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

[mothur.git] / sequenceparser.cpp

diff --git a/sequenceparser.cpp b/sequenceparser.cpp
index e60f19b0df0034de0c1a693d7c5a55c59ef0ef82..3eb508dd737e15d4f06cb04c202608b5d5e1fe7c 100644 (file)
--- a/sequenceparser.cpp
+++ b/sequenceparser.cpp
@@ -7,7 +7,7 @@
  *
  */
 
-#include "sequenceParser.h"
+#include "sequenceparser.h"
 
 
 /************************************************************/
@@ -37,13 +37,16 @@ SequenceParser::SequenceParser(string groupFile, string fastaFile, string nameFi
                m->openInputFile(fastaFile, in);
                
                map<string, string> seqName; //stores name -> sequence string so we can make new "unique" sequences when we parse the name file
+        int fastaCount = 0;
                while (!in.eof()) {
                        
                        if (m->control_pressed) { break; }
                        
                        Sequence seq(in); m->gobble(in);
+            fastaCount++;
+            if (m->debug) { if((fastaCount) % 1000 == 0){      m->mothurOut("[DEBUG]: reading seq " + toString(fastaCount) + "\n.");   } }
                        
-                       if (seq.getName() != "") {
+        if (seq.getName() != "") {
                                
                                 string group = groupMap->getGroup(seq.getName());
                                 if (group == "not found") {  error = 1; m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file and not in your groupfile, please correct."); m->mothurOutEndLine();  }
@@ -63,6 +66,8 @@ SequenceParser::SequenceParser(string groupFile, string fastaFile, string nameFi
                
                string first, second;
                int countName = 0;
+               set<string> thisnames1;
+               
                while(!inName.eof()) {
                        
                        if (m->control_pressed) { break; }
@@ -94,10 +99,12 @@ SequenceParser::SequenceParser(string groupFile, string fastaFile, string nameFi
                                        it = splitMap.find(group);
                                        if (it != splitMap.end()) { //adding seqs to this group
                                                (it->second) += "," + names[i];
+                                               thisnames1.insert(names[i]);
                                                countName++;
                                        }else { //first sighting of this group
                                                splitMap[group] = names[i];
                                                countName++;
+                                               thisnames1.insert(names[i]);
                                                
                                                //is this seq in the fasta file?
                                                if (i != 0) { //if not then we need to add a duplicate sequence to the seqs for this group so the new "fasta" and "name" files will match
@@ -129,8 +136,17 @@ SequenceParser::SequenceParser(string groupFile, string fastaFile, string nameFi
                inName.close();
                
                if (error == 1) { m->control_pressed = true; }
-               
+                       
                if (countName != (groupMap->getNumSeqs())) {
+                       vector<string> groupseqsnames = groupMap->getNamesSeqs();
+                       
+                       for (int i = 0; i < groupseqsnames.size(); i++) {
+                               set<string>::iterator itnamesfile = thisnames1.find(groupseqsnames[i]);
+                               if (itnamesfile == thisnames1.end()){
+                                       cout << "missing name " + groupseqsnames[i] << '\t' << allSeqsMap[groupseqsnames[i]] << endl;
+                               }
+                       }
+                       
                        m->mothurOutEndLine();
                        m->mothurOut("[ERROR]: Your name file contains " + toString(countName) + " valid sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
                        m->mothurOutEndLine();
@@ -240,6 +256,7 @@ vector<Sequence> SequenceParser::getSeqs(string g){
                        m->mothurOut("[ERROR]: No sequences available for group " + g + ", please correct."); m->mothurOutEndLine();
                }else {
                        seqForThisGroup = it->second;
+            if (m->debug) {  m->mothurOut("[DEBUG]: group " + g + " fasta file has " + toString(seqForThisGroup.size()) + " sequences.");  }
                }
                
                return seqForThisGroup; 
@@ -303,6 +320,7 @@ int SequenceParser::getSeqs(string g, string filename, bool uchimeFormat=false){
                                }
                                
                        }else { 
+                //m->mothurOut("Group " + g +  " contains " + toString(seqForThisGroup.size()) + " unique seqs.\n");
                                for (int i = 0; i < seqForThisGroup.size(); i++) {
                                        
                                        if(m->control_pressed) { out.close(); m->mothurRemove(filename); return 1; }
@@ -332,6 +350,7 @@ map<string, string> SequenceParser::getNameMap(string g){
                        m->mothurOut("[ERROR]: No nameMap available for group " + g + ", please correct."); m->mothurOutEndLine();
                }else {
                        nameMapForThisGroup = it->second;
+            if (m->debug) {  m->mothurOut("[DEBUG]: group " + g + " name file has " + toString(nameMapForThisGroup.size()) + " unique sequences.");  }
                }
                
                return nameMapForThisGroup;