Merge remote-tracking branch 'mothur/master'

[mothur.git] / removeseqscommand.cpp

diff --git a/removeseqscommand.cpp b/removeseqscommand.cpp
index 1fb94462ac48fa13ac871c6b9d3cf4f3ae2ceff6..00b94a9dac842ceeca4a2a0362d87f14ba12ca1b 100644 (file)
--- a/removeseqscommand.cpp
+++ b/removeseqscommand.cpp
@@ -10,6 +10,7 @@
 #include "removeseqscommand.h"
 #include "sequence.hpp"
 #include "listvector.hpp"
+#include "counttable.h"
 
 //**********************************************************************************************************************
 vector<string> RemoveSeqsCommand::setParameters(){     
@@ -71,7 +72,7 @@ string RemoveSeqsCommand::getOutputFileNameTag(string type, string inputName="")
             else if (type == "list")        {   outputFileName =  "pick" + m->getExtension(inputName);   }
             else if (type == "qfile")       {   outputFileName =  "pick" + m->getExtension(inputName);   }
             else if (type == "alignreport") {   outputFileName =  "pick.align.report";                   }
-            else if (type == "count")       {   outputFileName =  "pick.count.table";   }
+            else if (type == "count")       {   outputFileName =  "pick.count_table";   }
             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
         }
         return outputFileName;
@@ -405,6 +406,12 @@ int RemoveSeqsCommand::readFasta(){
                        if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
                        
                        Sequence currSeq(in);
+            
+            if (!dups) {//adjust name if needed
+                map<string, string>::iterator it = uniqueMap.find(currSeq.getName());
+                if (it != uniqueMap.end()) { currSeq.setName(it->second); }
+            }
+
                        name = currSeq.getName();
                        
                        if (name != "") {
@@ -412,7 +419,7 @@ int RemoveSeqsCommand::readFasta(){
                                if (names.count(name) == 0) {
                                        wroteSomething = true;
                                        
-                                       currSeq.printSequence(out);
+                    currSeq.printSequence(out);
                                }else {  removedCount++;  }
                        }
                        m->gobble(in);
@@ -476,6 +483,11 @@ int RemoveSeqsCommand::readQual(){
                        
                        m->gobble(in);
                        
+            if (!dups) {//adjust name if needed
+                map<string, string>::iterator it = uniqueMap.find(saveName);
+                if (it != uniqueMap.end()) { name = ">" + it->second; saveName = it->second; }
+            }
+            
                        if (names.count(saveName) == 0) {
                                wroteSomething = true;
                                
@@ -538,6 +550,14 @@ int RemoveSeqsCommand::readCount(){
         }
         in.close();
                out.close();
+        
+        //check for groups that have been eliminated
+        CountTable ct;
+        if (ct.testGroups(outputFileName)) {
+            ct.readTable(outputFileName);
+            ct.printTable(outputFileName);
+        }
+
                
                if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
                outputTypes["count"].push_back(outputFileName); outputNames.push_back(outputFileName);
@@ -686,6 +706,8 @@ int RemoveSeqsCommand::readName(){
                                                wroteSomething = true;
                                                
                                                out << validSecond[0] << '\t';
+                        //we are changing the unique name in the fasta file
+                        uniqueMap[firstCol] = validSecond[0];
                                                
                                                //you know you have at least one valid second since first column is valid
                                                for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
@@ -779,9 +801,15 @@ int RemoveSeqsCommand::readTax(){
                        in >> name;                             //read from first column
                        in >> tax;                      //read from second column
                        
+            if (!dups) {//adjust name if needed
+                map<string, string>::iterator it = uniqueMap.find(name);
+                if (it != uniqueMap.end()) { name = it->second; }
+            }
+            
                        //if this name is in the accnos file
                        if (names.count(name) == 0) {
                                wroteSomething = true;
+            
                                out << name << '\t' << tax << endl;
                        }else {  removedCount++;  }
                                        
@@ -831,6 +859,11 @@ int RemoveSeqsCommand::readAlign(){
                        if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
                        
                        in >> name;                             //read from first column
+            
+            if (!dups) {//adjust name if needed
+                map<string, string>::iterator it = uniqueMap.find(name);
+                if (it != uniqueMap.end()) { name = it->second; }
+            }
                        
                        //if this name is in the accnos file
                        if (names.count(name) == 0) {