public void compressMS2(List <TempIndex> tempIndexList, BlockIndex index)
{
if (converter.jobInfo.jobParams.threadAccelerate)
{
Hashtable table = Hashtable.Synchronized(new Hashtable());
//使用多线程处理数据提取与压缩
Parallel.For(0, tempIndexList.Count, (i, ParallelLoopState) =>
{
TempIndex tempIndex = tempIndexList[i];
TempScan ts = new TempScan(tempIndex.num, tempIndex.rt, tempIndex.tic);
if (converter.jobInfo.jobParams.includeCV)
{
ts.cvs = tempIndex.cvList;
}
converter.compress(converter.spectrumList.spectrum(tempIndex.num, true), ts);
table.Add(i, ts);
});
converter.outputWithOrder(table, index);
}
else
{
foreach (TempIndex tempIndex in tempIndexList)
{
TempScan ts = new TempScan(tempIndex.num, tempIndex.rt, tempIndex.tic);
if (converter.jobInfo.jobParams.includeCV)
{
ts.cvs = tempIndex.cvList;
}
converter.compress(converter.spectrumList.spectrum(tempIndex.num, true), ts);
converter.addToIndex(index, ts);
}
}
}
protected new void outputWithOrder(Hashtable table, BlockIndex index)
{
ArrayList keys = new ArrayList(table.Keys);
keys.Sort();
foreach (int key in keys)
{
TempScan tempScan = (TempScan)table[key];
addToIndex(index, tempScan);
}
}
//注意:本函数会操作startPosition这个全局变量
protected new void addToIndex(BlockIndex index, TempScan ts)
{
index.nums.Add(ts.num);
index.rts.Add(ts.rt);
index.mzs.Add(startPosition);
index.ints.Add(startPosition + ts.mzArrayBytes.Length);
index.cvList.Add(ts.cvs);
startPosition = startPosition + ts.mzArrayBytes.Length + ts.intArrayBytes.Length;
index.endPtr = startPosition;
airdStream.Write(ts.mzArrayBytes, 0, ts.mzArrayBytes.Length);
airdStream.Write(ts.intArrayBytes, 0, ts.intArrayBytes.Length);
}
//处理MS2,由于每一个MS1只跟随少量的MS2光谱图,因此DDA采集模式下MS2的压缩模式仍然使用Aird ZDPD的压缩算法
private void doWithMS2Block()
{
jobInfo.log("Start Processing MS2 List");
ArrayList keys = new ArrayList(ms2Table.Keys);
keys.Sort();
foreach (int key in keys)
{
List <TempIndex> tempIndexList = ms2Table[key] as List <TempIndex>;
//为每一组key创建一个Block
BlockIndex blockIndex = new BlockIndex();
blockIndex.level = 2;
blockIndex.startPtr = startPosition;
blockIndex.num = key;
//创建这一个block中每一个ms2的窗口序列
List <WindowRange> ms2Ranges = new List <WindowRange>();
jobInfo.log(null, "MS2:" + progress + "/" + ms2Table.Keys.Count);
progress++;
foreach (TempIndex index in tempIndexList)
{
WindowRange range = new WindowRange(index.mzStart, index.mzEnd, index.mz);
if (index.charge != 0)
{
range.charge = index.charge;
}
ms2Ranges.Add(range);
TempScan ts = new TempScan(index.num, index.rt, index.tic);
if (jobInfo.jobParams.includeCV)
{
ts.cvs = index.cvList;
}
compress(spectrumList.spectrum(index.num, true), ts);
blockIndex.nums.Add(ts.num);
blockIndex.rts.Add(ts.rt);
blockIndex.tics.Add(ts.tic);
if (jobInfo.jobParams.includeCV)
{
blockIndex.cvList.Add(ts.cvs);
}
blockIndex.mzs.Add(ts.mzArrayBytes.Length);
blockIndex.ints.Add(ts.intArrayBytes.Length);
startPosition = startPosition + ts.mzArrayBytes.Length + ts.intArrayBytes.Length;
airdStream.Write(ts.mzArrayBytes, 0, ts.mzArrayBytes.Length);
airdStream.Write(ts.intArrayBytes, 0, ts.intArrayBytes.Length);
}
blockIndex.rangeList = ms2Ranges;
blockIndex.endPtr = startPosition;
indexList.Add(blockIndex);
}
}
private void scan(int i)
{
Spectrum spectrum = spectrumList.spectrum(i, true);
//忽略所有MS1的谱图
if (spectrum.cvParamChild(CVID.MS_ms_level).value.ToString().Equals(MsLevel.MS1))
{
return;
}
double targetMz = getPrecursorIsolationWindowParams(spectrum, CVID.MS_isolation_window_target_m_z);
ArrayList ms2List = null;
if (rangeMap.Contains(targetMz))
{
ms2List = (ArrayList)ms2Map[targetMz];
}
else
{
double mz = getPrecursorIsolationWindowParams(spectrum, CVID.MS_isolation_window_target_m_z);
double lowerOffset = getPrecursorIsolationWindowParams(spectrum, CVID.MS_isolation_window_lower_offset);
double upperOffset = getPrecursorIsolationWindowParams(spectrum, CVID.MS_isolation_window_upper_offset);
WindowRange range = new WindowRange(mz - lowerOffset, mz + upperOffset, mz);
BlockIndex addIndex = new BlockIndex();
addIndex.setWindowRange(range);
rangeMap.Add(targetMz, addIndex);
ms2Map.Add(targetMz, ArrayList.Synchronized(new ArrayList()));
ms2List = (ArrayList)ms2Map[targetMz];
}
if (spectrum.scanList.scans.Count != 1)
{
return;
}
try
{
TempScan ts = new TempScan(i, parseRT(spectrum.scanList.scans[0]), parseTIC(spectrum));
if (jobInfo.jobParams.includeCV)
{
ts.cvs = CV.trans(spectrum);
}
compress(spectrum, ts);
ms2List.Add(ts);
}
catch (Exception exception)
{
jobInfo.log(exception.Message);
return;
}
}
private void parseAndStoreMS2Block()
{
jobInfo.log("Start Processing MS2 List");
foreach (double key in ms2Table.Keys)
{
List <TempIndex> tempIndexList = ms2Table[key] as List <TempIndex>;
//为每一个key组创建一个SwathBlock
BlockIndex swathIndex = new BlockIndex();
swathIndex.level = 2;
swathIndex.startPtr = startPosition;
//顺便创建一个WindowRanges,用以让Propro服务端快速获取全局的窗口数目和mz区间
WindowRange range = new WindowRange(tempIndexList[0].mzStart, tempIndexList[0].mzEnd, key);
swathIndex.setWindowRange(range);
ranges.Add(range);
jobInfo.log(null, "MS2:" + progress + "/" + ms2Table.Keys.Count);
progress++;
if (jobInfo.jobParams.threadAccelerate)
{
Hashtable table = Hashtable.Synchronized(new Hashtable());
//使用多线程处理数据提取与压缩
Parallel.For(0, tempIndexList.Count, (i, ParallelLoopState) =>
{
TempIndex index = tempIndexList[i];
TempScan ts = new TempScan(index.num, index.rt, index.tic);
compress(spectrumList.spectrum(index.num, true), ts);
table.Add(i, ts);
});
outputWithOrder(table, swathIndex);
}
else
{
foreach (TempIndex index in tempIndexList)
{
TempScan ts = new TempScan(index.pNum, index.rt, index.tic);
compress(spectrumList.spectrum(index.num, true), ts);
addToIndex(swathIndex, ts);
}
}
swathIndex.endPtr = startPosition;
indexList.Add(swathIndex);
jobInfo.log("MS2 Group Finished:" + progress + "/" + ms2Table.Keys.Count);
}
}
//注意:本函数会操作startPosition这个全局变量
public void addToIndex(BlockIndex index, object tempScan)
{
if (jobInfo.jobParams.useStackZDPD())
{
TempScanSZDPD ts = (TempScanSZDPD)tempScan;
index.nums.AddRange(ts.nums);
index.rts.AddRange(ts.rts);
index.tics.AddRange(ts.tics);
if (jobInfo.jobParams.includeCV)
{
index.cvList.AddRange(ts.cvs);
}
index.mzs.Add(ts.mzArrayBytes.Length);
index.ints.Add(ts.intArrayBytes.Length);
index.tags.Add(ts.tagArrayBytes.Length);
startPosition = startPosition + ts.mzArrayBytes.Length + ts.tagArrayBytes.Length + ts.intArrayBytes.Length;
airdStream.Write(ts.mzArrayBytes, 0, ts.mzArrayBytes.Length);
airdStream.Write(ts.tagArrayBytes, 0, ts.tagArrayBytes.Length);
airdStream.Write(ts.intArrayBytes, 0, ts.intArrayBytes.Length);
}
else
{
TempScan ts = (TempScan)tempScan;
index.nums.Add(ts.num);
index.rts.Add(ts.rt);
index.tics.Add(ts.tic);
if (jobInfo.jobParams.includeCV)
{
index.cvList.Add(ts.cvs);
}
index.mzs.Add(ts.mzArrayBytes.Length);
index.ints.Add(ts.intArrayBytes.Length);
startPosition = startPosition + ts.mzArrayBytes.Length + ts.intArrayBytes.Length;
airdStream.Write(ts.mzArrayBytes, 0, ts.mzArrayBytes.Length);
airdStream.Write(ts.intArrayBytes, 0, ts.intArrayBytes.Length);
}
}
public void compress(Spectrum spectrum, TempScan ts)
{
BinaryDataDouble mzData = spectrum.getMZArray().data;
BinaryDataDouble intData = spectrum.getIntensityArray().data;
var dataCount = mzData.Count;
int[] mzArray = new int[dataCount];
float[] intensityArray = new float[dataCount];
int j = 0;
for (int t = 0; t < dataCount; t++)
{
if (jobInfo.jobParams.ignoreZeroIntensity && intData[t] == 0)
{
continue;
}
int mz = Convert.ToInt32(mzData[t] * mzPrecision);
mzArray[j] = mz;
if (jobInfo.jobParams.log2)
{
intensityArray[j] = Convert.ToSingle(Math.Round(Math.Log(intData[t]) / log2, 3)); //取log10并且精确到小数点后3位
}
else
{
intensityArray[j] = Convert.ToSingle(Math.Round(intData[t], 1)); //精确到小数点后一位
}
j++;
}
int[] mzSubArray = new int[j];
Array.Copy(mzArray, mzSubArray, j);
float[] intensitySubArray = new float[j];
Array.Copy(intensityArray, intensitySubArray, j);
int[] compressedMzArray = CompressUtil.fastPforEncoder(mzSubArray);
ts.mzArrayBytes = CompressUtil.zlibEncoder(compressedMzArray);
ts.intArrayBytes = CompressUtil.zlibEncoder(intensitySubArray);
}
public void compressMS1(BlockIndex index)
{
if (converter.jobInfo.jobParams.threadAccelerate)
{
Hashtable table = Hashtable.Synchronized(new Hashtable());
//使用多线程处理数据提取与压缩
Parallel.For(0, converter.ms1List.Count, (i, ParallelLoopState) =>
{
converter.jobInfo.log(null, "MS1:" + i + "/" + converter.ms1List.Count);
TempIndex scanIndex = converter.ms1List[i];
TempScan ts = new TempScan(scanIndex.num, scanIndex.rt, scanIndex.tic);
if (converter.jobInfo.jobParams.includeCV)
{
ts.cvs = scanIndex.cvList;
}
converter.compress(converter.spectrumList.spectrum(scanIndex.num, true), ts);
table.Add(i, ts);
});
converter.outputWithOrder(table, index);
}
else
{
for (int i = 0; i < converter.ms1List.Count; i++)
{
converter.jobInfo.log(null, "MS1:" + i + "/" + converter.ms1List.Count);
TempIndex scanIndex = converter.ms1List[i];
TempScan ts = new TempScan(scanIndex.num, scanIndex.rt, scanIndex.tic);
if (converter.jobInfo.jobParams.includeCV)
{
ts.cvs = scanIndex.cvList;
}
converter.compress(converter.spectrumList.spectrum(scanIndex.num, true), ts);
converter.addToIndex(index, ts);
}
}
}
//解析MS1和MS2谱图
protected void doProgress()
{
var ms1Index = new BlockIndex(1);
var ms2Index = new BlockIndex(2);
jobInfo.log("Preprocessing:" + totalSize, "Preprocessing");
if (jobInfo.jobParams.threadAccelerate)
{
var ms1Table = Hashtable.Synchronized(new Hashtable());
var ms2Table = Hashtable.Synchronized(new Hashtable());
//使用多线程处理数据提取与压缩
Parallel.For(0, totalSize, (i, ParallelLoopState) =>
{
jobInfo.log(null, progress + "/" + totalSize);
progress++;
var spectrum = spectrumList.spectrum(i, true);
if (spectrum.scanList.scans.Count != 1)
{
ParallelLoopState.Break();
return;
}
var scan = spectrum.scanList.scans[0];
var ts = new TempScan(i, parseRT(scan), parseTIC(spectrum));
if (jobInfo.jobParams.includeCV)
{
ts.cvs = CV.trans(spectrum);
}
compress(spectrum, ts);
var msLevel = parseMsLevel(spectrum);
if (msLevel.Equals(MsLevel.MS1))
{
ms1Table.Add(i, ts);
}
else
{
if (activator == null)
{
parseActivator(spectrum.precursors[0].activation);
}
ms2Table.Add(i, ts);
}
});
outputWithOrder(ms1Table, ms1Index);
outputWithOrder(ms2Table, ms2Index);
}
else
{
for (var i = 0; i < totalSize; i++)
{
var spectrum = spectrumList.spectrum(i, true);
if (spectrum.scanList.scans.Count != 1)
{
continue;
}
var scan = spectrum.scanList.scans[0];
var ts = new TempScan(i, parseRT(scan), parseTIC(spectrum));
if (jobInfo.jobParams.includeCV)
{
ts.cvs = CV.trans(spectrum);
}
compress(spectrum, ts);
var msLevel = parseMsLevel(spectrum);
if (msLevel.Equals(MsLevel.MS1))
{
addToIndex(ms1Index, ts);
}
else
{
addToIndex(ms2Index, ts);
}
}
}
indexList.Add(ms1Index);
indexList.Add(ms2Index);
}
