helpString += "The insert parameter allows you to set a quality scores threshold. In the case where we are trying to decide whether to keep a base or remove it because the base is compared to a gap in the other fragment, if the base has a quality score equal to or below the threshold we eliminate it. Default=20.\n";
helpString += "The processors parameter allows you to specify how many processors you would like to use. The default is 1. \n";
helpString += "The allfiles parameter will create separate group and fasta file for each grouping. The default is F.\n";
+
helpString += "The trimoverlap parameter allows you to trim the sequences to only the overlapping section. The default is F.\n";
helpString += "The make.contigs command should be in the following format: \n";
helpString += "make.contigs(ffastq=yourForwardFastqFile, rfastq=yourReverseFastqFile, align=yourAlignmentMethod) \n";
outputTypes["fasta"] = tempOutNames;
outputTypes["group"] = tempOutNames;
outputTypes["report"] = tempOutNames;
- }
+ }
catch(exception& e) {
m->errorOut(e, "MakeContigsCommand", "MakeContigsCommand");
exit(1);
temp = validParameter.validFile(parameters, "allfiles", false); if (temp == "not found") { temp = "F"; }
allFiles = m->isTrue(temp);
+
temp = validParameter.validFile(parameters, "trimoverlap", false); if (temp == "not found") { temp = "F"; }
trimOverlap = m->isTrue(temp);
outputNames.push_back(compositeScrapFastaFile); outputTypes["fasta"].push_back(compositeScrapFastaFile);
}
+ map<string, int> totalGroupCounts;
+
for (int l = 0; l < filesToProcess.size(); l++) {
m->mothurOut("\n>>>>>\tProcessing " + filesToProcess[l][0][0] + " (file " + toString(l+1) + " of " + toString(filesToProcess.size()) + ")\t<<<<<\n");
+ groupCounts.clear();
+ groupMap.clear();
vector<vector<string> > fastaFileNames;
createOligosGroup = false;
string outputGroupFileName;
m->mothurOut("Making contigs...\n");
createProcesses(filesToProcess[l], outFastaFile, outScrapFastaFile, outMisMatchFile, fastaFileNames, l);
- m->mothurOut("Done.\n");
+ m->mothurOut("Here...\n");
//remove temp fasta and qual files
for (int i = 0; i < processors; i++) { for(int j = 0; j < filesToProcess[l][i].size(); j++) { m->mothurRemove(filesToProcess[l][i][j]); } }
ofstream out;
string thisGroupName = thisOutputDir + m->getRootName(m->getSimpleName(it->first));
- thisGroupName += getOutputFileName("group",variables); outputNames.push_back(thisGroupName); outputTypes["group"].push_back(thisGroupName);
+ thisGroupName += getOutputFileName("group",variables); outputNames.push_back(thisGroupName); outputTypes["group"].push_back(thisGroupName);
m->openOutputFile(thisGroupName, out);
while (!in.eof()){
if (m->control_pressed) { break; }
Sequence currSeq(in); m->gobble(in);
- out << currSeq.getName() << '\t' << it->second << endl;
+ out << currSeq.getName() << '\t' << it->second << endl;
}
- in.close();
out.close();
+ in.close();
}
}
}
if (filesToProcess.size() > 1) { //merge into large combo files
- if (createFileGroup || createOligosGroup) {
- if (l == 0) {
+ if (createFileGroup || createOligosGroup) {
+ if (l == 0) {
ofstream outCGroup;
m->openOutputFile(compositeGroupFile, outCGroup); outCGroup.close();
outputNames.push_back(compositeGroupFile); outputTypes["group"].push_back(compositeGroupFile);
m->appendFiles(outputGroupFileName, compositeGroupFile);
if (!allFiles) { m->mothurRemove(outputGroupFileName); }
else { outputNames.push_back(outputGroupFileName); outputTypes["group"].push_back(outputGroupFileName); }
+
+ for (map<string, int>::iterator itGroups = groupCounts.begin(); itGroups != groupCounts.end(); itGroups++) {
+ map<string, int>::iterator itTemp = totalGroupCounts.find(itGroups->first);
+ if (itTemp == totalGroupCounts.end()) { totalGroupCounts[itGroups->first] = itGroups->second; } //new group create it in totalGroups
+ else { itTemp->second += itGroups->second; } //existing group, update total
+ }
}
if (l == 0) { m->appendFiles(outMisMatchFile, compositeMisMatchFile); }
else { m->appendFilesWithoutHeaders(outMisMatchFile, compositeMisMatchFile); }
outputNames.push_back(outMisMatchFile); outputTypes["report"].push_back(outMisMatchFile);
}
}else {
+ totalGroupCounts = groupCounts;
outputNames.push_back(outFastaFile); outputTypes["fasta"].push_back(outFastaFile);
outputNames.push_back(outScrapFastaFile); outputTypes["fasta"].push_back(outScrapFastaFile);
outputNames.push_back(outMisMatchFile); outputTypes["report"].push_back(outMisMatchFile);
- if (createFileGroup || createOligosGroup) { outputNames.push_back(outputGroupFileName); outputTypes["group"].push_back(outputGroupFileName); }
+ if (createFileGroup || createOligosGroup) {
+ outputNames.push_back(outputGroupFileName); outputTypes["group"].push_back(outputGroupFileName);
+ }
}
+ m->mothurOut("Done.\n");
}
+
m->mothurOut("It took " + toString(time(NULL) - start) + " secs to process " + toString(numReads) + " sequences.\n");
if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
//output group counts
m->mothurOutEndLine();
int total = 0;
- if (groupCounts.size() != 0) { m->mothurOut("Group count: \n"); }
- for (map<string, int>::iterator it = groupCounts.begin(); it != groupCounts.end(); it++) {
+ if (totalGroupCounts.size() != 0) { m->mothurOut("Group count: \n"); }
+ for (map<string, int>::iterator it = totalGroupCounts.begin(); it != totalGroupCounts.end(); it++) {
total += it->second; m->mothurOut(it->first + "\t" + toString(it->second)); m->mothurOutEndLine();
}
if (total != 0) { m->mothurOut("Total of all groups is " + toString(total)); m->mothurOutEndLine(); }
}
}
}
-
+
num = driver(files[process],
outputFasta + toString(getpid()) + ".temp",
outputScrapFasta + toString(getpid()) + ".temp",
}
}
}
-
-
+
contigsData* tempcontig = new contigsData(group, files[h], (outputFasta + extension), (outputScrapFasta + extension), (outputMisMatches + extension), align, m, match, misMatch, gapOpen, gapExtend, insert, deltaq, barcodes, primers, tempFASTAFileNames, barcodeNameVector, primerNameVector, pdiffs, bdiffs, tdiffs, createOligosGroup, createFileGroup, allFiles, trimOverlap, h);
pDataArray.push_back(tempcontig);
}
m->gobble(in);
+ if (m->debug) { m->mothurOut("[DEBUG]: group = " + group + ", forward = " + forward + ", reverse = " + reverse + ".\n"); }
//check to make sure both are able to be opened
ifstream in2;
}
//roligo = reverseOligo(roligo);
+ if (m->debug) { m->mothurOut("[DEBUG]: reading - " + roligo + ".\n"); }
+
group = "";
// get rest of line in case there is a primer name
while (!in.eof()) {
char c = in.get();
- if (c == 10 || c == 13){ break; }
+ if (c == 10 || c == 13 || c == -1){ break; }
else if (c == 32 || c == 9){;} //space or tab
else { group += c; }
}
oligosPair newPrimer(foligo, roligo);
+
+ if (m->debug) { m->mothurOut("[DEBUG]: primer pair " + newPrimer.forward + " " + newPrimer.reverse + ", and group = " + group + ".\n"); }
//check for repeat barcodes
string tempPair = foligo+roligo;
group = "";
while (!in.eof()) {
char c = in.get();
- if (c == 10 || c == 13){ break; }
+ if (c == 10 || c == 13 || c == -1){ break; }
else if (c == 32 || c == 9){;} //space or tab
else { group += c; }
}
projects / mothur.git / blobdiff