*/
#include "getsharedotucommand.h"
-
+#include "sharedutilities.h"
//**********************************************************************************************************************
-vector<string> GetSharedOTUCommand::getValidParameters(){
+vector<string> GetSharedOTUCommand::setParameters(){
try {
- string Array[] = {"label","unique","shared","fasta","list","group","output","outputdir","inputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pfasta);
+ CommandParameter pgroup("group", "InputTypes", "", "", "none", "FNGLT", "none",false,true); parameters.push_back(pgroup);
+ CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,true); parameters.push_back(plist);
+ CommandParameter poutput("output", "Multiple", "accnos-default", "default", "", "", "",false,false); parameters.push_back(poutput);
+ CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
+ CommandParameter punique("unique", "String", "", "", "", "", "",false,false); parameters.push_back(punique);
+ CommandParameter pshared("shared", "String", "", "", "", "", "",false,false); parameters.push_back(pshared);
+ 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, "GetSharedOTUCommand", "getValidParameters");
+ m->errorOut(e, "GetSharedOTUCommand", "setParameters");
exit(1);
}
}
//**********************************************************************************************************************
-GetSharedOTUCommand::GetSharedOTUCommand(){
+string GetSharedOTUCommand::getHelpString(){
try {
- //initialize outputTypes
- vector<string> tempOutNames;
- outputTypes["fasta"] = tempOutNames;
- outputTypes["accnos"] = tempOutNames;
- outputTypes["sharedseqs"] = tempOutNames;
+ string helpString = "";
+ helpString += "The get.sharedseqs command parameters are list, group, label, unique, shared, output and fasta. The list and group parameters are required, unless you have valid current files.\n";
+ helpString += "The label parameter allows you to select what distance levels you would like output files for, and are separated by dashes.\n";
+ helpString += "The unique and shared parameters allow you to select groups you would like to know the shared info for, and are separated by dashes.\n";
+ helpString += "If you enter your groups under the unique parameter mothur will return the otus that contain ONLY sequences from those groups.\n";
+ helpString += "If you enter your groups under the shared parameter mothur will return the otus that contain sequences from those groups and may also contain sequences from other groups.\n";
+ helpString += "If you do not enter any groups then the get.sharedseqs command will return sequences that are unique to all groups in your group file.\n";
+ helpString += "The fasta parameter allows you to input a fasta file and outputs a fasta file for each distance level containing only the sequences that are in OTUs shared by the groups specified.\n";
+ helpString += "The output parameter allows you to output the list of names without the group and bin number added. \n";
+ helpString += "With this option you can use the names file as an input in get.seqs and remove.seqs commands. To do this enter output=accnos. \n";
+ helpString += "The get.sharedseqs command outputs a .names file for each distance level containing a list of sequences in the OTUs shared by the groups specified.\n";
+ helpString += "The get.sharedseqs command should be in the following format: get.sharedseqs(list=yourListFile, group=yourGroupFile, label=yourLabels, unique=yourGroups, fasta=yourFastafile, output=yourOutput).\n";
+ helpString += "Example get.sharedseqs(list=amazon.fn.list, label=unique-0.01, group= amazon.groups, unique=forest-pasture, fasta=amazon.fasta, output=accnos).\n";
+ helpString += "The output to the screen is the distance and the number of otus at that distance for the groups you specified.\n";
+ helpString += "The default value for label is all labels in your inputfile. The default for groups is all groups in your file.\n";
+ helpString += "Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabel).\n";
+ return helpString;
}
catch(exception& e) {
- m->errorOut(e, "GetSharedOTUCommand", "GetSharedOTUCommand");
+ m->errorOut(e, "GetSharedOTUCommand", "getHelpString");
exit(1);
}
}
//**********************************************************************************************************************
-vector<string> GetSharedOTUCommand::getRequiredParameters(){
+string GetSharedOTUCommand::getOutputFileNameTag(string type, string inputName=""){
try {
- string Array[] = {"list","group"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
- return myArray;
+ string outputFileName = "";
+ map<string, vector<string> >::iterator it;
+
+ //is this a type this command creates
+ it = outputTypes.find(type);
+ if (it == outputTypes.end()) { m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
+ else {
+ if (type == "fasta") { outputFileName = "shared.fasta"; }
+ else if (type == "accnos") { outputFileName = "accnos"; }
+ else if (type == "sharedseqs") { outputFileName = "shared.seqs"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true; }
+ }
+ return outputFileName;
}
catch(exception& e) {
- m->errorOut(e, "GetSharedOTUCommand", "getRequiredParameters");
+ m->errorOut(e, "GetSharedOTUCommand", "getOutputFileNameTag");
exit(1);
}
}
+
//**********************************************************************************************************************
-vector<string> GetSharedOTUCommand::getRequiredFiles(){
+GetSharedOTUCommand::GetSharedOTUCommand(){
try {
- vector<string> myArray;
- return myArray;
+ abort = true; calledHelp = true;
+ setParameters();
+ vector<string> tempOutNames;
+ outputTypes["fasta"] = tempOutNames;
+ outputTypes["accnos"] = tempOutNames;
+ outputTypes["sharedseqs"] = tempOutNames;
}
catch(exception& e) {
- m->errorOut(e, "GetSharedOTUCommand", "getRequiredFiles");
+ m->errorOut(e, "GetSharedOTUCommand", "GetSharedOTUCommand");
exit(1);
}
}
GetSharedOTUCommand::GetSharedOTUCommand(string option) {
try {
- globaldata = GlobalData::getInstance();
- abort = false;
+ abort = false; calledHelp = false;
unique = true;
allLines = 1;
- labels.clear();
//allow user to run help
- if(option == "help") { help(); abort = true; }
+ if(option == "help") { help(); abort = true; calledHelp = true; }
+ else if(option == "citation") { citation(); abort = true; calledHelp = true;}
else {
- //valid paramters for this command
- string Array[] = {"label","unique","shared","fasta","list","group","output","outputdir","inputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ vector<string> myArray = setParameters();
OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
//check for required parameters
listfile = validParameter.validFile(parameters, "list", true);
if (listfile == "not open") { abort = true; }
- else if (listfile == "not found") { listfile = ""; }
- else { globaldata->setListFile(listfile); globaldata->setFormat("list"); }
+ else if (listfile == "not found") {
+ listfile = m->getListFile();
+ if (listfile != "") { format = "list"; m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
+ else {
+ m->mothurOut("No valid current list file. You must provide a list file."); m->mothurOutEndLine();
+ abort = true;
+ }
+ }else { format = "list"; m->setListFile(listfile); }
groupfile = validParameter.validFile(parameters, "group", true);
if (groupfile == "not open") { abort = true; }
- else if (groupfile == "not found") { groupfile = ""; }
+ else if (groupfile == "not found") {
+ groupfile = m->getGroupFile();
+ if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
+ else {
+ m->mothurOut("No valid current group file. You must provide a group file."); m->mothurOutEndLine();
+ abort = true;
+ }
+ }else { m->setGroupFile(groupfile); }
if ((listfile == "") || (groupfile == "")) { m->mothurOut("The list and group parameters are required."); m->mothurOutEndLine(); abort = true; }
output = validParameter.validFile(parameters, "output", false);
if (output == "not found") { output = ""; }
+ else if (output == "default") { output = ""; }
groups = validParameter.validFile(parameters, "unique", false);
if (groups == "not found") { groups = ""; }
else {
userGroups = "unique." + groups;
m->splitAtDash(groups, Groups);
- globaldata->Groups = Groups;
+ m->setGroups(Groups);
}
else {
userGroups = groups;
m->splitAtDash(groups, Groups);
- globaldata->Groups = Groups;
+ m->setGroups(Groups);
unique = false;
}
fastafile = validParameter.validFile(parameters, "fasta", true);
if (fastafile == "not open") { abort = true; }
else if (fastafile == "not found") { fastafile = ""; }
-
+ else { m->setFastaFile(fastafile); }
}
}
}
//**********************************************************************************************************************
-void GetSharedOTUCommand::help(){
- try {
- m->mothurOut("The get.sharedseqs command parameters are list, group, label, unique, shared, output and fasta. The list and group parameters are required.\n");
- m->mothurOut("The label parameter allows you to select what distance levels you would like output files for, and are separated by dashes.\n");
- m->mothurOut("The unique and shared parameters allow you to select groups you would like to know the shared info for, and are separated by dashes.\n");
- m->mothurOut("If you enter your groups under the unique parameter mothur will return the otus that contain ONLY sequences from those groups.\n");
- m->mothurOut("If you enter your groups under the shared parameter mothur will return the otus that contain sequences from those groups and may also contain sequences from other groups.\n");
- m->mothurOut("If you do not enter any groups then the get.sharedseqs command will return sequences that are unique to all groups in your group file.\n");
- m->mothurOut("The fasta parameter allows you to input a fasta file and outputs a fasta file for each distance level containing only the sequences that are in OTUs shared by the groups specified.\n");
- m->mothurOut("The output parameter allows you to output the list of names without the group and bin number added. \n");
- m->mothurOut("With this option you can use the names file as an input in get.seqs and remove.seqs commands. To do this enter output=accnos. \n");
- m->mothurOut("The get.sharedseqs command outputs a .names file for each distance level containing a list of sequences in the OTUs shared by the groups specified.\n");
- m->mothurOut("The get.sharedseqs command should be in the following format: get.sabund(label=yourLabels, groups=yourGroups, fasta=yourFastafile, output=yourOutput).\n");
- m->mothurOut("Example get.sharedseqs(list=amazon.fn.list, label=unique-0.01, group=forest-pasture, fasta=amazon.fasta, output=accnos).\n");
- m->mothurOut("The output to the screen is the distance and the number of otus at that distance for the groups you specified.\n");
- m->mothurOut("The default value for label is all labels in your inputfile. The default for groups is all groups in your file.\n");
- m->mothurOut("Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabel).\n\n");
- }
- catch(exception& e) {
- m->errorOut(e, "GetSharedOTUCommand", "help");
- exit(1);
- }
-}
-
-//**********************************************************************************************************************
-
-GetSharedOTUCommand::~GetSharedOTUCommand(){}
-
-//**********************************************************************************************************************
-
int GetSharedOTUCommand::execute(){
try {
- if (abort == true) { return 0; }
+ if (abort == true) { if (calledHelp) { return 0; } return 2; }
groupMap = new GroupMap(groupfile);
int error = groupMap->readMap();
if (m->control_pressed) { delete groupMap; return 0; }
- globaldata->gGroupmap = groupMap;
-
if (Groups.size() == 0) {
- Groups = groupMap->namesOfGroups;
+ Groups = groupMap->getNamesOfGroups();
//make string for outputfile name
userGroups = "unique.";
for(int i = 0; i < Groups.size(); i++) { userGroups += Groups[i] + "-"; }
userGroups = userGroups.substr(0, userGroups.length()-1);
+ }else{
+ //sanity check for group names
+ SharedUtil util;
+ vector<string> namesOfGroups = groupMap->getNamesOfGroups();
+ util.setGroups(Groups, namesOfGroups);
+ groupMap->setNamesOfGroups(namesOfGroups);
}
//put groups in map to find easier
if (m->control_pressed) {
if (lastlist != NULL) { delete lastlist; }
- for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } outputTypes.clear();
+ for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } outputTypes.clear();
delete groupMap; return 0;
}
//reset groups parameter
- globaldata->Groups.clear();
+ m->clearGroups();
if (lastlist != NULL) { delete lastlist; }
- if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } delete groupMap; return 0; }
+ if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete groupMap; return 0; }
+
+ //set fasta file as new current fastafile
+ string current = "";
+ itTypes = outputTypes.find("fasta");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
+ }
+
+ if (output == "accnos") {
+ itTypes = outputTypes.find("accnos");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
+ }
+ }
m->mothurOutEndLine();
m->mothurOut("Output File Names: "); m->mothurOutEndLine();
if (outputDir == "") { outputDir += m->hasPath(listfile); }
if (output != "accnos") {
- outputFileNames = outputDir + m->getRootName(m->getSimpleName(listfile)) + shared->getLabel() + userGroups + ".shared.seqs";
+ outputFileNames = outputDir + m->getRootName(m->getSimpleName(listfile)) + shared->getLabel() + userGroups + "." + getOutputFileNameTag("sharedseqs");
}else {
- outputFileNames = outputDir + m->getRootName(m->getSimpleName(listfile)) + shared->getLabel() + userGroups + ".accnos";
+ outputFileNames = outputDir + m->getRootName(m->getSimpleName(listfile)) + shared->getLabel() + userGroups + "." + getOutputFileNameTag("accnos");
}
m->openOutputFile(outputFileNames, outNames);
//go through each bin, find out if shared
for (int i = 0; i < shared->getNumBins(); i++) {
- if (m->control_pressed) { outNames.close(); remove(outputFileNames.c_str()); return 0; }
+ if (m->control_pressed) { outNames.close(); m->mothurRemove(outputFileNames); return 0; }
bool uniqueOTU = true;
vector<string> namesOfSeqsInThisBin;
- string names = shared->get(i);
- while ((names.find_first_of(',') != -1)) {
- string name = names.substr(0,names.find_first_of(','));
- names = names.substr(names.find_first_of(',')+1, names.length());
+ string names = shared->get(i);
+ vector<string> binNames;
+ m->splitAtComma(names, binNames);
+ for(int j = 0; j < binNames.size(); j++) {
+ string name = binNames[j];
//find group
string seqGroup = groupMap->getGroup(name);
else { atLeastOne[seqGroup]++; }
}
- //get last name
- string seqGroup = groupMap->getGroup(names);
- if (output != "accnos") {
- namesOfSeqsInThisBin.push_back((names + "|" + seqGroup + "|" + toString(i+1)));
- }else { namesOfSeqsInThisBin.push_back(names); }
-
- if (seqGroup == "not found") { m->mothurOut(names + " is not in your groupfile. Please correct."); m->mothurOutEndLine(); exit(1); }
-
- //is this seq in one of hte groups we care about
- it = groupFinder.find(seqGroup);
- if (it == groupFinder.end()) { uniqueOTU = false; } //you have a sequence from a group you don't want
- else { atLeastOne[seqGroup]++; }
-
-
//make sure you have at least one seq from each group you want
bool sharedByAll = true;
map<string, int>::iterator it2;
outNames.close();
if (!wroteSomething) {
- remove(outputFileNames.c_str());
+ m->mothurRemove(outputFileNames);
string outputString = "\t" + toString(num) + " - No otus shared by groups";
string groupString = "";
//if fasta file provided output new fasta file
if ((fastafile != "") && wroteSomething) {
if (outputDir == "") { outputDir += m->hasPath(fastafile); }
- string outputFileFasta = outputDir + m->getRootName(m->getSimpleName(fastafile)) + shared->getLabel() + userGroups + ".shared.fasta";
+ string outputFileFasta = outputDir + m->getRootName(m->getSimpleName(fastafile)) + shared->getLabel() + userGroups + "." + getOutputFileNameTag("fasta");
ofstream outFasta;
m->openOutputFile(outputFileFasta, outFasta);
outputNames.push_back(outputFileFasta); outputTypes["fasta"].push_back(outputFileFasta);
projects / mothur.git / blobdiff