d);
}
}
if (spectrumFound)
{
// Try to determine the scan numbers by parsing inputFilePath
ExtractScanInfoFromDtaHeader(Path.GetFileName(inputFilePath), out var scanNumberStart, out var scanNumberEnd, out var scanCount);
spectrumInfoMsMsText.ScanNumber = scanNumberStart;
spectrumInfoMsMsText.ScanNumberEnd = scanNumberEnd;
spectrumInfoMsMsText.ScanCount = scanCount;
msmsDataList = new string[msMsDataList.Count];
msMsDataList.CopyTo(msmsDataList);
}
else
{
msmsDataList = new string[1];
}
if (!mAbortProcessing)
{
OperationComplete();
}
}
catch (Exception ex)
{
OnErrorEvent("Error in ReadSingleDtaFile", ex);
spectrumInfoMsMsText = new SpectrumInfoMsMsText();
msmsDataList = new string[1];
}
return(spectrumFound);
}
/// <summary>
/// Read a single mass spectrum
/// </summary>
/// <param name="reader"></param>
/// <param name="parentIonLineText"></param>
/// <param name="msMsDataList"></param>
/// <param name="spectrumInfoMsMsText"></param>
/// <param name="linesRead"></param>
/// <param name="lastProgressUpdateLine"></param>
/// <param name="mostRecentLineIn"></param>
/// <returns>if a valid spectrum is found, otherwise, false</returns>
private bool ReadSingleSpectrum(
TextReader reader,
string parentIonLineText,
out List <string> msMsDataList,
SpectrumInfoMsMsText spectrumInfoMsMsText,
ref int linesRead,
ref int lastProgressUpdateLine,
out string mostRecentLineIn)
{
var spectrumFound = false;
spectrumInfoMsMsText.ParentIonLineText = parentIonLineText;
parentIonLineText = parentIonLineText.Trim();
// Look for the first space
var charIndex = parentIonLineText.IndexOf(' ');
if (charIndex >= 1)
{
var parentIonInfo = parentIonLineText.Substring(0, charIndex);
if (double.TryParse(parentIonInfo, out var parentIonMH))
{
spectrumInfoMsMsText.ParentIonMH = parentIonMH;
var chargeText = parentIonLineText.Substring(charIndex + 1);
// See if chargeText contains another space
var spaceIndex = chargeText.IndexOf(' ');
if (spaceIndex > 0)
{
chargeText = chargeText.Substring(0, spaceIndex);
}
if (int.TryParse(chargeText, out var charge))
{
spectrumInfoMsMsText.ParentIonChargeCount = 1;
spectrumInfoMsMsText.ParentIonCharges[0] = charge;
// Note: DTA files have Parent Ion MH defined but not Parent Ion m/z
// Thus, compute .ParentIonMZ using .ParentIonMH
if (spectrumInfoMsMsText.ParentIonCharges[0] <= 1)
{
spectrumInfoMsMsText.ParentIonMZ = spectrumInfoMsMsText.ParentIonMH;
spectrumInfoMsMsText.ParentIonCharges[0] = 1;
}
else
{
spectrumInfoMsMsText.ParentIonMZ = ConvoluteMass(spectrumInfoMsMsText.ParentIonMH, 1, spectrumInfoMsMsText.ParentIonCharges[0]);
}
spectrumFound = true;
}
}
}
mostRecentLineIn = string.Empty;
msMsDataList = new List <string>();
if (!spectrumFound)
{
return(false);
}
// Read all of the MS/MS spectrum ions up to the next blank line or up to the next line starting with COMMENT_LINE_START_CHAR
while (reader.Peek() > -1)
{
var lineIn = reader.ReadLine();
linesRead++;
// See if lineIn is blank
if (lineIn != null)
{
mTotalBytesRead += lineIn.Length + 2;
mostRecentLineIn = lineIn;
if (lineIn.Trim().Length == 0 || lineIn.StartsWith(COMMENT_LINE_START_CHAR.ToString()))
{
break;
}
// Add to MS/MS data string list
msMsDataList.Add(lineIn.Trim());
AddNewRecentFileText(lineIn);
}
if (linesRead - lastProgressUpdateLine >= 250)
{
lastProgressUpdateLine = linesRead;
UpdateStreamReaderProgress();
}
}
return(true);
}
/// <summary>
/// Read the next spectrum from a .mgf file
/// </summary>
/// <param name="spectrumInfo"></param>
/// <returns>True if a spectrum is found, otherwise false</returns>
public override bool ReadNextSpectrum(out SpectrumInfo spectrumInfo)
{
var sepChars = new[] { ' ', '\t' };
var spectrumFound = false;
try
{
if (ReadingAndStoringSpectra || mCurrentSpectrum is null)
{
mCurrentSpectrum = new SpectrumInfoMsMsText();
}
else
{
mCurrentSpectrum.Clear();
}
var scanNumberFound = false;
// Initialize mCurrentMsMsDataList
if (mCurrentMsMsDataList is null)
{
mCurrentMsMsDataList = new List <string>();
}
else
{
mCurrentMsMsDataList.Clear();
}
if (mFileReader is null)
{
spectrumInfo = new SpectrumInfoMsMsText();
mErrorMessage = "Data file not currently open";
return(false);
}
AddNewRecentFileText(string.Empty, true, false);
{
mCurrentSpectrum.SpectrumTitleWithCommentChars = string.Empty;
mCurrentSpectrum.SpectrumTitle = string.Empty;
mCurrentSpectrum.MSLevel = 2;
}
var lastProgressUpdateLine = mInFileLineNumber;
while (!spectrumFound && mFileReader.Peek() > -1 && !mAbortProcessing)
{
var lineIn = mFileReader.ReadLine();
if (lineIn != null)
{
mTotalBytesRead += lineIn.Length + 2;
}
mInFileLineNumber++;
if (lineIn?.Trim().Length > 0)
{
AddNewRecentFileText(lineIn);
lineIn = lineIn.Trim();
// See if lineIn starts with the comment line start character (a pound sign, #)
if (lineIn.StartsWith(CommentLineStartChar.ToString()))
{
// Remove any comment characters at the start of lineIn
lineIn = lineIn.TrimStart(CommentLineStartChar).Trim();
// Look for LINE_START_MSMS in lineIn
// This will be present in MGF files created using Agilent's DataAnalysis software
if (lineIn.StartsWith(LINE_START_MSMS, StringComparison.OrdinalIgnoreCase))
{
lineIn = lineIn.Substring(LINE_START_MSMS.Length).Trim();
// Initialize these values
mCurrentSpectrum.ScanNumberEnd = 0;
mCurrentSpectrum.ScanCount = 1;
// Remove the # sign in front of the scan number
lineIn = lineIn.TrimStart('#').Trim();
// Look for the / sign and remove any text following it
// For example,
// ###MS: 4458/4486/
// ###MSMS: 4459/4488/
// The / sign is used to indicate that several MS/MS scans were combined to make the given spectrum; we'll just keep the first one
var charIndex = lineIn.IndexOf('/');
if (charIndex > 0)
{
string temp;
if (charIndex < lineIn.Length - 1)
{
temp = lineIn.Substring(charIndex + 1).Trim();
}
else
{
temp = string.Empty;
}
lineIn = lineIn.Substring(0, charIndex).Trim();
mCurrentSpectrum.ScanCount = 1;
if (temp.Length > 0)
{
while (true)
{
charIndex = temp.IndexOf('/');
if (charIndex > 0)
{
mCurrentSpectrum.ScanCount++;
if (charIndex < temp.Length - 1)
{
temp = temp.Substring(charIndex + 1).Trim();
}
else
{
temp = temp.Substring(0, charIndex).Trim();
break;
}
}
else
{
break;
}
}
if (IsNumber(temp))
{
mCurrentSpectrum.ScanNumberEnd = int.Parse(temp);
}
}
}
scanNumberFound = ExtractScanRange(lineIn, mCurrentSpectrum);
}
}
// Line does not start with a comment character
// Look for LINE_START_BEGIN_IONS in lineIn
else if (lineIn.StartsWith(LINE_START_BEGIN_IONS, StringComparison.OrdinalIgnoreCase))
{
if (!scanNumberFound)
{
// Need to update scanNumberStart
// Set it to one more than mScanNumberStartSaved
mCurrentSpectrum.ScanNumber = mScanNumberStartSaved + 1;
mCurrentSpectrum.ScanNumberEnd = mCurrentSpectrum.ScanNumber;
mCurrentSpectrum.SpectrumID = mCurrentSpectrum.ScanNumber;
mCurrentSpectrum.ScanCount = 1;
}
var parentIonFound = false;
// We have found an MS/MS scan
// Look for LINE_START_PEPMASS and LINE_START_CHARGE to determine the parent ion m/z and charge
while (mFileReader.Peek() > -1)
{
lineIn = mFileReader.ReadLine();
mInFileLineNumber++;
if (lineIn == null)
{
continue;
}
mTotalBytesRead += lineIn.Length + 2;
AddNewRecentFileText(lineIn);
if (lineIn.Trim().Length == 0)
{
continue;
}
lineIn = lineIn.Trim();
string[] splitLine;
if (lineIn.StartsWith(LINE_START_PEPMASS, StringComparison.OrdinalIgnoreCase))
{
// This line defines the peptide mass as an m/z value
// It may simply contain the m/z value, or it may also contain an intensity value
// The two values will be separated by a space or a tab
// We do not save the intensity value since it cannot be included in a .Dta file
lineIn = lineIn.Substring(LINE_START_PEPMASS.Length).Trim();
splitLine = lineIn.Split(sepChars);
if (splitLine.Length > 0 && IsNumber(splitLine[0]))
{
mCurrentSpectrum.ParentIonMZ = double.Parse(splitLine[0]);
parentIonFound = true;
}
else
{
// Invalid LINE_START_PEPMASS Line
// Ignore this entire scan
break;
}
}
else if (lineIn.StartsWith(LINE_START_CHARGE, StringComparison.OrdinalIgnoreCase))
{
// This line defines the peptide charge
// It may simply contain a single charge, like 1+ or 2+
// It may also contain two charges, as in 2+ and 3+
// Not all spectra in the MGF file will have a CHARGE= entry
lineIn = lineIn.Substring(LINE_START_CHARGE.Length).Trim();
// Remove any + signs in the line
lineIn = lineIn.Replace("+", string.Empty);
if (lineIn.IndexOf(' ') > 0)
{
// Multiple charges may be present
splitLine = lineIn.Split(sepChars);
var indexEnd = splitLine.Length - 1;
for (var index = 0; index <= indexEnd; index++)
{
// Step through the split line and add any numbers to the charge list
// Typically, splitLine(1) will contain "and"
if (IsNumber(splitLine[index].Trim()))
{
if (mCurrentSpectrum.ParentIonChargeCount < SpectrumInfoMsMsText.MAX_CHARGE_COUNT)
{
mCurrentSpectrum.ParentIonCharges[mCurrentSpectrum.ParentIonChargeCount] =
int.Parse(splitLine[index].Trim());
mCurrentSpectrum.ParentIonChargeCount++;
}
}
}
}
else if (IsNumber(lineIn))
{
mCurrentSpectrum.ParentIonChargeCount = 1;
mCurrentSpectrum.ParentIonCharges[0] = int.Parse(lineIn);
}
}
else if (lineIn.StartsWith(LINE_START_TITLE, StringComparison.OrdinalIgnoreCase))
{
mCurrentSpectrum.SpectrumTitle = lineIn;
lineIn = lineIn.Substring(LINE_START_TITLE.Length).Trim();
mCurrentSpectrum.SpectrumTitleWithCommentChars = lineIn;
if (!scanNumberFound)
{
// We didn't find a scan number in a ### MSMS: comment line
// Attempt to extract out the scan numbers from the Title
{
ExtractScanInfoFromDtaHeader(lineIn, out var scanNumberStart, out var scanNumberEnd, out var scanCount);
mCurrentSpectrum.ScanNumber = scanNumberStart;
mCurrentSpectrum.ScanNumberEnd = scanNumberEnd;
mCurrentSpectrum.ScanCount = scanCount;
}
}
}
else if (lineIn.StartsWith(LINE_START_END_IONS, StringComparison.OrdinalIgnoreCase))
{
// Empty ion list
break;
}
else if (lineIn.StartsWith(LINE_START_RT, StringComparison.OrdinalIgnoreCase))
{
lineIn = lineIn.Substring(LINE_START_RT.Length).Trim();
if (double.TryParse(lineIn, out var rtSeconds))
{
mCurrentSpectrum.RetentionTimeMin = (float)(rtSeconds / 60.0d);
}
}
else if (lineIn.StartsWith(LINE_START_SCANS, StringComparison.OrdinalIgnoreCase))
{
lineIn = lineIn.Substring(LINE_START_SCANS.Length).Trim();
scanNumberFound = ExtractScanRange(lineIn, mCurrentSpectrum);
}
else if (char.IsNumber(lineIn, 0))
{
// Found the start of the ion list
// Add to the MsMs data list
if (parentIonFound)
{
mCurrentMsMsDataList.Add(lineIn);
}
break;
}
}
if (parentIonFound && mCurrentMsMsDataList.Count > 0)
{
// We have determined the parent ion
// Note: MGF files have Parent Ion MZ defined but not Parent Ion MH
// Thus, compute .ParentIonMH using .ParentIonMZ
{
if (mCurrentSpectrum.ParentIonChargeCount >= 1)
{
mCurrentSpectrum.ParentIonMH = ConvoluteMass(mCurrentSpectrum.ParentIonMZ, mCurrentSpectrum.ParentIonCharges[0], 1);
}
else
{
mCurrentSpectrum.ParentIonMH = mCurrentSpectrum.ParentIonMZ;
}
}
// Read in the ions and populate mCurrentMsMsDataList
// Read all of the MS/MS spectrum ions up to the next blank line or up to LINE_START_END_IONS
while (mFileReader.Peek() > -1)
{
lineIn = mFileReader.ReadLine();
mInFileLineNumber++;
// See if lineIn is blank
if (lineIn != null)
{
mTotalBytesRead += lineIn.Length + 2;
AddNewRecentFileText(lineIn);
if (lineIn.Trim().Length > 0)
{
if (lineIn.Trim().StartsWith(LINE_START_END_IONS, StringComparison.OrdinalIgnoreCase))
{
break;
}
// Add to MS/MS data sting list
mCurrentMsMsDataList.Add(lineIn.Trim());
}
}
if (mInFileLineNumber - lastProgressUpdateLine >= 250)
{
lastProgressUpdateLine = mInFileLineNumber;
UpdateStreamReaderProgress();
}
}
spectrumFound = true;
mCurrentSpectrum.ClearMzAndIntensityData();
if (ReadTextDataOnly)
{
// Do not parse the text data to populate .MZList and .IntensityList
mCurrentSpectrum.PeaksCount = 0;
}
else
{
try
{
ParseMsMsDataList(mCurrentMsMsDataList, out var mzList, out var intensityList);
mCurrentSpectrum.PeaksCount = mzList.Count;
mCurrentSpectrum.StoreIons(mzList, intensityList);
mCurrentSpectrum.Validate(true, true);
}
catch (Exception ex)
{
mCurrentSpectrum.PeaksCount = 0;
spectrumFound = false;
}
}
}
// Copy the scan number to mScanNumberStartSaved
if (mCurrentSpectrum.ScanNumber > 0)
{
mScanNumberStartSaved = mCurrentSpectrum.ScanNumber;
}
}
}
if (mInFileLineNumber - lastProgressUpdateLine < 250 && !spectrumFound)
{
continue;
}
lastProgressUpdateLine = mInFileLineNumber;
if (mFileReader is StreamReader streamReader)
{
UpdateProgress(streamReader.BaseStream.Position / (double)streamReader.BaseStream.Length * 100.0d);
}
else if (mInFileStreamLength > 0L)
{
UpdateProgress(mTotalBytesRead / (double)mInFileStreamLength * 100.0d);
}
}
spectrumInfo = mCurrentSpectrum;
if (spectrumFound)
{
mScanCountRead++;
if (!ReadingAndStoringSpectra)
{
if (mInputFileStats.ScanCount < mScanCountRead)
{
mInputFileStats.ScanCount = mScanCountRead;
}
UpdateFileStats(mInputFileStats.ScanCount, spectrumInfo.ScanNumber, false);
}
}
}
catch (Exception ex)
{
OnErrorEvent("Error in ReadNextSpectrum", ex);
spectrumInfo = new SpectrumInfoMsMsText();
}
return(spectrumFound);
}
