#include "sequence.hpp"
#include "listvector.hpp"
#include "sharedutilities.h"
+#include "inputdata.h"
//**********************************************************************************************************************
-vector<string> GetGroupsCommand::getValidParameters(){
+vector<string> GetGroupsCommand::setParameters(){
try {
- string Array[] = {"fasta","name", "group", "accnos", "groups","list","taxonomy","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 pshared("shared", "InputTypes", "", "", "none", "FNGLT-sharedGroup", "none",false,false); parameters.push_back(pshared);
+ CommandParameter pname("name", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pname);
+ CommandParameter pgroup("group", "InputTypes", "", "", "none", "FNGLT-sharedGroup", "none",false,false); parameters.push_back(pgroup);
+ CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(plist);
+ CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(ptaxonomy);
+ CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(paccnos);
+ CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
+ 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, "GetGroupsCommand", "getValidParameters");
+ m->errorOut(e, "GetGroupsCommand", "setParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+string GetGroupsCommand::getHelpString(){
+ try {
+ string helpString = "";
+ helpString += "The get.groups command selects sequences from a specfic group or set of groups from the following file types: fasta, name, group, list, taxonomy or shared file.\n";
+ helpString += "It outputs a file containing the sequences in the those specified groups, or a sharedfile containing only those groups.\n";
+ helpString += "The get.groups command parameters are accnos, fasta, name, group, list, taxonomy, shared and groups. The group parameter is required, unless you have a current group file, or are using a shared file.\n";
+ helpString += "You must also provide an accnos containing the list of groups to get or set the groups parameter to the groups you wish to select.\n";
+ helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like. You can separate group names with dashes.\n";
+ helpString += "The get.groups command should be in the following format: get.groups(accnos=yourAccnos, fasta=yourFasta, group=yourGroupFile).\n";
+ helpString += "Example get.groups(accnos=amazon.accnos, fasta=amazon.fasta, group=amazon.groups).\n";
+ helpString += "or get.groups(groups=pasture, fasta=amazon.fasta, group=amazon.groups).\n";
+ helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
+ return helpString;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "getHelpString");
exit(1);
}
}
//**********************************************************************************************************************
GetGroupsCommand::GetGroupsCommand(){
try {
- abort = true; calledHelp = true;
+ abort = true; calledHelp = true;
+ setParameters();
vector<string> tempOutNames;
outputTypes["fasta"] = tempOutNames;
outputTypes["taxonomy"] = tempOutNames;
outputTypes["name"] = tempOutNames;
outputTypes["group"] = tempOutNames;
outputTypes["list"] = tempOutNames;
+ outputTypes["shared"] = tempOutNames;
}
catch(exception& e) {
m->errorOut(e, "GetGroupsCommand", "GetGroupsCommand");
}
}
//**********************************************************************************************************************
-vector<string> GetGroupsCommand::getRequiredParameters(){
- try {
- string Array[] = {"group"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
- return myArray;
- }
- catch(exception& e) {
- m->errorOut(e, "GetGroupsCommand", "getRequiredParameters");
- exit(1);
- }
-}
-//**********************************************************************************************************************
-vector<string> GetGroupsCommand::getRequiredFiles(){
- try {
- vector<string> myArray;
- return myArray;
- }
- catch(exception& e) {
- m->errorOut(e, "GetGroupsCommand", "getRequiredFiles");
- exit(1);
- }
-}
-//**********************************************************************************************************************
GetGroupsCommand::GetGroupsCommand(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 {
- //valid paramters for this command
- string Array[] = {"fasta","name", "group", "accnos", "groups", "list","taxonomy","outputdir","inputdir" };
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ vector<string> myArray = setParameters();
OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
outputTypes["name"] = tempOutNames;
outputTypes["group"] = tempOutNames;
outputTypes["list"] = tempOutNames;
+ outputTypes["shared"] = tempOutNames;
//if the user changes the output directory command factory will send this info to us in the output parameter
//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("shared");
+ //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["shared"] = inputDir + it->second; }
+ }
}
//check for required parameters
accnosfile = validParameter.validFile(parameters, "accnos", true);
if (accnosfile == "not open") { abort = true; }
- else if (accnosfile == "not found") { accnosfile = ""; }
+ else if (accnosfile == "not found") { accnosfile = ""; }
+ else { m->setAccnosFile(accnosfile); }
fastafile = validParameter.validFile(parameters, "fasta", true);
if (fastafile == "not open") { abort = true; }
- else if (fastafile == "not found") { fastafile = ""; }
+ else if (fastafile == "not found") { fastafile = ""; }
+ else { m->setFastaFile(fastafile); }
namefile = validParameter.validFile(parameters, "name", true);
if (namefile == "not open") { abort = true; }
else if (namefile == "not found") { namefile = ""; }
+ else { m->setNameFile(namefile); }
groupfile = validParameter.validFile(parameters, "group", true);
if (groupfile == "not open") { abort = true; }
- else if (groupfile == "not found") { groupfile = ""; m->mothurOut("You must provide a group file."); m->mothurOutEndLine(); abort = true; }
+ else if (groupfile == "not found") {
+ //if there is a current group file, use it
+ groupfile = m->getGroupFile();
+ if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
+ else { m->mothurOut("You have no current groupfile and the group parameter is required."); m->mothurOutEndLine(); abort = true; }
+ }else { m->setGroupFile(groupfile); }
listfile = validParameter.validFile(parameters, "list", true);
if (listfile == "not open") { abort = true; }
else if (listfile == "not found") { listfile = ""; }
+ else { m->setListFile(listfile); }
taxfile = validParameter.validFile(parameters, "taxonomy", true);
if (taxfile == "not open") { abort = true; }
else if (taxfile == "not found") { taxfile = ""; }
+ else { m->setTaxonomyFile(taxfile); }
groups = validParameter.validFile(parameters, "groups", false);
if (groups == "not found") { groups = ""; }
- else {
+ else {
m->splitAtDash(groups, Groups);
+ m->setGroups(Groups);
+ }
+
+ sharedfile = validParameter.validFile(parameters, "shared", true);
+ if (sharedfile == "not open") { sharedfile = ""; abort = true; }
+ else if (sharedfile == "not found") { sharedfile = ""; }
+ else { m->setSharedFile(sharedfile); }
+
+ groupfile = validParameter.validFile(parameters, "group", true);
+ if (groupfile == "not open") { groupfile = ""; abort = true; }
+ else if (groupfile == "not found") { groupfile = ""; }
+ else { m->setGroupFile(groupfile); }
+
+ if ((sharedfile == "") && (groupfile == "")) {
+ //is there are current file available for any of these?
+ if ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")) {
+ //give priority to group, then shared
+ groupfile = m->getGroupFile();
+ if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
+ else {
+ sharedfile = m->getSharedFile();
+ if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
+ else {
+ m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
+ }
+ }
+ }else {
+ //give priority to shared, then group
+ sharedfile = m->getSharedFile();
+ if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
+ else {
+ groupfile = m->getGroupFile();
+ if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
+ else {
+ m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
+ }
+ }
+ }
}
if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
- if ((fastafile == "") && (namefile == "") && (groupfile == "") && (listfile == "") && (taxfile == "")) { m->mothurOut("You must provide at least one of the following: fasta, name, taxonomy, group or list."); m->mothurOutEndLine(); abort = true; }
+ if ((fastafile == "") && (namefile == "") && (groupfile == "") && (sharedfile == "") && (listfile == "") && (taxfile == "")) { m->mothurOut("You must provide at least one of the following: fasta, name, taxonomy, group, shared or list."); m->mothurOutEndLine(); abort = true; }
+ if ((groupfile == "") && ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != ""))) { m->mothurOut("If using a fasta, name, taxonomy, group or list, then you must provide a group file."); m->mothurOutEndLine(); abort = true; }
+
}
}
}
//**********************************************************************************************************************
-void GetGroupsCommand::help(){
- try {
- m->mothurOut("The get.groups command selects sequences from a specfic group or set of groups from the following file types: fasta, name, group, list, taxonomy.\n");
- m->mothurOut("It outputs a file containing the sequences in the those specified groups.\n");
- m->mothurOut("The get.groups command parameters are accnos, fasta, name, group, list, taxonomy and groups. The group parameter is required.\n");
- m->mothurOut("You must also provide an accnos containing the list of groups to get or set the groups parameter to the groups you wish to select.\n");
- m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like. You can separate group names with dashes.\n");
- m->mothurOut("The get.groups command should be in the following format: get.groups(accnos=yourAccnos, fasta=yourFasta, group=yourGroupFile).\n");
- m->mothurOut("Example get.groups(accnos=amazon.accnos, fasta=amazon.fasta, group=amazon.groups).\n");
- m->mothurOut("or get.groups(groups=pasture, fasta=amazon.fasta, group=amazon.groups).\n");
- m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
- }
- catch(exception& e) {
- m->errorOut(e, "GetGroupsCommand", "help");
- exit(1);
- }
-}
-
-//**********************************************************************************************************************
-
int GetGroupsCommand::execute(){
try {
if (abort == true) { if (calledHelp) { return 0; } return 2; }
- groupMap = new GroupMap(groupfile);
- groupMap->readMap();
-
//get groups you want to remove
if (accnosfile != "") { readAccnos(); }
- //make sure groups are valid
- //takes care of user setting groupNames that are invalid or setting groups=all
- SharedUtil* util = new SharedUtil();
- util->setGroups(Groups, groupMap->namesOfGroups);
- delete util;
-
- //fill names with names of sequences that are from the groups we want to remove
- fillNames();
+ if (groupfile != "") {
+ groupMap = new GroupMap(groupfile);
+ groupMap->readMap();
+
+ //make sure groups are valid
+ //takes care of user setting groupNames that are invalid or setting groups=all
+ SharedUtil* util = new SharedUtil();
+ vector<string> gNamesOfGroups = groupMap->getNamesOfGroups();
+ util->setGroups(Groups, gNamesOfGroups);
+ groupMap->setNamesOfGroups(gNamesOfGroups);
+ delete util;
+
+ //fill names with names of sequences that are from the groups we want to remove
+ fillNames();
+
+ delete groupMap;
+ }
- if (m->control_pressed) { delete groupMap; return 0; }
+ if (m->control_pressed) { return 0; }
//read through the correct file and output lines you want to keep
if (namefile != "") { readName(); }
if (groupfile != "") { readGroup(); }
if (listfile != "") { readList(); }
if (taxfile != "") { readTax(); }
+ if (sharedfile != "") { readShared(); }
- if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
+ if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
- m->mothurOut("Selected " + toString(names.size()) + " sequences. From the groups: "); m->mothurOutEndLine();
- for (int i = 0; i < Groups.size(); i++) { m->mothurOut(Groups[i]); m->mothurOutEndLine(); }
- m->mothurOutEndLine();
if (outputNames.size() != 0) {
m->mothurOutEndLine();
if (itTypes != outputTypes.end()) {
if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
}
+
+ itTypes = outputTypes.find("shared");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
+ }
}
return 0;
string name;
bool wroteSomething = false;
+ int selectedCount = 0;
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
Sequence currSeq(in);
name = currSeq.getName();
wroteSomething = true;
currSeq.printSequence(out);
+ selectedCount++;
}
}
m->gobble(in);
if (wroteSomething == false) { m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine(); }
outputTypes["fasta"].push_back(outputFileName); outputNames.push_back(outputFileName);
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
+
+
return 0;
}
exit(1);
}
}
-
+//**********************************************************************************************************************
+int GetGroupsCommand::readShared(){
+ try {
+ string thisOutputDir = outputDir;
+ if (outputDir == "") { thisOutputDir += m->hasPath(sharedfile); }
+
+ InputData input(sharedfile, "sharedfile");
+ vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
+
+ bool wroteSomething = false;
+
+ while(lookup[0] != NULL) {
+
+ string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + lookup[0]->getLabel() + ".pick" + m->getExtension(sharedfile);
+ ofstream out;
+ m->openOutputFile(outputFileName, out);
+ outputTypes["shared"].push_back(outputFileName); outputNames.push_back(outputFileName);
+
+ if (m->control_pressed) { out.close(); m->mothurRemove(outputFileName); for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; }
+
+ lookup[0]->printHeaders(out);
+
+ for (int i = 0; i < lookup.size(); i++) {
+ out << lookup[i]->getLabel() << '\t' << lookup[i]->getGroup() << '\t';
+ lookup[i]->print(out);
+ wroteSomething = true;
+
+ }
+
+ //get next line to process
+ //prevent memory leak
+ for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
+ lookup = input.getSharedRAbundVectors();
+
+ out.close();
+ }
+
+ if (wroteSomething == false) { m->mothurOut("Your file contains only the groups you wish to remove."); m->mothurOutEndLine(); }
+
+ string groupsString = "";
+ for (int i = 0; i < Groups.size()-1; i++) { groupsString += Groups[i] + ", "; }
+ groupsString += Groups[Groups.size()-1];
+
+ m->mothurOut("Selected groups: " + groupsString + " from your shared file."); m->mothurOutEndLine();
+
+ return 0;
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "GetGroupsCommand", "readShared");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
int GetGroupsCommand::readList(){
try {
m->openInputFile(listfile, in);
bool wroteSomething = false;
+ int selectedCount = 0;
while(!in.eof()){
+
+ selectedCount = 0;
+
//read in list vector
ListVector list(in);
//for each bin
for (int i = 0; i < list.getNumBins(); i++) {
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
//parse out names that are in accnos file
string binnames = list.get(i);
binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
//if that name is in the .accnos file, add it
- if (names.count(name) != 0) { newNames += name + ","; }
+ if (names.count(name) != 0) { newNames += name + ","; selectedCount++; }
}
//get last name
- if (names.count(binnames) != 0) { newNames += binnames + ","; }
+ if (names.count(binnames) != 0) { newNames += binnames + ","; selectedCount++; }
//if there are names in this bin add to new list
if (newNames != "") {
if (wroteSomething == false) { m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine(); }
outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
+
return 0;
}
string name, firstCol, secondCol;
bool wroteSomething = false;
+ int selectedCount = 0;
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
in >> firstCol; m->gobble(in);
in >> secondCol;
}
}
+ selectedCount += validSecond.size();
+
//if the name in the first column is in the set then print it and any other names in second column also in set
if (names.count(firstCol) != 0) {
if (wroteSomething == false) { m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine(); }
outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your name file."); m->mothurOutEndLine();
+
return 0;
}
catch(exception& e) {
string name, group;
bool wroteSomething = false;
+ int selectedCount = 0;
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
in >> name; //read from first column
in >> group; //read from second column
if (names.count(name) != 0) {
wroteSomething = true;
out << name << '\t' << group << endl;
+ selectedCount++;
}
m->gobble(in);
if (wroteSomething == false) { m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine(); }
outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
+ m->mothurOut("Selected " + toString(selectedCount) + " sequences from your group file."); m->mothurOutEndLine();
+
return 0;
}
catch(exception& e) {
bool wroteSomething = false;
while(!in.eof()){
- if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str()); return 0; }
+ if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName); return 0; }
in >> name; //read from first column
in >> tax; //read from second column
}
in.close();
+ m->setGroups(Groups);
+
}
catch(exception& e) {
m->errorOut(e, "GetGroupsCommand", "readAccnos");
projects / mothur.git / blobdiff