/* Fetch the product ID from the given device path */
static inline uint16_t pidFromDevPath(std::string const& devpath)
{
try
{
auto t1 = searchOne(devpath, "VID_.*PID_([0-9a-fA-F]{4})");
if (t1.empty()) // prevent unnecessary exceptions
return 0;
return static_cast<uint16_t>(std::stoi(t1, nullptr, 16));
}
catch (std::invalid_argument &)
{
return 0;
}
}
void searchOneMaskTrim(struct dnaSeq *seq, boolean isProt,
struct genoFind *gf, FILE *outFile,
struct hash *maskHash,
long long *retTotalSize, int *retCount)
/* Search a single sequence against a single genoFind index. */
{
boolean maskQuery = (qMask != NULL);
boolean lcMask = (qMask != NULL && sameWord(qMask, "lower"));
Bits *qMaskBits = maskQuerySeq(seq, isProt, maskQuery, lcMask);
struct dnaSeq trimmedSeq;
ZeroVar(&trimmedSeq);
trimSeq(seq, &trimmedSeq);
if (qType == gftRna || qType == gftRnaX)
memSwapChar(trimmedSeq.dna, trimmedSeq.size, 'u', 't');
searchOne(&trimmedSeq, gf, outFile, isProt, maskHash, qMaskBits);
*retTotalSize += seq->size;
*retCount += 1;
bitFree(&qMaskBits);
}
// search - runs the searches
void SpectraSTMzXMLSearchTask::search() {
if (!m_params.indexCacheAll) {
// Not caching all entries. In this case, the queries have to be sorted by precursor m/z first, such that
// the cached window slides from low to high precursor m/z only once. (Otherwise, the cached entries will need to be swapped
// in and out repeatedly, defeating the purpose of caching.)
// There is a catch however. To be able to search out of order, one has to keep many mzXML files open, and most systems
// have a max file opened limit. In such case, we will need to divide the mzXML files into smaller batches. The library will
// will need to be read numBatches times, but the tradeoff is we won't need to keep all entries cached in memory by selecting
// the indexCacheAll option.
// Divide the mzXML files into equal batches of at most MAX_NUM_OPEN_FILES files.
unsigned int numBatches = ((unsigned int)m_searchFileNames.size() - 1) / MAX_NUM_OPEN_FILES + 1;
unsigned int batchStart = 0;
for (unsigned int b = 0; b < numBatches; b++) {
m_batchBoundaries.push_back(batchStart);
batchStart += (unsigned int)m_searchFileNames.size() / numBatches;
}
m_batchBoundaries.push_back((unsigned int)m_searchFileNames.size());
// For each batch, sort all spectra by precursor m/z, open the files, and set the search in motion
for (unsigned int batch = 0; batch < (unsigned int)m_batchBoundaries.size() - 1; batch++) {
// this will do the sorting. the vector m_scans will be populated with the sorted scans.
prepareSortedSearch((unsigned int)batch);
// open the output files and print the headers (e.g. the xml definitions, ms run info, etc)
for (unsigned int n = m_batchBoundaries[batch]; n < m_batchBoundaries[batch + 1]; n++) {
m_outputs[n]->openFile();
m_outputs[n]->printHeader();
}
// tracking search progress
ProgressCount pc(!g_quiet && !g_verbose, 1, (int)(m_scans.size()));
string msg("Searching");
pc.start(msg);
// create searches from the m_scans one-by-one, and search them
for (vector<pair<unsigned int, rampScanInfo*> >::iterator i = m_scans.begin(); i != m_scans.end(); i++) {
searchOne((*i).first, (*i).second);
pc.increment();
// done. we can delete the rampScanInfo object now.
delete (*i).second;
}
pc.done();
// log the search of the batch
stringstream searchLogss;
searchLogss << "Searched sorted scans ";
searchLogss << "(Max " << m_numScansInFile << " scans; " << m_numSearchedInFile << " searched, ";
searchLogss << m_numLikelyGoodInFile << " likely good; ";
if (m_numNotSelectedInFile > 0) {
searchLogss << m_numNotSelectedInFile << " not selected; ";
}
searchLogss << m_numFailedFilterInFile << " failed filter; " << m_numMissingInFile << " missing; " << m_numMS1InFile << " MS1)";
g_log->log("MZXML SEARCH", searchLogss.str());
// done with this batch. close the files so that we can open more files in the next batch
for (unsigned int n = m_batchBoundaries[batch]; n < m_batchBoundaries[batch + 1]; n++) {
delete (m_files[n].second); // the cramp objects, which will close the mzXML files
m_files[n].second = NULL;
m_outputs[n]->printFooter(); // the output files
m_outputs[n]->closeFile();
}
}
} else {
// This is the case where we're caching everything anyway. In this case, it is not necessary to sort
// by precursor m/z before searching. We simply open the files one by one and search the queries
// in the order they are read.
for (unsigned int n = 0; n < (unsigned int)m_searchFileNames.size(); n++) {
// open the file using cRamp
cRamp* cramp = new cRamp(m_searchFileNames[n].c_str());
if (!cramp->OK()) {
g_log->error("MZXML SEARCH", "Cannot open file \"" + m_searchFileNames[n] + "\". File skipped.");
delete (cramp);
continue;
}
// Read the run info to extract the number of scans
rampRunInfo* runInfo = cramp->getRunInfo();
if (!runInfo) {
// probably an empty file...
g_log->error("MZXML SEARCH", "Cannot open file \"" + m_searchFileNames[n] + "\". File skipped.");
delete (cramp);
continue;
}
rampInstrumentInfo* instr = cramp->getInstrumentInfo();
if (instr) {
m_outputs[n]->setInstrInfo(instr);
// delete (instr);
}
// open the output file
m_outputs[n]->openFile();
m_outputs[n]->printHeader();
int numScans = cramp->getLastScan();
delete (runInfo);
// parse out the file name to determine the query prefix. Note that the query string
// has the form <mzXML file name>.<scan num>.<scan num>.0
FileName fn;
parseFileName(m_searchFileNames[n], fn);
// m_files is a vector of (FileName, cRamp*)
m_files[n].first = fn;
m_files[n].second = cramp;
ProgressCount pc(!g_quiet && !g_verbose, 1, numScans);
stringstream msg;
msg << "Searching \"" << m_searchFileNames[n] << "\" " << "(" << n + 1 << " of " << m_searchFileNames.size() << ")";
pc.start(msg.str());
m_numScansInFile = numScans;
m_numNotSelectedInFile = 0;
m_numMissingInFile = 0;
m_numMS1InFile = 0;
m_numFailedFilterInFile = 0;
m_numSearchedInFile = 0;
m_numLikelyGoodInFile = 0;
for (int k = 1; k <= numScans; k++) {
pc.increment();
// Filter out all scans not in selected list (in this case,
// the selected list contains a list of scan numbers as strings
if (!m_searchAll && !isInSelectedList(SpectraSTQuery::constructQueryName(fn.name, k, 0))) {
m_numNotSelectedInFile++;
continue;
}
// get the scan header (no peak list) first to check whether it's MS2.
// it'd be a waste of time if we read all scans, including MS1
rampScanInfo* scanInfo = cramp->getScanHeaderInfo(k);
// check to make sure the scan is good, and is not MS1
if (!scanInfo || (!m_isMzData && scanInfo->m_data.acquisitionNum != k)) {
m_numMissingInFile++;
if (scanInfo) delete (scanInfo);
continue;
}
if (scanInfo->m_data.msLevel == 1) {
m_numMS1InFile++;
delete (scanInfo);
continue;
}
// now we can search
searchOne(n, scanInfo);
// done, can delete scanInfo
delete scanInfo;
}
pc.done();
// log the search of this file
stringstream searchLogss;
searchLogss << "Searched \"" << m_searchFileNames[n] + "\" ";
searchLogss << "(Max " << m_numScansInFile << " scans; " << m_numSearchedInFile << " searched, ";
searchLogss << m_numLikelyGoodInFile << " likely good; ";
if (m_numNotSelectedInFile > 0) {
searchLogss << m_numNotSelectedInFile << " not selected; ";
}
searchLogss << m_numFailedFilterInFile << " failed filter; " << m_numMissingInFile << " missing; " << m_numMS1InFile << " MS1)";
g_log->log("MZXML SEARCH", searchLogss.str());
// we can delete the cRamp object now that we're done with this file.
// this is in contrast to the case where we're opening all the files at once for
// sorting -- in that case the cRamp objects will be deleted at the end of all
// searches
delete (m_files[n].second);
m_files[n].second = NULL;
m_outputs[n]->printFooter();
m_outputs[n]->closeFile(); // just so we won't hit the File Open limit if there are too many files
}
}
m_searchTaskStats.logStats();
}