#include "readtreecommand.h"
+//**********************************************************************************************************************
+vector<string> ReadTreeCommand::getValidParameters(){
+ try {
+ string Array[] = {"tree","group","name","outputdir","inputdir"};
+ vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ReadTreeCommand", "getValidParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+vector<string> ReadTreeCommand::getRequiredParameters(){
+ try {
+ string Array[] = {"tree"};
+ vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ReadTreeCommand", "getRequiredParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+vector<string> ReadTreeCommand::getRequiredFiles(){
+ try {
+ vector<string> myArray;
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ReadTreeCommand", "getRequiredFiles");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
ReadTreeCommand::ReadTreeCommand(string option) {
try {
groupfile = validParameter.validFile(parameters, "group", true);
if (groupfile == "not open") { abort = true; }
- else if (groupfile == "not found") { groupfile = ""; m->mothurOut("group is a required parameter for the read.tree command."); m->mothurOutEndLine(); abort = true; }
- else {
+ else if (groupfile == "not found") {
+ groupfile = "";
+
+ m->mothurOut("You have not provided a group file. I am assumming all sequence are from the same group."); m->mothurOutEndLine();
+
+ if (treefile != "") { Tree* tree = new Tree(treefile); delete tree; } //extracts names from tree to make faked out groupmap
+
+ globaldata->setGroupFile(groupfile);
+ //read in group map info.
+ treeMap = new TreeMap();
+ for (int i = 0; i < globaldata->Treenames.size(); i++) { treeMap->addSeq(globaldata->Treenames[i], "Group1"); }
+ globaldata->gTreemap = treeMap;
+
+ }else {
globaldata->setGroupFile(groupfile);
//read in group map info.
treeMap = new TreeMap(groupfile);
m->mothurOut("It also must be run before using the parsimony command, unless you are using the randomtree parameter.\n");
m->mothurOut("The read.tree command parameters are tree, group and name.\n");
m->mothurOut("The read.tree command should be in the following format: read.tree(tree=yourTreeFile, group=yourGroupFile).\n");
- m->mothurOut("The tree and group parameters are both required.\n");
+ m->mothurOut("The tree and group parameters are both required, if no group file is given then one group is assumed.\n");
m->mothurOut("The name parameter allows you to enter a namefile.\n");
m->mothurOut("Note: No spaces between parameter labels (i.e. tree), '=' and parameters (i.e.yourTreefile).\n\n");
}
}
-// Sarah, isn't this checking already done when assigning the sequences to the groups? it makes read.tree
-// wicked slow... For some reason my tree is coming in here eventhough the number of sequences in the tree
-// agrees with the number of lines in the name file and the number of sequences represented by the name file
-// agrees with the number of sequences in the group file
-
- //output any names that are in group file but not in tree
+ //if you provide a namefile we will use the numNames in the namefile as long as the number of unique match the tree names size.
+ int numNamesInTree;
+ if (namefile != "") {
+ if (numUniquesInName == globaldata->Treenames.size()) { numNamesInTree = nameMap.size(); }
+ else { numNamesInTree = globaldata->Treenames.size(); }
+ }else { numNamesInTree = globaldata->Treenames.size(); }
-// if (globaldata->Treenames.size() < treeMap->getNumSeqs()) {
-// cout << "in here" << endl;
-// for (int i = 0; i < treeMap->namesOfSeqs.size(); i++) {
-// //is that name in the tree?
-// int count = 0;
-// for (int j = 0; j < globaldata->Treenames.size(); j++) {
-// if (treeMap->namesOfSeqs[i] == globaldata->Treenames[j]) { break; } //found it
-// count++;
-// }
-//
-// if (m->control_pressed) {
-// for (int i = 0; i < T.size(); i++) { delete T[i]; }
-// globaldata->gTree.clear();
-// delete globaldata->gTreemap;
-// return 0;
-// }
-//
-// //then you did not find it so report it
-// if (count == globaldata->Treenames.size()) {
-// //if it is in your namefile then don't remove
-// map<string, string>::iterator it = nameMap.find(treeMap->namesOfSeqs[i]);
-//
-// if (it == nameMap.end()) {
-// m->mothurOut(treeMap->namesOfSeqs[i] + " is in your groupfile and not in your tree. It will be disregarded."); m->mothurOutEndLine();
-// treeMap->removeSeq(treeMap->namesOfSeqs[i]);
-// i--; //need this because removeSeq removes name from namesOfSeqs
-// }
-// }
-// }
-//
-// globaldata->gTreemap = treeMap;
-// }
+ //output any names that are in group file but not in tree
+ if (numNamesInTree < treeMap->getNumSeqs()) {
+ for (int i = 0; i < treeMap->namesOfSeqs.size(); i++) {
+ //is that name in the tree?
+ int count = 0;
+ for (int j = 0; j < globaldata->Treenames.size(); j++) {
+ if (treeMap->namesOfSeqs[i] == globaldata->Treenames[j]) { break; } //found it
+ count++;
+ }
+
+ if (m->control_pressed) {
+ for (int i = 0; i < T.size(); i++) { delete T[i]; }
+ globaldata->gTree.clear();
+ delete globaldata->gTreemap;
+ return 0;
+ }
+
+ //then you did not find it so report it
+ if (count == globaldata->Treenames.size()) {
+ //if it is in your namefile then don't remove
+ map<string, string>::iterator it = nameMap.find(treeMap->namesOfSeqs[i]);
+
+ if (it == nameMap.end()) {
+ m->mothurOut(treeMap->namesOfSeqs[i] + " is in your groupfile and not in your tree. It will be disregarded."); m->mothurOutEndLine();
+ treeMap->removeSeq(treeMap->namesOfSeqs[i]);
+ i--; //need this because removeSeq removes name from namesOfSeqs
+ }
+ }
+ }
+
+ globaldata->gTreemap = treeMap;
+ }
+
return 0;
}
catch(exception& e) {
int ReadTreeCommand::readNamesFile() {
try {
globaldata->names.clear();
+ numUniquesInName = 0;
ifstream in;
m->openInputFile(namefile, in);
while(!in.eof()) {
in >> first >> second; m->gobble(in);
+ numUniquesInName++;
+
itNames = globaldata->names.find(first);
if (itNames == globaldata->names.end()) {
globaldata->names[first] = second;
vector<string> dupNames;
m->splitAtComma(second, dupNames);
- for (int i = 0; i < dupNames.size(); i++) { nameMap[dupNames[i]] = dupNames[i]; }
- }else { m->mothurOut(first + " has already been seen in namefile, disregarding names file."); m->mothurOutEndLine(); in.close(); globaldata->names.clear(); return 1; }
+ for (int i = 0; i < dupNames.size(); i++) { nameMap[dupNames[i]] = dupNames[i]; if ((groupfile == "") && (i != 0)) { globaldata->gTreemap->addSeq(dupNames[i], "Group1"); } }
+ }else { m->mothurOut(first + " has already been seen in namefile, disregarding names file."); m->mothurOutEndLine(); in.close(); globaldata->names.clear(); namefile = ""; return 1; }
}
in.close();
projects / mothur.git / blobdiff