else {
//valid paramters for this command
- string AlignArray[] = {"fasta", "start", "end", "maxambig", "maxhomop", "minlength", "maxlength", "name", "group"};
+ string AlignArray[] = {"fasta", "start", "end", "maxambig", "maxhomop", "minlength", "maxlength",
+ "name", "group", "alignreport"};
vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
OptionParser parser(option);
//check for required parameters
fastafile = validParameter.validFile(parameters, "fasta", true);
- if (fastafile == "not found") { cout << "fasta is a required parameter for the screen.seqs command." << endl; abort = true; }
+ if (fastafile == "not found") { mothurOut("fasta is a required parameter for the screen.seqs command."); mothurOutEndLine(); abort = true; }
else if (fastafile == "not open") { abort = true; }
-
+
groupfile = validParameter.validFile(parameters, "group", true);
if (groupfile == "not open") { abort = true; }
else if (groupfile == "not found") { groupfile = ""; }
if (namefile == "not open") { abort = true; }
else if (namefile == "not found") { namefile = ""; }
-
+ alignreport = validParameter.validFile(parameters, "alignreport", true);
+ if (alignreport == "not open") { abort = true; }
+ else if (alignreport == "not found") { alignreport = ""; }
+
//check for optional parameter and set defaults
// ...at some point should added some additional type checking...
string temp;
- temp = validParameter.validFile(parameters, "start", false); if (temp == "not found") { temp = "-1"; }
+ temp = validParameter.validFile(parameters, "start", false); if (temp == "not found") { temp = "-1"; }
convert(temp, startPos);
- temp = validParameter.validFile(parameters, "end", false); if (temp == "not found") { temp = "-1"; }
+ temp = validParameter.validFile(parameters, "end", false); if (temp == "not found") { temp = "-1"; }
convert(temp, endPos);
temp = validParameter.validFile(parameters, "maxambig", false); if (temp == "not found") { temp = "-1"; }
temp = validParameter.validFile(parameters, "maxhomop", false); if (temp == "not found") { temp = "-1"; }
convert(temp, maxHomoP);
- temp = validParameter.validFile(parameters, "minlength", false); if (temp == "not found") { temp = "-1"; }
+ temp = validParameter.validFile(parameters, "minlength", false); if (temp == "not found") { temp = "-1"; }
convert(temp, minLength);
- temp = validParameter.validFile(parameters, "maxlength", false); if (temp == "not found") { temp = "-1"; }
+ temp = validParameter.validFile(parameters, "maxlength", false); if (temp == "not found") { temp = "-1"; }
convert(temp, maxLength);
}
}
catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the ScreenSeqsCommand class Function ScreenSeqsCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ errorOut(e, "ScreenSeqsCommand", "ScreenSeqsCommand");
exit(1);
}
- catch(...) {
- cout << "An unknown error has occurred in the ScreenSeqsCommand class function ScreenSeqsCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
}
//**********************************************************************************************************************
void ScreenSeqsCommand::help(){
try {
- cout << "The screen.seqs command reads a fastafile and creates ....." << "\n";
- cout << "The screen.seqs command parameters are fasta, start, end, maxambig, maxhomop, minlength, maxlength, name, and group." << "\n";
- cout << "The fasta parameter is required." << "\n";
- cout << "The start parameter .... The default is -1." << "\n";
- cout << "The end parameter .... The default is -1." << "\n";
- cout << "The maxambig parameter .... The default is -1." << "\n";
- cout << "The maxhomop parameter .... The default is -1." << "\n";
- cout << "The minlength parameter .... The default is -1." << "\n";
- cout << "The maxlength parameter .... The default is -1." << "\n";
- cout << "The name parameter allows you to provide a namesfile, and the group parameter allows you to provide a groupfile." << "\n";
- cout << "The screen.seqs command should be in the following format: " << "\n";
- cout << "screen.seqs(fasta=yourFastaFile, name=youNameFile, group=yourGroupFIle, start=yourStart, end=yourEnd, maxambig=yourMaxambig, " << "\n";
- cout << "maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength) " << "\n";
- cout << "Example screen.seqs(fasta=abrecovery.fasta, name=abrecovery.names, group=abrecovery.groups, start=..., end=..., maxambig=..., maxhomop=..., minlength=..., maxlength=...)." << "\n";
- cout << "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta)." << "\n" << "\n";
+ mothurOut("The screen.seqs command reads a fastafile and creates .....\n");
+ mothurOut("The screen.seqs command parameters are fasta, start, end, maxambig, maxhomop, minlength, maxlength, name, and group.\n");
+ mothurOut("The fasta parameter is required.\n");
+ mothurOut("The start parameter .... The default is -1.\n");
+ mothurOut("The end parameter .... The default is -1.\n");
+ mothurOut("The maxambig parameter .... The default is -1.\n");
+ mothurOut("The maxhomop parameter .... The default is -1.\n");
+ mothurOut("The minlength parameter .... The default is -1.\n");
+ mothurOut("The maxlength parameter .... The default is -1.\n");
+ mothurOut("The name parameter allows you to provide a namesfile, and the group parameter allows you to provide a groupfile.\n");
+ mothurOut("The screen.seqs command should be in the following format: \n");
+ mothurOut("screen.seqs(fasta=yourFastaFile, name=youNameFile, group=yourGroupFIle, start=yourStart, end=yourEnd, maxambig=yourMaxambig, \n");
+ mothurOut("maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength) \n");
+ mothurOut("Example screen.seqs(fasta=abrecovery.fasta, name=abrecovery.names, group=abrecovery.groups, start=..., end=..., maxambig=..., maxhomop=..., minlength=..., maxlength=...).\n");
+ mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
}
catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the ScreenSeqsCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ errorOut(e, "ScreenSeqsCommand", "help");
exit(1);
}
- catch(...) {
- cout << "An unknown error has occurred in the ScreenSeqsCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
}
//***************************************************************************************************************
while(!inFASTA.eof()){
Sequence currSeq(inFASTA);
- bool goodSeq = 1; // innocent until proven guilty
- if(goodSeq == 1 && startPos != -1 && startPos < currSeq.getStartPos()) { goodSeq = 0; }
- if(goodSeq == 1 && endPos != -1 && endPos > currSeq.getEndPos()) { goodSeq = 0; }
- if(goodSeq == 1 && maxAmbig != -1 && maxAmbig < currSeq.getAmbigBases()) { goodSeq = 0; }
- if(goodSeq == 1 && maxHomoP != -1 && maxHomoP < currSeq.getLongHomoPolymer()) { goodSeq = 0; }
- if(goodSeq == 1 && minLength != -1 && minLength > currSeq.getNumBases()) { goodSeq = 0; }
- if(goodSeq == 1 && maxLength != -1 && maxLength < currSeq.getNumBases()) { goodSeq = 0; }
-
- if(goodSeq == 1){
- currSeq.printSequence(goodSeqOut);
- }
- else{
- currSeq.printSequence(badSeqOut);
- badSeqNames.insert(currSeq.getName());
+ if (currSeq.getName() != "") {
+ bool goodSeq = 1; // innocent until proven guilty
+ if(goodSeq == 1 && startPos != -1 && startPos < currSeq.getStartPos()) { goodSeq = 0; }
+ if(goodSeq == 1 && endPos != -1 && endPos > currSeq.getEndPos()) { goodSeq = 0; }
+ if(goodSeq == 1 && maxAmbig != -1 && maxAmbig < currSeq.getAmbigBases()) { goodSeq = 0; }
+ if(goodSeq == 1 && maxHomoP != -1 && maxHomoP < currSeq.getLongHomoPolymer()) { goodSeq = 0; }
+ if(goodSeq == 1 && minLength != -1 && minLength > currSeq.getNumBases()) { goodSeq = 0; }
+ if(goodSeq == 1 && maxLength != -1 && maxLength < currSeq.getNumBases()) { goodSeq = 0; }
+
+ if(goodSeq == 1){
+ currSeq.printSequence(goodSeqOut);
+ }
+ else{
+ currSeq.printSequence(badSeqOut);
+ badSeqNames.insert(currSeq.getName());
+ }
}
gobble(inFASTA);
}
- if(namefile != ""){
- screenNameGroupFile(badSeqNames);
- }
- else if(groupfile != ""){
- screenGroupFile(badSeqNames);
- }
+ if(namefile != "" && groupfile != "") { screenNameGroupFile(badSeqNames); } // this screens both names and groups
+ else if(namefile != "") { screenNameGroupFile(badSeqNames); }
+ else if(groupfile != "") { screenGroupFile(badSeqNames); } // this screens just the groups
+ if(alignreport != "") { screenAlignReport(badSeqNames); }
+ goodSeqOut.close();
+ badSeqOut.close();
+ inFASTA.close();
return 0;
}
catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the ScreenSeqsCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ errorOut(e, "ScreenSeqsCommand", "execute");
exit(1);
}
- catch(...) {
- cout << "An unknown error has occurred in the ScreenSeqsCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
-
}
//***************************************************************************************************************
goodNameOut.close();
badNameOut.close();
+ //we were unable to remove some of the bad sequences
+ if (badSeqNames.size() != 0) {
+ for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
+ mothurOut("Your namefile does not include the sequence " + *it + " please correct.");
+ mothurOutEndLine();
+ }
+ }
+
if(groupfile != ""){
ifstream inputGroups;
inputGroups.close();
goodGroupOut.close();
badGroupOut.close();
+
+ //we were unable to remove some of the bad sequences
+ if (badSeqGroups.size() != 0) {
+ for (it = badSeqGroups.begin(); it != badSeqGroups.end(); it++) {
+ mothurOut("Your namefile does not include the sequence " + *it + " please correct.");
+ mothurOutEndLine();
+ }
+ }
}
}
}
gobble(inputGroups);
}
+
+ //we were unable to remove some of the bad sequences
+ if (badSeqNames.size() != 0) {
+ for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
+ mothurOut("Your groupfile does not include the sequence " + *it + " please correct.");
+ mothurOutEndLine();
+ }
+ }
+
inputGroups.close();
goodGroupOut.close();
badGroupOut.close();
//***************************************************************************************************************
+void ScreenSeqsCommand::screenAlignReport(set<string> badSeqNames){
+
+ ifstream inputAlignReport;
+ openInputFile(alignreport, inputAlignReport);
+ string seqName, group;
+ set<string>::iterator it;
+
+ string goodAlignReportFile = getRootName(alignreport) + "good" + getExtension(alignreport);
+ string badAlignReportFile = getRootName(alignreport) + "bad" + getExtension(alignreport);
+
+ ofstream goodAlignReportOut; openOutputFile(goodAlignReportFile, goodAlignReportOut);
+ ofstream badAlignReportOut; openOutputFile(badAlignReportFile, badAlignReportOut);
+
+ while (!inputAlignReport.eof()) { // need to copy header
+ char c = inputAlignReport.get();
+ goodAlignReportOut << c;
+ badAlignReportOut << c;
+ if (c == 10 || c == 13){ break; }
+ }
+
+ while(!inputAlignReport.eof()){
+ inputAlignReport >> seqName;
+ it = badSeqNames.find(seqName);
+ string line;
+ while (!inputAlignReport.eof()) { // need to copy header
+ char c = inputAlignReport.get();
+ line += c;
+ if (c == 10 || c == 13){ break; }
+ }
+
+ if(it != badSeqNames.end()){
+ badSeqNames.erase(it);
+ badAlignReportOut << seqName << '\t' << line;;
+ }
+ else{
+ goodAlignReportOut << seqName << '\t' << line;
+ }
+ gobble(inputAlignReport);
+ }
+
+ //we were unable to remove some of the bad sequences
+ if (badSeqNames.size() != 0) {
+ for (it = badSeqNames.begin(); it != badSeqNames.end(); it++) {
+ mothurOut("Your file does not include the sequence " + *it + " please correct.");
+ mothurOutEndLine();
+ }
+ }
+
+ inputAlignReport.close();
+ goodAlignReportOut.close();
+ badAlignReportOut.close();
+
+}
+
+//***************************************************************************************************************
+
projects / mothur.git / blobdiff