From 9a808b553f14a25f9c5e94da92a7077a57054e8e Mon Sep 17 00:00:00 2001
From: Sarah Westcott <mothur.westcott@gmail.com>
Date: Tue, 25 Feb 2014 11:57:05 -0500
Subject: [PATCH] =?utf8?q?fixed=20bug=20in=20make.biom=20with=20picrust=20?=
 =?utf8?q?that=20didn=E2=80=99t=20allow=20for=20unclassified=20OTUs.=20fix?=
 =?utf8?q?ed=20bug=20in=20fastq.info=20that=20quit=20after=20100000=20seqs?=
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

---
 makebiomcommand.cpp    | 49 ++++++++++++++++++++++++++++++------------
 parsefastaqcommand.cpp |  1 -
 2 files changed, 35 insertions(+), 15 deletions(-)

diff --git a/makebiomcommand.cpp b/makebiomcommand.cpp
index 4c35092..36bda15 100644
--- a/makebiomcommand.cpp
+++ b/makebiomcommand.cpp
@@ -680,8 +680,17 @@ int MakeBiomCommand::getGreenGenesOTUIDs(vector<SharedRAbundVector*>& lookup, ma
         for (map<string, string>::iterator it = labelTaxMap.begin(); it != labelTaxMap.end(); it++) { //maps label -> consensus taxonomy
             if (m->control_pressed) { break; }
             
+            string OTUTaxonomy = it->second;
+            
+            //remove confidences
+            m->removeConfidences(OTUTaxonomy);
+            
+            //remove unclassifieds to match template
+            int thisPos = OTUTaxonomy.find("unclassified;");
+            if (thisPos != string::npos) {  OTUTaxonomy = OTUTaxonomy.substr(0, thisPos);  }
+            
             //get list of reference ids that map to this taxonomy
-            vector<string> referenceIds = phylo.getSeqs(it->second);
+            vector<string> referenceIds = phylo.getSeqs(OTUTaxonomy);
             
             if (m->control_pressed) { break; }
             
@@ -724,6 +733,7 @@ int MakeBiomCommand::getGreenGenesOTUIDs(vector<SharedRAbundVector*>& lookup, ma
         map<string, string> newLabelTaxMap;
         //loop through ggOTUID list combining mothur otus and adjusting labels
         //ggOTUIDs = 16097 -> <OTU01, OTU10, OTU22>
+        
         for (map<string, vector<string> >::iterator itMap = ggOTUIDs.begin(); itMap != ggOTUIDs.end(); itMap++) {
 			if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
             
@@ -737,14 +747,18 @@ int MakeBiomCommand::getGreenGenesOTUIDs(vector<SharedRAbundVector*>& lookup, ma
             vector<int> abunds; abunds.resize(lookup.size(), 0);
             string mergeString = "";
             vector<float> boots; boots.resize(scores.size(), 0);
+            bool scoresNULL = false;
             for (int j = 0; j < itMap->second.size(); j++) { //<OTU01, OTU10, OTU22>
-                //merge bootstrap scores
-                vector<string> scores;
-                vector<string> taxonomies = parseTax(it->second, scores);
-                for (int i = 0; i < boots.size(); i++) {
-                    float tempScore; m->mothurConvert(scores[i], tempScore);
-                    boots[i] += tempScore;
-                }
+                
+                if (scores[0] != "null") {
+                    //merge bootstrap scores
+                    vector<string> scores;
+                    vector<string> taxonomies = parseTax(it->second, scores);
+                    for (int i = 0; i < boots.size(); i++) {
+                        float tempScore; m->mothurConvert(scores[i], tempScore);
+                        boots[i] += tempScore;
+                    }
+                }else { scoresNULL = true; }
                 
                 //merge abunds
                 mergeString += (itMap->second)[j] + " ";
@@ -757,14 +771,21 @@ int MakeBiomCommand::getGreenGenesOTUIDs(vector<SharedRAbundVector*>& lookup, ma
             
             //average scores
             //add merged otu to new lookup
-            for (int j = 0; j < boots.size(); j++) { boots[j] /= (float) itMap->second.size(); }
-            
-            //assemble new taxomoy
             string newTaxString = "";
-            for (int j = 0; j < boots.size(); j++) {
-                newTaxString += taxonomies[j] + "(" + toString(boots[j]) + ");";
+            if (!scoresNULL) {
+                for (int j = 0; j < boots.size(); j++) { boots[j] /= (float) itMap->second.size(); }
+            
+                //assemble new taxomoy
+                for (int j = 0; j < boots.size(); j++) {
+                    newTaxString += taxonomies[j] + "(" + toString(boots[j]) + ");";
+                }
+            }else {
+                //assemble new taxomoy
+                for (int j = 0; j < taxonomies.size(); j++) {
+                    newTaxString += taxonomies[j] + ";";
+                }
             }
-
+            
             //set new gg otu id to taxonomy. OTU01 -> k__Bacteria becomes 16097 -> k__Bacteria
             //find taxonomy of this otu
             newLabelTaxMap[itMap->first] = newTaxString;
diff --git a/parsefastaqcommand.cpp b/parsefastaqcommand.cpp
index c3bd393..a509ae9 100644
--- a/parsefastaqcommand.cpp
+++ b/parsefastaqcommand.cpp
@@ -308,7 +308,6 @@ int ParseFastaQCommand::execute(){
                 }
                 //report progress
                 if((count+1) % 10000 == 0){	m->mothurOut(toString(count+1)); m->mothurOutEndLine();		}
-                if(count > 100000){	break;	}
                 count++;
 			}
 		}
-- 
2.39.2