fixed bug with dist.shared subsampling. added mode parameter to dist.shared so...

[mothur.git] / removeseqscommand.cpp

diff --git a/removeseqscommand.cpp b/removeseqscommand.cpp
index c7dd5c227622cb3fe0cbf764ff266fdbdea21271..2abe9ef78a8c76354d377087fa7a23deb27b4d25 100644 (file)
--- a/removeseqscommand.cpp
+++ b/removeseqscommand.cpp
@@ -191,19 +191,22 @@ RemoveSeqsCommand::RemoveSeqsCommand(string option)  {
                                        m->mothurOut("You have no valid accnos file and accnos is required."); m->mothurOutEndLine(); 
                                        abort = true;
                                }  
-                       }       
+                       }else { m->setAccnosFile(accnosfile); } 
                        
                        fastafile = validParameter.validFile(parameters, "fasta", true);
-                       if (fastafile == "not open") { abort = true; }
+                       if (fastafile == "not open") { fastafile = ""; abort = true; }
                        else if (fastafile == "not found") {  fastafile = "";  }        
-                       
+                       else { m->setFastaFile(fastafile); }
+                                                                  
                        namefile = validParameter.validFile(parameters, "name", true);
-                       if (namefile == "not open") { abort = 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") { abort = true; }
-                       else if (groupfile == "not found") {  groupfile = "";  }        
+                       else if (groupfile == "not found") {  groupfile = "";  }
+                       else { m->setGroupFile(groupfile); }
                        
                        alignfile = validParameter.validFile(parameters, "alignreport", true);
                        if (alignfile == "not open") { abort = true; }
@@ -212,15 +215,17 @@ RemoveSeqsCommand::RemoveSeqsCommand(string option)  {
                        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); }
                        
                        qualfile = validParameter.validFile(parameters, "qfile", true);
                        if (qualfile == "not open") { abort = true; }
                        else if (qualfile == "not found") {  qualfile = "";  }                  
-
+                       else { m->setQualFile(qualfile); }
                        
                        string usedDups = "true";
                        string temp = validParameter.validFile(parameters, "dups", false);      
@@ -232,6 +237,10 @@ RemoveSeqsCommand::RemoveSeqsCommand(string option)  {
                        
                        if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == "") && (qualfile == ""))  { m->mothurOut("You must provide at least one of the following: fasta, name, group, taxonomy, quality, alignreport or list."); m->mothurOutEndLine(); abort = true; }
                        
+                       if ((fastafile != "") && (namefile == "")) {
+                               vector<string> files; files.push_back(fastafile);
+                               parser.getNameFile(files);
+                       }
                }
 
        }
@@ -261,10 +270,8 @@ int RemoveSeqsCommand::execute(){
                if (taxfile != "")                      {               readTax();              }
                if (qualfile != "")                     {               readQual();             }
                
-               if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
-               
-               m->mothurOut("Removed " + toString(names.size()) + " sequences."); m->mothurOutEndLine();
-               
+               if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
+       
                if (outputNames.size() != 0) {
                        m->mothurOutEndLine();
                        m->mothurOut("Output File Names: "); m->mothurOutEndLine();
@@ -328,9 +335,10 @@ int RemoveSeqsCommand::readFasta(){
                string name;
                
                bool wroteSomething = false;
+               int removedCount = 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();
@@ -341,7 +349,7 @@ int RemoveSeqsCommand::readFasta(){
                                        wroteSomething = true;
                                        
                                        currSeq.printSequence(out);
-                               }
+                               }else {  removedCount++;  }
                        }
                        m->gobble(in);
                }
@@ -351,6 +359,8 @@ int RemoveSeqsCommand::readFasta(){
                if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
                outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
                
+               m->mothurOut("Removed " + toString(removedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
+               
                return 0;
                
        }
@@ -374,6 +384,7 @@ int RemoveSeqsCommand::readQual(){
                string name;
                
                bool wroteSomething = false;
+               int removedCount = 0;
                
                
                while(!in.eof()){       
@@ -405,7 +416,7 @@ int RemoveSeqsCommand::readQual(){
                                wroteSomething = true;
                                
                                out << name << endl << scores;
-                       }
+                       }else {  removedCount++;  }
                        
                        m->gobble(in);
                }
@@ -416,6 +427,8 @@ int RemoveSeqsCommand::readQual(){
                if (wroteSomething == false) { m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
                outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName); 
                
+               m->mothurOut("Removed " + toString(removedCount) + " sequences from your quality file."); m->mothurOutEndLine();
+               
                return 0;
                
        }
@@ -438,8 +451,12 @@ int RemoveSeqsCommand::readList(){
                m->openInputFile(listfile, in);
                
                bool wroteSomething = false;
+               int removedCount = 0;
                
                while(!in.eof()){
+                       
+                       removedCount = 0;
+                       
                        //read in list vector
                        ListVector list(in);
                        
@@ -449,7 +466,7 @@ int RemoveSeqsCommand::readList(){
                        
                        //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);
@@ -461,10 +478,12 @@ int RemoveSeqsCommand::readList(){
                                        
                                        //if that name is in the .accnos file, add it
                                        if (names.count(name) == 0) {  newNames += name + ",";  }
+                                       else {  removedCount++;  }
                                }
                        
                                //get last name
                                if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
+                               else {  removedCount++;  }
 
                                //if there are names in this bin add to new list
                                if (newNames != "") {  
@@ -486,7 +505,9 @@ int RemoveSeqsCommand::readList(){
                
                if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
                outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
-                               
+               
+               m->mothurOut("Removed " + toString(removedCount) + " sequences from your list file."); m->mothurOutEndLine();
+               
                return 0;
 
        }
@@ -510,10 +531,10 @@ int RemoveSeqsCommand::readName(){
                string name, firstCol, secondCol;
                
                bool wroteSomething = false;
-                               
+               int removedCount = 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;                        
@@ -530,7 +551,9 @@ int RemoveSeqsCommand::readName(){
                        
                        if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
                                for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
+                               removedCount += parsedNames.size();
                        }else {
+                               removedCount += parsedNames.size()-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) {
                                        
@@ -566,6 +589,8 @@ int RemoveSeqsCommand::readName(){
                if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
                outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
                
+               m->mothurOut("Removed " + toString(removedCount) + " sequences from your name file."); m->mothurOutEndLine();
+               
                return 0;
        }
        catch(exception& e) {
@@ -589,9 +614,10 @@ int RemoveSeqsCommand::readGroup(){
                string name, group;
                
                bool wroteSomething = false;
+               int removedCount = 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
@@ -600,7 +626,7 @@ int RemoveSeqsCommand::readGroup(){
                        if (names.count(name) == 0) {
                                wroteSomething = true;
                                out << name << '\t' << group << endl;
-                       }
+                       }else {  removedCount++;  }
                                        
                        m->gobble(in);
                }
@@ -610,6 +636,9 @@ int RemoveSeqsCommand::readGroup(){
                if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
                outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
                
+               m->mothurOut("Removed " + toString(removedCount) + " sequences from your group file."); m->mothurOutEndLine();
+
+               
                return 0;
        }
        catch(exception& e) {
@@ -631,9 +660,10 @@ int RemoveSeqsCommand::readTax(){
                string name, tax;
                
                bool wroteSomething = false;
+               int removedCount = 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 >> tax;                      //read from second column
@@ -642,7 +672,7 @@ int RemoveSeqsCommand::readTax(){
                        if (names.count(name) == 0) {
                                wroteSomething = true;
                                out << name << '\t' << tax << endl;
-                       }
+                       }else {  removedCount++;  }
                                        
                        m->gobble(in);
                }
@@ -652,6 +682,8 @@ int RemoveSeqsCommand::readTax(){
                if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
                outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
                
+               m->mothurOut("Removed " + toString(removedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
+               
                return 0;
        }
        catch(exception& e) {
@@ -675,6 +707,7 @@ int RemoveSeqsCommand::readAlign(){
                string name, junk;
                
                bool wroteSomething = false;
+               int removedCount = 0;
                
                //read column headers
                for (int i = 0; i < 16; i++) {  
@@ -684,7 +717,7 @@ int RemoveSeqsCommand::readAlign(){
                out << endl;
                
                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
                        
@@ -702,6 +735,7 @@ int RemoveSeqsCommand::readAlign(){
                                out << endl;
                                
                        }else {//still read just don't do anything with it
+                               removedCount++;  
                                
                                //read rest
                                for (int i = 0; i < 15; i++) {  
@@ -718,6 +752,9 @@ int RemoveSeqsCommand::readAlign(){
                if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
                outputTypes["alignreport"].push_back(outputFileName); outputNames.push_back(outputFileName);
                
+               m->mothurOut("Removed " + toString(removedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
+
+               
                return 0;
                
        }