vector<string> setParameters();
string getCommandName() { return "chimera.uchime"; }
string getCommandCategory() { return "Sequence Processing"; }
- string getOutputFileNameTag(string, string);
+
string getHelpString();
- string getCitation() { return "uchime by Robert C. Edgar\nhttp://drive5.com/uchime\nThis code is donated to the public domain.\nhttp://www.mothur.org/wiki/Chimera.uchime\nEdgar,R.C., Haas,B.J., Clemente,J.C., Quince,C. and Knight,R. (2011), UCHIME improves sensitivity and speed of chimera detection, Bioinformatics, in press.\n"; }
+ string getOutputPattern(string);
+ string getCitation() { return "uchime by Robert C. Edgar\nhttp://drive5.com/uchime\nThis code was donated to the public domain.\nEdgar,R.C., Haas,B.J., Clemente,J.C., Quince,C. and Knight,R. (2011), UCHIME improves sensitivity and speed of chimera detection. Bioinformatics 27:2194.\nhttp://www.mothur.org/wiki/Chimera.uchime\n"; }
string getDescription() { return "detect chimeric sequences"; }
int execute();
int driver(string, string, string, string, int&);
int createProcesses(string, string, string, string, int&);
- bool abort, useAbskew, chimealns, useMinH, useMindiv, useXn, useDn, useXa, useChunks, useMinchunk, useIdsmoothwindow, useMinsmoothid, useMaxp, skipgaps, skipgaps2, useMinlen, useMaxlen, ucl, useQueryfract, hasCount, hasName;
- string fastafile, groupfile, templatefile, outputDir, namefile, countfile, abskew, minh, mindiv, xn, dn, xa, chunks, minchunk, idsmoothwindow, minsmoothid, maxp, minlen, maxlen, queryfract, uchimeLocation;
+ bool abort, useAbskew, chimealns, useMinH, useMindiv, useXn, useDn, useXa, useChunks, useMinchunk, useIdsmoothwindow, useMinsmoothid, useMaxp, skipgaps, skipgaps2, useMinlen, useMaxlen, ucl, useQueryfract, hasCount, hasName, dups;
+ string fastafile, groupfile, templatefile, outputDir, namefile, countfile, abskew, minh, mindiv, xn, dn, xa, chunks, minchunk, idsmoothwindow, minsmoothid, maxp, minlen, maxlen, queryfract, uchimeLocation, strand;
int processors;
SequenceParser* sparser;
int deconvoluteResults(map<string, string>&, string, string, string);
int driverGroups(string, string, string, string, int, int, vector<string>);
int createProcessesGroups(string, string, string, string, vector<string>, string, string, string);
+ int prepFile(string filename, string);
};
int threadID, count, numChimeras;
vector<string> groups;
bool useAbskew, chimealns, useMinH, useMindiv, useXn, useDn, useXa, useChunks, useMinchunk, useIdsmoothwindow, useMinsmoothid, useMaxp, skipgaps, skipgaps2, useMinlen, useMaxlen, ucl, useQueryfract, hasCount;
- string abskew, minh, mindiv, xn, dn, xa, chunks, minchunk, idsmoothwindow, minsmoothid, maxp, minlen, maxlen, queryfract;
+ string abskew, minh, mindiv, xn, dn, xa, chunks, minchunk, idsmoothwindow, minsmoothid, maxp, minlen, maxlen, queryfract, strand;
uchimeData(){}
uchimeData(string o, string uloc, string t, string file, string f, string n, string g, string ac, string al, vector<string> gr, MothurOut* mout, int st, int en, int tid) {
hasCount = hc;
}
- void setVariables(string abske, string min, string mindi, string x, string d, string xa2, string chunk, string minchun, string idsmoothwindo, string minsmoothi, string max, string minle, string maxle, string queryfrac) {
+ void setVariables(string abske, string min, string mindi, string x, string d, string xa2, string chunk, string minchun, string idsmoothwindo, string minsmoothi, string max, string minle, string maxle, string queryfrac, string stra) {
abskew = abske;
minh = min;
mindiv = mindi;
+ strand = stra;
xn = x;
dn = d;
xa = xa2;
if (pDataArray->hasCount) {
error = cparser->getSeqs(pDataArray->groups[i], pDataArray->filename, true); if ((error == 1) || pDataArray->m->control_pressed) { delete cparser; return 0; }
}else {
- error = cparser->getSeqs(pDataArray->groups[i], pDataArray->filename, true); if ((error == 1) || pDataArray->m->control_pressed) { delete parser; return 0; }
+ error = parser->getSeqs(pDataArray->groups[i], pDataArray->filename, true); if ((error == 1) || pDataArray->m->control_pressed) { delete parser; return 0; }
}
//int numSeqs = driver((outputFName + groups[i]), filename, (accnos+ groups[i]), (alns+ groups[i]), numChimeras);
cPara.push_back(tempa);
}
+ if (pDataArray->strand != "") {
+ char* tempA = new char[9];
+ *tempA = '\0'; strncat(tempA, "--strand", 8);
+ cPara.push_back(tempA);
+ char* tempa = new char[pDataArray->strand.length()+1];
+ *tempa = '\0'; strncat(tempa, pDataArray->strand.c_str(), pDataArray->strand.length());
+ cPara.push_back(tempa);
+ }
+
if (pDataArray->useAbskew) {
char* tempskew = new char[9];
*tempskew = '\0'; strncat(tempskew, "--abskew", 8);
strncat(tempUchime, uchimeCommand.c_str(), uchimeCommand.length());
cPara.push_back(tempUchime);
- char* tempIn = new char[8];
- *tempIn = '\0'; strncat(tempIn, "--input", 7);
- //strcpy(tempIn, "--input");
- cPara.push_back(tempIn);
- char* temp = new char[filename.length()+1];
- *temp = '\0'; strncat(temp, filename.c_str(), filename.length());
- //strcpy(temp, filename.c_str());
- cPara.push_back(temp);
-
- //add reference file
+ string outputFileName = filename.substr(1, filename.length()-2) + ".uchime_formatted";
+ //prepFile(filename.substr(1, filename.length()-2), outputFileName);
+ //prepFile(filename, outputFileName);
+ /******************************************/
+ ifstream in23;
+ pDataArray->m->openInputFile((filename.substr(1, filename.length()-2)), in23);
+
+ ofstream out23;
+ pDataArray->m->openOutputFile(outputFileName, out23);
+
+ int fcount = 0;
+ while (!in23.eof()) {
+ if (pDataArray->m->control_pressed) { break; }
+
+ Sequence seq(in23); pDataArray->m->gobble(in23);
+
+ if (seq.getName() != "") { seq.printSequence(out23); fcount++; }
+ }
+ in23.close();
+ out23.close();
+ /******************************************/
+
+ filename = outputFileName;
+ filename = "\"" + filename + "\"";
+
+ //add reference file
char* tempRef = new char[5];
//strcpy(tempRef, "--db");
*tempRef = '\0'; strncat(tempRef, "--db", 4);
//strcpy(tempR, templatefile.c_str());
*tempR = '\0'; strncat(tempR, templatefile.c_str(), templatefile.length());
cPara.push_back(tempR);
+
+ char* tempIn = new char[8];
+ *tempIn = '\0'; strncat(tempIn, "--input", 7);
+ //strcpy(tempIn, "--input");
+ cPara.push_back(tempIn);
+ char* temp = new char[filename.length()+1];
+ *temp = '\0'; strncat(temp, filename.c_str(), filename.length());
+ //strcpy(temp, filename.c_str());
+ cPara.push_back(temp);
char* tempO = new char[12];
*tempO = '\0'; strncat(tempO, "--uchimeout", 11);
cPara.push_back(tempa);
}
+ if (pDataArray->strand != "") {
+ char* tempA = new char[9];
+ *tempA = '\0'; strncat(tempA, "--strand", 8);
+ cPara.push_back(tempA);
+ char* tempa = new char[pDataArray->strand.length()+1];
+ *tempa = '\0'; strncat(tempa, pDataArray->strand.c_str(), pDataArray->strand.length());
+ cPara.push_back(tempa);
+ }
+
if (pDataArray->useAbskew) {
char* tempskew = new char[9];
*tempskew = '\0'; strncat(tempskew, "--abskew", 8);
for (int j = 0; j < cPara.size(); j++) { uchimeParameters[j] = cPara[j]; commandString += toString(cPara[j]) + " "; }
//int numArgs = cPara.size();
+ commandString = "\"" + commandString + "\"";
+
//uchime_main(numArgs, uchimeParameters);
//cout << "commandString = " << commandString << endl;
if (pDataArray->m->debug) { pDataArray->m->mothurOut("[DEBUG]: uchime command = " + commandString + ".\n"); }
in.close();
out.close();
+ if (fcount != totalSeqs) { pDataArray->m->mothurOut("[ERROR]: process " + toString(pDataArray->threadID) + " only processed " + toString(pDataArray->count) + " of " + toString(pDataArray->end) + " sequences assigned to it, quitting. \n"); pDataArray->m->control_pressed = true; }
+
if (pDataArray->m->control_pressed) { return 0; }
pDataArray->m->mothurOutEndLine(); pDataArray->m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(totalSeqs) + " sequences."); pDataArray->m->mothurOutEndLine();
pDataArray->count = totalSeqs;
pDataArray->numChimeras = numChimeras;
+
return totalSeqs;
}
projects / mothur.git / blobdiff