A7FE7C401330EA1000F7B327 /* getcurrentcommand.cpp in Sources */ = {isa = PBXBuildFile; fileRef = A7FE7C3F1330EA1000F7B327 /* getcurrentcommand.cpp */; };
A7FE7E6D13311EA400F7B327 /* setcurrentcommand.cpp in Sources */ = {isa = PBXBuildFile; fileRef = A7FE7E6C13311EA400F7B327 /* setcurrentcommand.cpp */; };
A7FF19F2140FFDA500AD216D /* trimoligos.cpp in Sources */ = {isa = PBXBuildFile; fileRef = A7FF19F1140FFDA500AD216D /* trimoligos.cpp */; };
+ A7FFB558142CA02C004884F2 /* summarytaxcommand.cpp in Sources */ = {isa = PBXBuildFile; fileRef = A7FFB557142CA02C004884F2 /* summarytaxcommand.cpp */; };
/* End PBXBuildFile section */
/* Begin PBXCopyFilesBuildPhase section */
A7FE7E6C13311EA400F7B327 /* setcurrentcommand.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = setcurrentcommand.cpp; sourceTree = "<group>"; };
A7FF19F0140FFDA500AD216D /* trimoligos.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = trimoligos.h; sourceTree = "<group>"; };
A7FF19F1140FFDA500AD216D /* trimoligos.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = trimoligos.cpp; sourceTree = "<group>"; };
+ A7FFB556142CA02C004884F2 /* summarytaxcommand.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = summarytaxcommand.h; sourceTree = "<group>"; };
+ A7FFB557142CA02C004884F2 /* summarytaxcommand.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = summarytaxcommand.cpp; sourceTree = "<group>"; };
C6A0FF2C0290799A04C91782 /* mothur.1 */ = {isa = PBXFileReference; lastKnownFileType = text.man; path = mothur.1; sourceTree = "<group>"; };
/* End PBXFileReference section */
A7E9B7E412D37EC400DA6239 /* sffinfocommand.h */,
A7E9B7E312D37EC400DA6239 /* sffinfocommand.cpp */,
A7E9B7F312D37EC400DA6239 /* sharedcommand.h */,
+ A7E9B7F212D37EC400DA6239 /* sharedcommand.cpp */,
A7E9B82812D37EC400DA6239 /* shhhercommand.h */,
A7E9B82712D37EC400DA6239 /* shhhercommand.cpp */,
- A7E9B7F212D37EC400DA6239 /* sharedcommand.cpp */,
A7E9B84012D37EC400DA6239 /* splitabundcommand.h */,
A7E9B83F12D37EC400DA6239 /* splitabundcommand.cpp */,
A7E9B84212D37EC400DA6239 /* splitgroupscommand.h */,
A7E9B85712D37EC400DA6239 /* summarycommand.cpp */,
A7E9B85A12D37EC400DA6239 /* summarysharedcommand.h */,
A7E9B85912D37EC400DA6239 /* summarysharedcommand.cpp */,
+ A7FFB556142CA02C004884F2 /* summarytaxcommand.h */,
+ A7FFB557142CA02C004884F2 /* summarytaxcommand.cpp */,
A7E9B85C12D37EC400DA6239 /* systemcommand.h */,
A7E9B85B12D37EC400DA6239 /* systemcommand.cpp */,
A7E9B86312D37EC400DA6239 /* treegroupscommand.h */,
A795840D13F13CD900F201D5 /* countgroupscommand.cpp in Sources */,
A7FF19F2140FFDA500AD216D /* trimoligos.cpp in Sources */,
A7F9F5CF141A5E500032F693 /* sequenceparser.cpp in Sources */,
+ A7FFB558142CA02C004884F2 /* summarytaxcommand.cpp in Sources */,
);
runOnlyForDeploymentPostprocessing = 0;
};
}
/**************************************************************************************************/
-void PhyloSummary::summarize(string userTfile){
+int PhyloSummary::summarize(string userTfile){
try {
ifstream in;
//read in users taxonomy file and add sequences to tree
string name, tax;
+ int numSeqs = 0;
while(!in.eof()){
in >> name >> tax; m->gobble(in);
addSeqToTree(name, tax);
+ numSeqs++;
if (m->control_pressed) { break; }
}
in.close();
+
+ return numSeqs;
}
catch(exception& e) {
m->errorOut(e, "PhyloSummary", "summarize");
int level = 0;
+ //are there confidence scores, if so remove them
+ if (seqTaxonomy.find_first_of('(') != -1) { removeConfidences(seqTaxonomy); }
+
while (seqTaxonomy != "") {
if (m->control_pressed) { return 0; }
int level = 0;
+ //are there confidence scores, if so remove them
+ if (seqTaxonomy.find_first_of('(') != -1) { removeConfidences(seqTaxonomy); }
+
while (seqTaxonomy != "") {
if (m->control_pressed) { return 0; }
out << endl;
int totalChildrenInTree = 0;
+ map<string, int>::iterator itGroup;
map<string,int>::iterator it;
for(it=tree[0].children.begin();it!=tree[0].children.end();it++){
- if (tree[it->second].total != 0) { totalChildrenInTree++; }
+ if (tree[it->second].total != 0) {
+ totalChildrenInTree++;
+ tree[0].total += tree[it->second].total;
+
+ if (groupmap != NULL) {
+ vector<string> mGroups = groupmap->getNamesOfGroups();
+ for (int i = 0; i < mGroups.size(); i++) { tree[0].groupCount[mGroups[i]] += tree[it->second].groupCount[mGroups[i]]; }
+ }
+ }
}
//print root
out << tree[0].level << "\t" << tree[0].rank << "\t" << tree[0].name << "\t" << totalChildrenInTree << "\t" << tree[0].total << "\t";
- map<string, int>::iterator itGroup;
+
if (groupmap != NULL) {
//for (itGroup = tree[0].groupCount.begin(); itGroup != tree[0].groupCount.end(); itGroup++) {
// out << itGroup->second << '\t';
}
catch(exception& e) {
- m->errorOut(e, "PhyloSummary", "print");
+ m->errorOut(e, "PhyloSummary", "readTreeStruct");
+ exit(1);
+ }
+}
+/**************************************************************************************************/
+void PhyloSummary::removeConfidences(string& tax) {
+ try {
+
+ string taxon;
+ string newTax = "";
+
+ while (tax.find_first_of(';') != -1) {
+ //get taxon
+ taxon = tax.substr(0,tax.find_first_of(';'));
+
+ int pos = taxon.find_first_of('(');
+ if (pos != -1) {
+ taxon = taxon.substr(0, pos); //rip off confidence
+ }
+
+ taxon += ";";
+
+ tax = tax.substr(tax.find_first_of(';')+1, tax.length());
+ newTax += taxon;
+ }
+
+ tax = newTax;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "PhyloSummary", "removeConfidences");
exit(1);
}
}
-
/**************************************************************************************************/
--- /dev/null
+/*
+ * summarytaxcommand.cpp
+ * Mothur
+ *
+ * Created by westcott on 9/23/11.
+ * Copyright 2011 Schloss Lab. All rights reserved.
+ *
+ */
+
+#include "summarytaxcommand.h"
+#include "phylosummary.h"
+
+//**********************************************************************************************************************
+vector<string> SummaryTaxCommand::setParameters(){
+ try {
+ CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptaxonomy);
+ CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
+ CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
+ CommandParameter preftaxonomy("reftaxonomy", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(preftaxonomy);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+
+ vector<string> myArray;
+ for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummaryTaxCommand", "setParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+string SummaryTaxCommand::getHelpString(){
+ try {
+ string helpString = "";
+ helpString += "The summary.tax command reads a taxonomy file and an optional name file, and summarizes the taxonomy information.\n";
+ helpString += "The summary.tax command parameters are taxonomy, group and name. taxonomy is required, unless you have a valid current taxonomy file.\n";
+ helpString += "The name parameter allows you to enter a name file associated with your taxonomy file. \n";
+ helpString += "The group parameter allows you add a group file so you can have the summary totals broken up by group.\n";
+ helpString += "The reftaxonomy parameter allows you give the name of the reference taxonomy file used when you classified your sequences. It is not required, but providing it will keep the rankIDs in the summary file static.\n";
+ helpString += "The summary.tax command should be in the following format: \n";
+ helpString += "summary.tax(taxonomy=yourTaxonomyFile) \n";
+ helpString += "Note: No spaces between parameter labels (i.e. taxonomy), '=' and parameters (i.e.yourTaxonomyFile).\n";
+ return helpString;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummaryTaxCommand", "getHelpString");
+ exit(1);
+ }
+}
+
+//**********************************************************************************************************************
+SummaryTaxCommand::SummaryTaxCommand(){
+ try {
+ abort = true; calledHelp = true;
+ setParameters();
+ vector<string> tempOutNames;
+ outputTypes["summary"] = tempOutNames;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummaryTaxCommand", "SummaryTaxCommand");
+ exit(1);
+ }
+}
+//***************************************************************************************************************
+
+SummaryTaxCommand::SummaryTaxCommand(string option) {
+ try {
+ abort = false; calledHelp = false;
+
+ //allow user to run help
+ if(option == "help") { help(); abort = true; calledHelp = true; }
+ else if(option == "citation") { citation(); abort = true; calledHelp = true;}
+
+ else {
+ vector<string> myArray = setParameters();
+
+ OptionParser parser(option);
+ map<string,string> parameters = parser.getParameters();
+
+ ValidParameters validParameter;
+ map<string,string>::iterator it;
+
+ //check to make sure all parameters are valid for command
+ for (it = parameters.begin(); it != parameters.end(); it++) {
+ if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
+ }
+
+ //if the user changes the input directory command factory will send this info to us in the output parameter
+ string inputDir = validParameter.validFile(parameters, "inputdir", false);
+ if (inputDir == "not found"){ inputDir = ""; }
+ else {
+ string path;
+ it = parameters.find("taxonomy");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["taxonomy"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("name");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["name"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("group");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["group"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("reftaxonomy");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["reftaxonomy"] = inputDir + it->second; }
+ }
+
+ }
+
+ //initialize outputTypes
+ vector<string> tempOutNames;
+ outputTypes["summary"] = tempOutNames;
+
+ //check for required parameters
+ taxfile = validParameter.validFile(parameters, "taxonomy", true);
+ if (taxfile == "not open") { abort = true; }
+ else if (taxfile == "not found") {
+ taxfile = m->getTaxonomyFile();
+ if (taxfile != "") { m->mothurOut("Using " + taxfile + " as input file for the taxonomy parameter."); m->mothurOutEndLine(); }
+ else { m->mothurOut("You have no current taxonomy file and the taxonomy parameter is required."); m->mothurOutEndLine(); abort = true; }
+ }else { m->setTaxonomyFile(taxfile); }
+
+ namefile = validParameter.validFile(parameters, "name", true);
+ if (namefile == "not open") { namefile = ""; abort = true; }
+ else if (namefile == "not found") { namefile = ""; }
+ else { m->setNameFile(namefile); }
+
+ groupfile = validParameter.validFile(parameters, "group", true);
+ if (groupfile == "not open") { groupfile = ""; abort = true; }
+ else if (groupfile == "not found") { groupfile = ""; }
+ else { m->setGroupFile(groupfile); }
+
+ refTaxonomy = validParameter.validFile(parameters, "reftaxonomy", true);
+ if (refTaxonomy == "not found") { refTaxonomy = ""; m->mothurOut("reftaxonomy is not required, but if given will keep the rankIDs in the summary file static."); m->mothurOutEndLine(); }
+ else if (refTaxonomy == "not open") { refTaxonomy = ""; abort = true; }
+
+ //if the user changes the output directory command factory will send this info to us in the output parameter
+ outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
+ outputDir = "";
+ outputDir += m->hasPath(taxfile); //if user entered a file with a path then preserve it
+ }
+ }
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummaryTaxCommand", "SummaryTaxCommand");
+ exit(1);
+ }
+}
+//***************************************************************************************************************
+
+int SummaryTaxCommand::execute(){
+ try{
+
+ if (abort == true) { if (calledHelp) { return 0; } return 2; }
+ int start = time(NULL);
+
+ PhyloSummary* taxaSum;
+ if (refTaxonomy != "") {
+ taxaSum = new PhyloSummary(refTaxonomy, groupfile);
+ }else {
+ taxaSum = new PhyloSummary(groupfile);
+ }
+
+ if (m->control_pressed) { delete taxaSum; return 0; }
+
+ int numSeqs = 0;
+ if (namefile == "") { numSeqs = taxaSum->summarize(taxfile); }
+ else {
+ map<string, vector<string> > nameMap;
+ map<string, vector<string> >::iterator itNames;
+ m->readNames(namefile, nameMap);
+
+ if (m->control_pressed) { delete taxaSum; return 0; }
+
+ ifstream in;
+ m->openInputFile(taxfile, in);
+
+ //read in users taxonomy file and add sequences to tree
+ string name, taxon;
+
+ while(!in.eof()){
+ in >> name >> taxon; m->gobble(in);
+
+ itNames = nameMap.find(name);
+
+ if (itNames == nameMap.end()) {
+ m->mothurOut("[ERROR]: " + name + " is not in your name file please correct."); m->mothurOutEndLine(); exit(1);
+ }else{
+ for (int i = 0; i < itNames->second.size(); i++) {
+ numSeqs++;
+ taxaSum->addSeqToTree(itNames->second[i], taxon); //add it as many times as there are identical seqs
+ }
+ itNames->second.clear();
+ nameMap.erase(itNames->first);
+ }
+ }
+ in.close();
+ }
+
+ if (m->control_pressed) { delete taxaSum; return 0; }
+
+ //print summary file
+ ofstream outTaxTree;
+ string summaryFile = outputDir + m->getRootName(m->getSimpleName(taxfile)) + "tax.summary";
+ m->openOutputFile(summaryFile, outTaxTree);
+ taxaSum->print(outTaxTree);
+ outTaxTree.close();
+
+ delete taxaSum;
+
+ if (m->control_pressed) { m->mothurRemove(summaryFile); return 0; }
+
+ m->mothurOutEndLine();
+ m->mothurOut("It took " + toString(time(NULL) - start) + " secs to create the summary file for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
+ m->mothurOutEndLine();
+ m->mothurOut("Output File Name: "); m->mothurOutEndLine();
+ m->mothurOut(summaryFile); m->mothurOutEndLine(); outputNames.push_back(summaryFile); outputTypes["summary"].push_back(summaryFile);
+ m->mothurOutEndLine();
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummaryTaxCommand", "execute");
+ exit(1);
+ }
+}
+/**************************************************************************************/
+
+
projects / mothur.git / commitdiff