added distance command and filterseqs

diff --git a/Mothur.xcodeproj/project.pbxproj b/Mothur.xcodeproj/project.pbxproj
index d29d52e10c177f12f9721223058495cfd28fffde..7beb6cbfe80f60a8db0c7a74dbcb02e5762e4f45 100644 (file)
--- a/Mothur.xcodeproj/project.pbxproj
+++ b/Mothur.xcodeproj/project.pbxproj
@@ -41,7 +41,6 @@
                378C1B050FB0644E004D63F5 /* readclustal.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 378C1AF20FB0644D004D63F5 /* readclustal.cpp */; };
                378C1B060FB0644E004D63F5 /* readfasta.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 378C1AF40FB0644D004D63F5 /* readfasta.cpp */; };
                378C1B070FB0644E004D63F5 /* readnexus.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 378C1AF60FB0644D004D63F5 /* readnexus.cpp */; };
-               378C1B080FB0644E004D63F5 /* readseqscommand.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 378C1AF90FB0644D004D63F5 /* readseqscommand.cpp */; };
                378C1B090FB0644E004D63F5 /* readseqsphylip.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 378C1AFB0FB0644D004D63F5 /* readseqsphylip.cpp */; };
                378C1B0A0FB0644E004D63F5 /* sequencedb.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 378C1AFD0FB0644D004D63F5 /* sequencedb.cpp */; };
                378C1B0B0FB0644E004D63F5 /* sharedjackknife.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 378C1AFF0FB0644D004D63F5 /* sharedjackknife.cpp */; };
@@ -49,6 +48,7 @@
                379293C30F2DE73400B9034A /* treemap.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 379293C20F2DE73400B9034A /* treemap.cpp */; };
                379294700F2E191800B9034A /* parsimonycommand.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 3792946F0F2E191800B9034A /* parsimonycommand.cpp */; };
                3792948A0F2E258500B9034A /* parsimony.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 379294890F2E258500B9034A /* parsimony.cpp */; };
+               379643ED0FB9B5A80081FDB6 /* readseqs.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 379643EB0FB9B5A80081FDB6 /* readseqs.cpp */; };
                37AD4CE40F28AEA300AA2D49 /* sharedlistvector.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 37AD4CE30F28AEA300AA2D49 /* sharedlistvector.cpp */; };
                37AD4DBB0F28E2FE00AA2D49 /* tree.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 37AD4DBA0F28E2FE00AA2D49 /* tree.cpp */; };
                37AD4DCA0F28F3DD00AA2D49 /* readtree.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 37AD4DC90F28F3DD00AA2D49 /* readtree.cpp */; };
@@ -225,8 +225,6 @@
                378C1AF60FB0644D004D63F5 /* readnexus.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = readnexus.cpp; sourceTree = "<group>"; };
                378C1AF70FB0644D004D63F5 /* readnexus.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = readnexus.h; sourceTree = "<group>"; };
                378C1AF80FB0644D004D63F5 /* readnexusal.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = readnexusal.h; sourceTree = "<group>"; };
-               378C1AF90FB0644D004D63F5 /* readseqscommand.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = readseqscommand.cpp; sourceTree = "<group>"; };
-               378C1AFA0FB0644D004D63F5 /* readseqscommand.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = readseqscommand.h; sourceTree = "<group>"; };
                378C1AFB0FB0644D004D63F5 /* readseqsphylip.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = readseqsphylip.cpp; sourceTree = "<group>"; };
                378C1AFC0FB0644D004D63F5 /* readseqsphylip.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = readseqsphylip.h; sourceTree = "<group>"; };
                378C1AFD0FB0644D004D63F5 /* sequencedb.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = sequencedb.cpp; sourceTree = "<group>"; };
@@ -242,6 +240,8 @@
                379294880F2E258500B9034A /* parsimony.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = parsimony.h; sourceTree = "<group>"; };
                379294890F2E258500B9034A /* parsimony.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = parsimony.cpp; sourceTree = "<group>"; };
                3792948D0F2E271100B9034A /* treecalculator.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = treecalculator.h; sourceTree = "<group>"; };
+               379643EB0FB9B5A80081FDB6 /* readseqs.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = readseqs.cpp; sourceTree = "<group>"; };
+               379643EC0FB9B5A80081FDB6 /* readseqs.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = readseqs.h; sourceTree = "<group>"; };
                37AD4CE20F28AEA300AA2D49 /* sharedlistvector.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = sharedlistvector.h; sourceTree = "<group>"; };
                37AD4CE30F28AEA300AA2D49 /* sharedlistvector.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = sharedlistvector.cpp; sourceTree = "<group>"; };
                37AD4DB90F28E2FE00AA2D49 /* tree.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = tree.h; sourceTree = "<group>"; };
@@ -484,7 +484,6 @@
                                7E4130F70F8E58FA00381DD0 /* dlibshuff.h */,
                                7E4130F60F8E58FA00381DD0 /* dlibshuff.cpp */,
                                37D927D60F21331F001D4494 /* display.h */,
-                               37C753DF0FB3492400DBD02E /* dist.h */,
                                37D927D80F21331F001D4494 /* engine.hpp */,
                                37D927D70F21331F001D4494 /* engine.cpp */,
                                37D928B10F213472001D4494 /* errorcheckor */,
@@ -514,24 +513,7 @@
                                37D928060F21331F001D4494 /* rarefact.h */,
                                37D928050F21331F001D4494 /* rarefact.cpp */,
                                37D928090F21331F001D4494 /* rarefactioncurvedata.h */,
-                               378C1AF30FB0644D004D63F5 /* readclustal.h */,
-                               378C1AF20FB0644D004D63F5 /* readclustal.cpp */,
-                               375AA1340F9E433D008EF9B8 /* readcolumn.h */,
-                               375AA1330F9E433D008EF9B8 /* readcolumn.cpp */,
-                               378C1AF50FB0644D004D63F5 /* readfasta.h */,
-                               378C1AF40FB0644D004D63F5 /* readfasta.cpp */,
-                               37D928130F21331F001D4494 /* readmatrix.hpp */,
-                               378C1AF70FB0644D004D63F5 /* readnexus.h */,
-                               378C1AF60FB0644D004D63F5 /* readnexus.cpp */,
-                               378C1AF80FB0644D004D63F5 /* readnexusal.h */,
-                               375AA1360F9E433D008EF9B8 /* readotu.h */,
-                               375AA1350F9E433D008EF9B8 /* readotu.cpp */,
-                               375AA1380F9E433D008EF9B8 /* readphylip.h */,
-                               375AA1370F9E433D008EF9B8 /* readphylip.cpp */,
-                               378C1AFC0FB0644D004D63F5 /* readseqsphylip.h */,
-                               378C1AFB0FB0644D004D63F5 /* readseqsphylip.cpp */,
-                               37AD4DC80F28F3DD00AA2D49 /* readtree.h */,
-                               37AD4DC90F28F3DD00AA2D49 /* readtree.cpp */,
+                               3796441D0FB9B9650081FDB6 /* read */,
                                37D928210F21331F001D4494 /* shared.h */,
                                37D928200F21331F001D4494 /* shared.cpp */,
                                211C38310F961DD400FEE541 /* sharedutilities.h */,
@@ -555,6 +537,33 @@
                        name = Products;
                        sourceTree = "<group>";
                };
+               3796441D0FB9B9650081FDB6 /* read */ = {
+                       isa = PBXGroup;
+                       children = (
+                               378C1AF30FB0644D004D63F5 /* readclustal.h */,
+                               378C1AF20FB0644D004D63F5 /* readclustal.cpp */,
+                               375AA1340F9E433D008EF9B8 /* readcolumn.h */,
+                               375AA1330F9E433D008EF9B8 /* readcolumn.cpp */,
+                               378C1AF50FB0644D004D63F5 /* readfasta.h */,
+                               378C1AF40FB0644D004D63F5 /* readfasta.cpp */,
+                               37D928130F21331F001D4494 /* readmatrix.hpp */,
+                               378C1AF70FB0644D004D63F5 /* readnexus.h */,
+                               378C1AF60FB0644D004D63F5 /* readnexus.cpp */,
+                               378C1AF80FB0644D004D63F5 /* readnexusal.h */,
+                               375AA1360F9E433D008EF9B8 /* readotu.h */,
+                               375AA1350F9E433D008EF9B8 /* readotu.cpp */,
+                               375AA1380F9E433D008EF9B8 /* readphylip.h */,
+                               375AA1370F9E433D008EF9B8 /* readphylip.cpp */,
+                               378C1AFC0FB0644D004D63F5 /* readseqsphylip.h */,
+                               378C1AFB0FB0644D004D63F5 /* readseqsphylip.cpp */,
+                               379643EB0FB9B5A80081FDB6 /* readseqs.cpp */,
+                               379643EC0FB9B5A80081FDB6 /* readseqs.h */,
+                               37AD4DC80F28F3DD00AA2D49 /* readtree.h */,
+                               37AD4DC90F28F3DD00AA2D49 /* readtree.cpp */,
+                       );
+                       name = read;
+                       sourceTree = "<group>";
+               };
                37D928A60F2133C0001D4494 /* calculators */ = {
                        isa = PBXGroup;
                        children = (
@@ -573,6 +582,7 @@
                                37D927BF0F21331F001D4494 /* chao1.cpp */,
                                7EC3D4510FA0FFF900338DA5 /* coverage.h */,
                                7EC3D4500FA0FFF900338DA5 /* coverage.cpp */,
+                               37C753DF0FB3492400DBD02E /* dist.h */,
                                37C753F00FB34AE800DBD02E /* eachgapdist.h */,
                                37C753F40FB34C0300DBD02E /* eachgapignore.h */,
                                EB9303F70F53517300E8EF26 /* geom.h */,
@@ -710,8 +720,6 @@
                                372E12950F263D5A0095CF7E /* readdistcommand.cpp */,
                                372E126E0F26365B0095CF7E /* readotucommand.h */,
                                372E126F0F26365B0095CF7E /* readotucommand.cpp */,
-                               378C1AFA0FB0644D004D63F5 /* readseqscommand.h */,
-                               378C1AF90FB0644D004D63F5 /* readseqscommand.cpp */,
                                37E5F4900F2A3DA800F8D827 /* readtreecommand.h */,
                                37E5F4910F2A3DA800F8D827 /* readtreecommand.cpp */,
                                37D928270F21331F001D4494 /* sharedcommand.h */,
@@ -965,12 +973,12 @@
                                378C1B050FB0644E004D63F5 /* readclustal.cpp in Sources */,
                                378C1B060FB0644E004D63F5 /* readfasta.cpp in Sources */,
                                378C1B070FB0644E004D63F5 /* readnexus.cpp in Sources */,
-                               378C1B080FB0644E004D63F5 /* readseqscommand.cpp in Sources */,
                                378C1B090FB0644E004D63F5 /* readseqsphylip.cpp in Sources */,
                                378C1B0A0FB0644E004D63F5 /* sequencedb.cpp in Sources */,
                                378C1B0B0FB0644E004D63F5 /* sharedjackknife.cpp in Sources */,
                                378C1B0C0FB0644E004D63F5 /* sharedmarczewski.cpp in Sources */,
                                37C753CE0FB3415200DBD02E /* distancecommand.cpp in Sources */,
+                               379643ED0FB9B5A80081FDB6 /* readseqs.cpp in Sources */,
                        );
                        runOnlyForDeploymentPostprocessing = 0;
                };

diff --git a/commandfactory.cpp b/commandfactory.cpp
index e357e6cd70c45ec8c0fdeaa4100d8473b7fcebda..ea43c48e6b84e4c2e83de0c067e081c2a88da49f 100644 (file)
--- a/commandfactory.cpp
+++ b/commandfactory.cpp
@@ -11,7 +11,6 @@
 #include "readdistcommand.h"
 #include "readtreecommand.h"
 #include "readotucommand.h"
-#include "readseqscommand.h"
 #include "clustercommand.h"
 #include "parselistcommand.h"
 #include "collectcommand.h"
@@ -41,6 +40,7 @@
 #include "treegroupscommand.h"
 #include "bootstrapsharedcommand.h"
 #include "concensuscommand.h"
+#include "distancecommand.h"
 
 
 /***********************************************************/
@@ -66,7 +66,6 @@ Command* CommandFactory::getCommand(string commandName){
 
                         if(commandName == "read.dist")                         {       command = new ReadDistCommand();                        }
                else if(commandName == "read.otu")                              {       command = new ReadOtuCommand();                         }
-               else if(commandName == "read.seqs")                             {       command = new ReadSeqsCommand();                        }
                else if(commandName == "read.tree")                             {       command = new ReadTreeCommand();                        }
                else if(commandName == "cluster")                               {       command = new ClusterCommand();                         }
                else if(commandName == "deconvolute")                   {       command = new DeconvoluteCommand();                     }
@@ -93,6 +92,7 @@ Command* CommandFactory::getCommand(string commandName){
                else if(commandName == "tree.shared")                   {   command = new TreeGroupCommand();                   }
                else if(commandName == "bootstrap.shared")              {   command = new BootSharedCommand();                  }
                else if(commandName == "concensus")                             {   command = new ConcensusCommand();                   }
+               else if(commandName == "distance")                              {   command = new DistanceCommand();                    }
                else                                                                                    {       command = new NoCommand();                                      }
 
                return command;

diff --git a/distancecommand.cpp b/distancecommand.cpp
index c20ddb91aecf7a4c0f7460cce4b11601c64ac597..0714ab3623181a86d5e0485b5a4e6613144a6b66 100644 (file)
--- a/distancecommand.cpp
+++ b/distancecommand.cpp
@@ -21,10 +21,8 @@ DistanceCommand::DistanceCommand(){
                globaldata = GlobalData::getInstance();
                validCalculator = new ValidCalculators();
                ends = globaldata->getEnds();
-               seqDB = globaldata->gSequenceDB;
                convert(globaldata->getProcessors(), processors);
                convert(globaldata->getCutOff(), cutoff);
-               distFile = getRootName(globaldata->getFastaFile()) + "dist";
                
                int i;
                if (ends != "T") {
@@ -52,7 +50,6 @@ DistanceCommand::DistanceCommand(){
                        }
                }
                                
-                               
                //reset calc for next command
                globaldata->setCalc("");
 
@@ -71,9 +68,28 @@ DistanceCommand::DistanceCommand(){
 
 int DistanceCommand::execute(){
        try {
+               
+               //read file
+               string filename = globaldata->inputFileName;
+               
+               if(globaldata->getFastaFile() != "") {
+                       readSeqs =  new ReadFasta(filename); }
+               else if(globaldata->getNexusFile() != "") {
+                       readSeqs = new ReadNexus(filename); }
+               else if(globaldata->getClustalFile() != "") {
+                       readSeqs = new ReadClustal(filename); }
+               else if(globaldata->getPhylipFile() != "") {
+                       readSeqs = new ReadPhylip(filename); }
+                       
+               readSeqs->read();
+               seqDB = readSeqs->getDB();
+       
                int numSeqs = seqDB->getNumSeqs();
                
-               system(("rm "+distFile).c_str() );
+               string distFile = getRootName(globaldata->getFastaFile()) + "dist";
+               
+               remove(distFile.c_str());
+               
                if(processors == 1){
                        driver(distCalculator, seqDB, 0, numSeqs, distFile, cutoff);    
                }       
@@ -82,13 +98,13 @@ int DistanceCommand::execute(){
                        int pid = fork();
                        if(pid > 0){
                                driver(distCalculator, seqDB, 0, (numSeqs/sqrt(2)), distFile + "tempa", cutoff);        
-//                             system(("cat " + distFile + "tempa" + " >> " + distFile).c_str());
-//                             system(("rm " + distFile + "tempa").c_str());                           
+                               appendFiles((distFile+"tempa"), distFile);
+                               remove((distFile + "tempa").c_str());   
                        }
                        else{
                                driver(distCalculator, seqDB, (numSeqs/sqrt(2)), numSeqs, distFile + "tempb", cutoff);  
-//                             system(("cat " + distFile + "tempb" + " >> " + distFile).c_str());
-//                             system(("rm " + distFile + "tempb").c_str());                           
+                               appendFiles((distFile+"tempb"), distFile);
+                               remove((distFile + "tempb").c_str());   
                        }
                        wait(NULL);
 
@@ -99,37 +115,20 @@ int DistanceCommand::execute(){
                                int pid2 = fork();
                                if(pid2 > 0){
                                        driver(distCalculator, seqDB, 0, sqrt(3) * numSeqs / 3, distFile + "tempa", cutoff);
-                                       #ifdef HAVE_CAT
-                                               system(("cat " + distFile + "tempa" + " >> " + distFile).c_str());
-                                       #else
-                                               #ifdef HAVE_COPY
-//get system call from pat system(("copy " + distFile + "tempa").c_str());
-                                               #else
-                                                       cout << "Sorry but I can't continue because this operating system doesn't appear to support the cat() or copy() system calls." << endl;
-                                               #endif
-                                       #endif
-                                       
-                                       #ifdef HAVE_RM
-                                               system(("rm " + distFile + "tempa").c_str());   
-                                       #else
-                                               #ifdef HAVE_ERASE
-                                                       system(("erase " + distFile + "tempa").c_str());
-                                               #else
-                                                       cout << "Sorry but I can't remove the required files because this operating system doesn't appear to support the rm() or erase() system calls." << endl;
-                                               #endif
-                                       #endif
+                                       appendFiles(distFile+"tempa", distFile);
+                                       remove((distFile + "tempa").c_str());   
                                }
                                else{
                                        driver(distCalculator, seqDB, sqrt(3) * numSeqs / 3, sqrt(6) * numSeqs / 3, distFile + "tempb", cutoff);        
-                                       system(("cat " + distFile + "tempb" + " >> " + distFile).c_str());
-                                       system(("rm " + distFile + "tempb").c_str());                           
+                                       appendFiles(distFile+"tempb", distFile);
+                                       remove((distFile + "tempb").c_str());                           
                                }
                                wait(NULL);
                        }
                        else{
                                driver(distCalculator, seqDB, sqrt(6) * numSeqs / 3, numSeqs, distFile + "tempc", cutoff);      
-                               system(("cat " + distFile + "tempc" + " >> " + distFile).c_str());
-                               system(("rm " + distFile + "tempc").c_str());                           
+                               appendFiles(distFile+"tempc", distFile);
+                               remove((distFile + "tempc").c_str());                   
                        }
                        wait(NULL);
                }
@@ -139,13 +138,17 @@ int DistanceCommand::execute(){
                                int pid2 = fork();
                                if(pid2 > 0){
                                        driver(distCalculator, seqDB, 0, numSeqs / 2, distFile + "tempa", cutoff);      
-                                       system(("cat " + distFile + "tempa" + " >> " + distFile).c_str());
-                                       system(("rm " + distFile + "tempa").c_str());                           
+                                       //system(("cat " + distFile + "tempa" + " >> " + distFile).c_str());
+                                       appendFiles(distFile+"tempa", distFile);
+                                       //system(("rm " + distFile + "tempa").c_str()); 
+                                       remove((distFile + "tempa").c_str());                   
                                }
                                else{
                                        driver(distCalculator, seqDB, numSeqs / 2, (numSeqs/sqrt(2)), distFile + "tempb", cutoff);      
-                                       system(("cat " + distFile + "tempb" + " >> " + distFile).c_str());
-                                       system(("rm " + distFile + "tempb").c_str());                           
+                                       //system(("cat " + distFile + "tempb" + " >> " + distFile).c_str());
+                                       appendFiles(distFile+"tempb", distFile);
+                                       //system(("rm " + distFile + "tempb").c_str());
+                                       remove((distFile + "tempb").c_str());                           
                                }
                                wait(NULL);
                        }
@@ -153,13 +156,17 @@ int DistanceCommand::execute(){
                                int pid3 = fork();
                                if(pid3 > 0){
                                        driver(distCalculator, seqDB, (numSeqs/sqrt(2)), (sqrt(3) * numSeqs / 2), distFile + "tempc", cutoff);  
-                                       system(("cat " + distFile + "tempc" + " >> " + distFile).c_str());
-                                       system(("rm " + distFile + "tempc").c_str());                           
+                                       //system(("cat " + distFile + "tempc" + " >> " + distFile).c_str());
+                                       appendFiles(distFile+"tempc", distFile);
+                                       //system(("rm " + distFile + "tempc").c_str());
+                                       remove((distFile + "tempc").c_str());                           
                                }
                                else{
                                        driver(distCalculator, seqDB, (sqrt(3) * numSeqs / 2), numSeqs, distFile + "tempd", cutoff);    
-                                       system(("cat " + distFile + "tempd" + " >> " + distFile).c_str());
-                                       system(("rm " + distFile + "tempd").c_str());                           
+                                       //system(("cat " + distFile + "tempd" + " >> " + distFile).c_str());
+                                       appendFiles(distFile+"tempd", distFile);
+                                       //system(("rm " + distFile + "tempd").c_str());
+                                       remove((distFile + "tempd").c_str());                           
                                }
                                wait(NULL);
                        }
@@ -194,12 +201,13 @@ int DistanceCommand::driver(Dist* distCalculator, SequenceDB* align, int startLi
        
                for(int i=startLine;i<endLine;i++){
                
-                       for(int j=0;j<i;j++){
-                       
+                       for(int j=0;j<i;j++){   
                                distCalculator->calcDist(align->get(i), align->get(j));
                                double dist = distCalculator->getDist();
+
                                if(dist <= cutoff){
-                               distFile << align->get(i).getName() << ' ' << align->get(j).getName() << ' ' << dist << endl;
+                                       distFile << align->get(i).getName() << ' ' << align->get(j).getName() << ' ' << dist << endl;
+//cout << align->get(i).getName() << ' ' << align->get(j).getName() << ' ' << dist << endl;
                                }
                        
                        }
@@ -224,4 +232,35 @@ int DistanceCommand::driver(Dist* distCalculator, SequenceDB* align, int startLi
 }
 
 /**************************************************************************************************/
-
+void DistanceCommand::appendFiles(string temp, string filename) {
+       try{
+               ofstream output;
+               ifstream input;
+               
+               //open output file in append mode
+               openOutputFileAppend(filename, output);
+               
+               //open temp file for reading
+               openInputFile(temp, input);
+               
+               string line;
+               //read input file and write to output file
+               while(input.eof() != true) {
+                       getline(input, line); //getline removes the newline char
+                       if (line != "") {
+                               output << line << endl;   // Appending back newline char 
+                       }
+               }       
+                       
+               input.close();
+               output.close();
+       }
+       catch(exception& e) {
+               cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function appendFiles. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+       catch(...) {
+               cout << "An unknown error has occurred in the DistanceCommand class function appendFiles. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }       
+}

diff --git a/distancecommand.h b/distancecommand.h
index a43e3955c9ec8184040653389b6a2f61ad2e94ba..34dd65e66ee0984d298e68ba0cc587343d191d5b 100644 (file)
--- a/distancecommand.h
+++ b/distancecommand.h
@@ -16,6 +16,10 @@
 #include "validcalculator.h"
 #include "dist.h"
 #include "sequencedb.h"
+#include "readfasta.h"
+#include "readnexus.h"
+#include "readclustal.h"
+#include "readseqsphylip.h"
 
 
 using namespace std;
@@ -32,12 +36,14 @@ private:
        ValidCalculators* validCalculator;
        Dist* distCalculator;
        SequenceDB* seqDB;
+       ReadSeqs* readSeqs;
        ofstream out;
        string outputFileName;
-       string ends, distFile;
+       string ends;
        int processors;
        float cutoff;
        
+       void appendFiles(string, string);
        int driver(Dist*, SequenceDB*, int, int, string, float);
 
 };

diff --git a/eachgapdist.h b/eachgapdist.h
index 767c4bf819aa25f2bee0c4f40a99374a30bb35e5..f4f00c16d4b41632f2c435d921ddc5bb3554edad 100644 (file)
--- a/eachgapdist.h
+++ b/eachgapdist.h
@@ -22,24 +22,24 @@ public:
                int start = 0;
                
                for(int i=0; i<A.getLength(); i++){
-                       if(A.getAligned()[i] != '.' || B.getAligned()[i] != '.'){
+                       if(A.getUnaligned()[i] != '.' || B.getUnaligned()[i] != '.'){
                                start = i;
                                break;
                        }
                }
                
                for(int i=start;i<A.getLength();i++){
-                       if((A.getAligned()[i] == '-' || B.getAligned()[i] == '-') && (A.getAligned()[i] == '.' || B.getAligned()[i] == '.')){}
-                       else if(A.getAligned()[i] == '.' && B.getAligned()[i] == '.'){
+                       if((A.getUnaligned()[i] == '-' || B.getUnaligned()[i] == '-') && (A.getUnaligned()[i] == '.' || B.getUnaligned()[i] == '.')){}
+                       else if(A.getUnaligned()[i] == '.' && B.getUnaligned()[i] == '.'){
                                break;  
                        }
-                       else if(A.getAligned()[i] != '-' && B.getAligned()[i] != '-'){
-                               if(A.getAligned()[i] != B.getAligned()[i]){
+                       else if(A.getUnaligned()[i] != '-' && B.getUnaligned()[i] != '-'){
+                               if(A.getUnaligned()[i] != B.getUnaligned()[i]){
                                        diff++;
                                }
                                length++;
                        }
-                       else if(A.getAligned()[i] != '-' || B.getAligned()[i] != '-'){
+                       else if(A.getUnaligned()[i] != '-' || B.getUnaligned()[i] != '-'){
                                diff++;
                                length++;
                        }

diff --git a/eachgapignore.h b/eachgapignore.h
index 9b9d1673249d454bf1a7ce19f41d9b568568ccb6..edc1cab3e4a1bbaf006df491ce1cd79da69db046 100644 (file)
--- a/eachgapignore.h
+++ b/eachgapignore.h
@@ -25,7 +25,7 @@ public:
                int end = 0;
                
                for(int i=0;i<A.getLength();i++){
-                       if(A.getAligned()[i] == '.' || B.getAligned()[i] == '.' || A.getAligned()[i] == '-' || B.getAligned()[i] == '-'){
+                       if(A.getUnaligned()[i] == '.' || B.getUnaligned()[i] == '.' || A.getUnaligned()[i] == '-' || B.getUnaligned()[i] == '-'){
                        }
                        else{
                                start = i;
@@ -33,7 +33,7 @@ public:
                        }
                }
                for(int i=A.getLength()-1;i>=0;i--){
-                       if(A.getAligned()[i] == '.' || B.getAligned()[i] == '.' || A.getAligned()[i] == '-' || B.getAligned()[i] == '-'){
+                       if(A.getUnaligned()[i] == '.' || B.getUnaligned()[i] == '.' || A.getUnaligned()[i] == '-' || B.getUnaligned()[i] == '-'){
                        }
                        else{
                                end = i;
@@ -42,17 +42,17 @@ public:
                }
                
                for(int i=start;i<=end;i++){
-                       if(A.getAligned()[i] == '-' && B.getAligned()[i] == '-'){}
-                       else if(A.getAligned()[i] == '.' || B.getAligned()[i] == '.'){
+                       if(A.getUnaligned()[i] == '-' && B.getUnaligned()[i] == '-'){}
+                       else if(A.getUnaligned()[i] == '.' || B.getUnaligned()[i] == '.'){
                                break;  
                        }
-                       else if(A.getAligned()[i] != '-' && B.getAligned()[i] != '-'){
-                               if(A.getAligned()[i] != B.getAligned()[i]){
+                       else if(A.getUnaligned()[i] != '-' && B.getUnaligned()[i] != '-'){
+                               if(A.getUnaligned()[i] != B.getUnaligned()[i]){
                                        diff++;
                                }
                                length++;
                        }
-                       else if(A.getAligned()[i] != '-' || B.getAligned()[i] != '-'){
+                       else if(A.getUnaligned()[i] != '-' || B.getUnaligned()[i] != '-'){
                                diff++;
                                length++;
                        }

diff --git a/errorchecking.cpp b/errorchecking.cpp
index 803a83635071d01f35c964aa4a5b424895d08fd2..f1279a4aaa57df50ab3f6606da3387d89e0718c0 100644 (file)
--- a/errorchecking.cpp
+++ b/errorchecking.cpp
@@ -188,8 +188,6 @@ bool ErrorCheck::checkInput(string input) {
                        }
                }else if (commandName == "read.tree") { 
                        validateTreeFiles(); //checks the treefile and groupfile parameters
-               }else if (commandName == "read.seqs") {
-                       if ((fastafile == "") && (nexusfile == "") && (clustalfile == "") && (phylipfile == "")) { cout << "You must enter a fastafile, nexusfile, or clustalfile with the read.seqs() command." << endl; return false; }
                }else if (commandName == "deconvolute") {
                        if (fastafile == "") { cout << "You must enter a fastafile with the deconvolute() command." << endl; return false; }
                        validateReadFiles();
@@ -248,10 +246,11 @@ bool ErrorCheck::checkInput(string input) {
                        }
                }
                
-               if (commandName == "filter.seqs"){ 
-                       if ((globaldata->getFastaFile() == "") && (globaldata->getNexusFile() == "") && (globaldata->getClustalFile() == "") && (globaldata->getPhylipFile() == "")) {
+               if ((commandName == "filter.seqs") || (commandName == "distance")) { 
+                       if ((fastafile == "") && (nexusfile == "") && (clustalfile == "") && (phylipfile == "")) {
                                 cout << "You must read either a fasta, nexus, clustal, or phylip file before you can use the filter.seqs command." << endl; return false; 
                        }
+                       validateSeqsFiles();
                }
                
                if ((commandName == "bin.seqs")) { 
@@ -510,6 +509,71 @@ void ErrorCheck::validateReadPhil() {
 }
 /*******************************************************/
 
+/******************************************************/
+//This function checks to make sure the user entered appropriate
+// format parameters on a distfile read
+void ErrorCheck::validateSeqsFiles() {
+       try {
+               ifstream filehandle;
+               int ableToOpen;
+               
+               //checks to make sure only one file type is given
+               if (phylipfile != "") { 
+                       if ((nexusfile != "") || (fastafile != "") || (clustalfile != "")) { 
+                               cout << "You may enter ONLY ONE of the following: phylip, fasta, nexus or clustal." << endl; errorFree = false; }
+                       else {
+                               ableToOpen = openInputFile(phylipfile, filehandle);
+                               filehandle.close();
+                               if (ableToOpen == 1) { //unable to open
+                                       errorFree = false;
+                               }
+                       }
+               }else if (nexusfile != "") { 
+                       if ((phylipfile != "") || (fastafile != "") || (clustalfile != "")) { 
+                               cout << "You may enter ONLY ONE of the following: phylip, fasta, nexus or clustal." << endl; errorFree = false; }
+                       else {
+                               ableToOpen = openInputFile(nexusfile, filehandle);
+                               filehandle.close();
+                               if (ableToOpen == 1) { //unable to open
+                                       errorFree = false;
+                               }
+                       }
+               }else if (fastafile != "") { 
+                       if ((phylipfile != "") || (nexusfile != "") || (clustalfile != "")) { 
+                               cout << "You may enter ONLY ONE of the following: phylip, fasta, nexus or clustal." << endl; errorFree = false; }
+                       else {
+                               ableToOpen = openInputFile(fastafile, filehandle);
+                               filehandle.close();
+                               if (ableToOpen == 1) { //unable to open
+                                       errorFree = false;
+                               }
+                       }
+               }else if (clustalfile != "") { 
+                       if ((phylipfile != "") || (nexusfile != "") || (fastafile != "")) { 
+                               cout << "You may enter ONLY ONE of the following: phylip, fasta, nexus or clustal." << endl; errorFree = false; }
+                       else {
+                               ableToOpen = openInputFile(clustalfile, filehandle);
+                               filehandle.close();
+                               if (ableToOpen == 1) { //unable to open
+                                       errorFree = false;
+                               }
+                       }
+
+               }
+               
+       }
+       catch(exception& e) {
+               cout << "Standard Error: " << e.what() << " has occurred in the ErrorCheck class Function validateSeqsFiles. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+       catch(...) {
+               cout << "An unknown error has occurred in the ErrorCheck class function validateSeqsFiles. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+}
+
+/*******************************************************/
+
 /******************************************************/
 //This function checks to make sure the user entered appropriate
 // format parameters on a bin.seq command

diff --git a/errorchecking.h b/errorchecking.h
index 0f6dbb590bd4c66873697e12eb920900cb973f7e..8de923499a5f84d520063d337c4be688a57d21ad 100644 (file)
--- a/errorchecking.h
+++ b/errorchecking.h
@@ -31,6 +31,7 @@ class ErrorCheck {
                void validateParseFiles();
                void validateTreeFiles();
                void validateBinFiles();
+               void validateSeqsFiles();
                void clear();
                void refresh();
                string phylipfile, columnfile, listfile, rabundfile, sabundfile, namefile, groupfile, orderfile, fastafile, nexusfile, clustalfile, treefile, sharedfile, cutoff, format; 

diff --git a/filterseqscommand.cpp b/filterseqscommand.cpp
index dec53a5b6f72d5f365295cfc71386bbe7797d0b2..6170d0043252b8283728db25b59a4adc5b7930b2 100644 (file)
--- a/filterseqscommand.cpp
+++ b/filterseqscommand.cpp
@@ -8,15 +8,13 @@
  */
 
 #include "filterseqscommand.h"
-#include <iostream>
-#include <fstream>
 
 /**************************************************************************************/
 void FilterSeqsCommand::doTrump() {
        trump = globaldata->getTrump();
        for(int i = 0; i < db->size(); i++) {
                Sequence cur = db->get(i);
-               string curAligned = cur.getAligned();
+               string curAligned = cur.getUnaligned();
                for(int j = 0; j < curAligned.length(); j++) {
                        string curChar = curAligned.substr(j, 1);
                        if(curChar.compare(trump) == 0) 
@@ -39,7 +37,7 @@ void FilterSeqsCommand::doSoft() {
        
        for(int i = 0; i < db->size(); i++) {
                Sequence cur = db->get(i);
-               string curAligned = cur.getAligned();
+               string curAligned = cur.getUnaligned();
                
                for(int j = 0; j < curAligned.length(); j++) {
                        string curChar = curAligned.substr(j, 1);
@@ -99,8 +97,24 @@ void FilterSeqsCommand::doFilter() {
 int FilterSeqsCommand::execute() {     
        try {
                globaldata = GlobalData::getInstance();
-               db = globaldata->gSequenceDB;
+               filename = globaldata->inputFileName;
                
+               if(globaldata->getFastaFile() != "") {
+                       readSeqs =  new ReadFasta(filename); }
+               else if(globaldata->getNexusFile() != "") {
+                       readSeqs = new ReadNexus(filename); }
+               else if(globaldata->getClustalFile() != "") {
+                       readSeqs = new ReadClustal(filename); }
+               else if(globaldata->getPhylipFile() != "") {
+                       readSeqs = new ReadPhylip(filename); }
+                       
+               readSeqs->read();
+               db = readSeqs->getDB();
+               
+               //for(int i = 0; i < db->size(); i++) {
+//                     cout << db->get(i).getLength() << "\n" << db->get(i).getName() << ": " << db->get(i).getUnaligned() << "\n\n";
+//             }
+
                for(int i = 0; i < db->get(0).getLength(); i++) 
                        columnsToRemove.push_back(false);
                
@@ -109,7 +123,6 @@ int FilterSeqsCommand::execute() {
                        doTrump();
                else if(globaldata->getSoft().compare("") != 0)
                        doSoft();
-                       
                else if(globaldata->getFilter().compare("") != 0) 
                        doFilter();
                
@@ -121,11 +134,13 @@ int FilterSeqsCommand::execute() {
 //                     else
 //                             cout << "false\n";
 //             }
+
+
                //Creating the new SequenceDB 
                SequenceDB newDB;
                for(int i = 0; i < db->size(); i++) {
                        Sequence curSeq = db->get(i);
-                       string curAligned = curSeq.getAligned();
+                       string curAligned = curSeq.getUnaligned();
                        string curName = curSeq.getName();
                        string newAligned = "";
                        for(int j = 0; j < curAligned.length(); j++) 
@@ -136,11 +151,16 @@ int FilterSeqsCommand::execute() {
                        newDB.add(newSeq);
                }
                
+               string newFileName = getRootName(filename) + "filter.fa";
                ofstream outfile;
-               outfile.open("filtertest.txt");
+               outfile.open(newFileName.c_str());
                newDB.print(outfile);
                outfile.close();
                        
+               globaldata->clear();
+               //delete db;
+               //delete newDB;
+               
                return 0;
        }
        catch(exception& e) {
@@ -151,5 +171,4 @@ int FilterSeqsCommand::execute() {
                cout << "An unknown error has occurred in the FilterSeqsCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
                exit(1);
        }
-}
-/**************************************************************************************/
+}
\ No newline at end of file

diff --git a/filterseqscommand.h b/filterseqscommand.h
index 9e703ceeb62c2d610776c7f245980de651ab28a1..06eaa6db7e2deb8b74344d7f408f09734cbf43cf 100644 (file)
--- a/filterseqscommand.h
+++ b/filterseqscommand.h
@@ -35,10 +35,7 @@ private:
        GlobalData* globaldata;
        string filename, trump, filter;
        
-       ReadFasta* readFasta;
-       ReadNexus* readNexus;
-       ReadClustal* readClustal;
-       ReadPhylip* readPhylip;
+       ReadSeqs* readSeqs;
        
        vector<bool> columnsToRemove;
        SequenceDB* db;

diff --git a/ignoregaps.h b/ignoregaps.h
index 8127b6205e42d18516561565e8a4eb29ccd56ab6..aa106c9b1bbc6bcc1cde4be9dd02fb5876b3ef85 100644 (file)
--- a/ignoregaps.h
+++ b/ignoregaps.h
@@ -27,7 +27,7 @@ public:
                int end = 0;
                
                for(int i=0;i<A.getLength();i++){
-                       if(A.getAligned()[i] == '.' || B.getAligned()[i] == '.' || A.getAligned()[i] == '-' || B.getAligned()[i] == '-'){
+                       if(A.getUnaligned()[i] == '.' || B.getUnaligned()[i] == '.' || A.getUnaligned()[i] == '-' || B.getUnaligned()[i] == '-'){
                        }
                        else{
                                start = i;
@@ -35,7 +35,7 @@ public:
                        }
                }
                for(int i=A.getLength()-1;i>=0;i--){
-                       if(A.getAligned()[i] == '.' || B.getAligned()[i] == '.' || A.getAligned()[i] == '-' || B.getAligned()[i] == '-'){
+                       if(A.getUnaligned()[i] == '.' || B.getUnaligned()[i] == '.' || A.getUnaligned()[i] == '-' || B.getUnaligned()[i] == '-'){
                        }
                        else{
                                end = i;
@@ -44,11 +44,11 @@ public:
                }
                
                for(int i=start; i<=end; i++){
-                       if(A.getAligned()[i] == '.' || B.getAligned()[i] == '.'){
+                       if(A.getUnaligned()[i] == '.' || B.getUnaligned()[i] == '.'){
                                break;
                        }
-                       else if((A.getAligned()[i] != '-' && B.getAligned()[i] != '-')){
-                               if(A.getAligned()[i] != B.getAligned()[i]){
+                       else if((A.getUnaligned()[i] != '-' && B.getUnaligned()[i] != '-')){
+                               if(A.getUnaligned()[i] != B.getUnaligned()[i]){
                                        diff++;
                                }
                                length++;

diff --git a/mothur.h b/mothur.h
index e4a9bfc69e23ee143fab64726e4631c0cd95580c..2c9595ebe53c4055179f2297372bb1e691fff8ab 100644 (file)
--- a/mothur.h
+++ b/mothur.h
@@ -273,6 +273,21 @@ inline int openOutputFile(string fileName, ofstream& fileHandle){
        }
 
 }
+/***********************************************************************/
+
+inline int openOutputFileAppend(string fileName, ofstream& fileHandle){
+       
+       fileHandle.open(fileName.c_str(), ios::app);
+       if(!fileHandle) {
+               cerr << "Error: Could not open " << fileName << endl;
+               return 1;
+       }
+       else {
+               return 0;
+       }
+
+}
+
 
 /***********************************************************************/
 

diff --git a/onegapdist.h b/onegapdist.h
index b6b257cb2f452027f19ff06e199f432076a2ef5b..ef290fe22b91f0ac1e1fbc9a29bc86e86eea8511 100644 (file)
--- a/onegapdist.h
+++ b/onegapdist.h
@@ -26,7 +26,7 @@ public:
                int start = 0;
                
                for(int i=0;i<A.getLength();i++){
-                       if(A.getAligned()[i] == '.' && B.getAligned()[i] == '.'){
+                       if(A.getUnaligned()[i] == '.' && B.getUnaligned()[i] == '.'){
                        }
                        else{
                                start = i;
@@ -34,8 +34,8 @@ public:
                        }
                }
                for(int i=start;i<A.getLength();i++){
-                       if((A.getAligned()[i] == '-' || B.getAligned()[i] == '-') && (A.getAligned()[i] == '.' || B.getAligned()[i] == '.')){}
-                       else if(A.getAligned()[i] == '-' && B.getAligned()[i] != '-' && B.getAligned()[i] != '.'){
+                       if((A.getUnaligned()[i] == '-' || B.getUnaligned()[i] == '-') && (A.getUnaligned()[i] == '.' || B.getUnaligned()[i] == '.')){}
+                       else if(A.getUnaligned()[i] == '-' && B.getUnaligned()[i] != '-' && B.getUnaligned()[i] != '.'){
                                if(openGapA == 0){
                                        difference++;
                                        minLength++;
@@ -43,7 +43,7 @@ public:
                                        openGapB = 0;
                                }
                        }
-                       else if(A.getAligned()[i] != '-' && B.getAligned()[i] == '-' && A.getAligned()[i] != '.'){
+                       else if(A.getUnaligned()[i] != '-' && B.getUnaligned()[i] == '-' && A.getUnaligned()[i] != '.'){
                                if(openGapB == 0){
                                        difference++;
                                        minLength++;
@@ -51,8 +51,8 @@ public:
                                        openGapB = 1;
                                }
                        }
-                       else if(A.getAligned()[i] != '-' && B.getAligned()[i] != '-'){
-                               if(A.getAligned()[i] != B.getAligned()[i]){
+                       else if(A.getUnaligned()[i] != '-' && B.getUnaligned()[i] != '-'){
+                               if(A.getUnaligned()[i] != B.getUnaligned()[i]){
                                        difference++;
                                        minLength++;
                                        openGapA = 0;
@@ -64,7 +64,7 @@ public:
                                        openGapB = 0;
                                }
                        }
-                       else if(A.getAligned()[i] == '.' && B.getAligned()[i] == '.'){
+                       else if(A.getUnaligned()[i] == '.' && B.getUnaligned()[i] == '.'){
                                break;
                        }
                }

diff --git a/onegapignore.h b/onegapignore.h
index ee6cc5267931816a654c99f4e748d19bb2940901..e280ce0ea45e3c828205a5d8872a47cacf8fcc02 100644 (file)
--- a/onegapignore.h
+++ b/onegapignore.h
@@ -26,7 +26,7 @@ public:
                int end = 0;
                
                for(int i=0;i<A.getLength();i++){
-                       if(A.getAligned()[i] == '.' || B.getAligned()[i] == '.' || A.getAligned()[i] == '-' || B.getAligned()[i] == '-'){
+                       if(A.getUnaligned()[i] == '.' || B.getUnaligned()[i] == '.' || A.getUnaligned()[i] == '-' || B.getUnaligned()[i] == '-'){
                        }
                        else{
                                start = i;
@@ -34,7 +34,7 @@ public:
                        }
                }
                for(int i=A.getLength()-1;i>=0;i--){
-                       if(A.getAligned()[i] == '.' || B.getAligned()[i] == '.' || A.getAligned()[i] == '-' || B.getAligned()[i] == '-'){
+                       if(A.getUnaligned()[i] == '.' || B.getUnaligned()[i] == '.' || A.getUnaligned()[i] == '-' || B.getUnaligned()[i] == '-'){
                        }
                        else{
                                end = i;
@@ -44,8 +44,8 @@ public:
                
                
                for(int i=start;i<=end;i++){
-                       if(A.getAligned()[i] == '-' && B.getAligned()[i] == '-'){}
-                       else if(A.getAligned()[i] == '-' && B.getAligned()[i] != '-'){
+                       if(A.getUnaligned()[i] == '-' && B.getUnaligned()[i] == '-'){}
+                       else if(A.getUnaligned()[i] == '-' && B.getUnaligned()[i] != '-'){
                                if(openGapA == 0){
                                        difference++;
                                        minLength++;
@@ -53,7 +53,7 @@ public:
                                        openGapB = 0;
                                }
                        }
-                       else if(A.getAligned()[i] != '-' && B.getAligned()[i] == '-'){
+                       else if(A.getUnaligned()[i] != '-' && B.getUnaligned()[i] == '-'){
                                if(openGapB == 0){
                                        difference++;
                                        minLength++;
@@ -61,8 +61,8 @@ public:
                                        openGapB = 1;
                                }
                        }
-                       else if(A.getAligned()[i] != '-' && B.getAligned()[i] != '-'){
-                               if(A.getAligned()[i] != B.getAligned()[i]){
+                       else if(A.getUnaligned()[i] != '-' && B.getUnaligned()[i] != '-'){
+                               if(A.getUnaligned()[i] != B.getUnaligned()[i]){
                                        difference++;
                                        minLength++;
                                        openGapA = 0;

diff --git a/readclustal.cpp b/readclustal.cpp
index 2f31539623d91656f26740d803795752e156e177..f42b3597a12d5dbe25d07f6d6007bb24de416f61 100644 (file)
--- a/readclustal.cpp
+++ b/readclustal.cpp
@@ -12,11 +12,8 @@
 #include <fstream>
 
 /*******************************************************************************/
-ReadClustal::ReadClustal(string file) {
+ReadClustal::ReadClustal(string file) : ReadSeqs(file){
        try {
-               openInputFile(file, filehandle);
-               clustalFile = file;
-               globaldata = GlobalData::getInstance();
        }
        catch(exception& e) {
                cout << "Standard Error: " << e.what() << " has occurred in the ReadTree class Function ReadTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
@@ -43,6 +40,7 @@ void ReadClustal::read() {
        
        int count = 0;
        int numSeqs = 0;
+       int lastSeqLength = 0;
        bool firstDone = false; 
        
        while(!filehandle.eof()) {
@@ -61,16 +59,20 @@ void ReadClustal::read() {
 
                if(name.find_first_of("*") == -1) {
                        filehandle >> sequence;
+                       lastSeqLength = sequence.length();
                        if(!firstDone) {
                                Sequence newSeq(name, sequence);
                                sequencedb.add(newSeq);
                        } 
                        else 
-                               sequencedb.set(count, sequencedb.get(count).getAligned() + sequence);
+                               sequencedb.set(count, sequencedb.get(count).getUnaligned() + sequence);
                                
                        count++;
                }
        }
+       if(count == 1)
+               sequencedb.set(0, sequencedb.get(0).getUnaligned().substr(0, sequencedb.get(0).getUnaligned().length() - lastSeqLength));
+               
        filehandle.close();
 }
 

diff --git a/readclustal.h b/readclustal.h
index e09e900cbc19f2439708cbd64a75f4bc54039c86..547112c0791c6e65de957187ef8bdfe1a06ad3ef 100644 (file)
--- a/readclustal.h
+++ b/readclustal.h
@@ -11,28 +11,20 @@
  */
 using namespace std;
 
+#include "readseqs.h"
 #include "globaldata.hpp"
 #include "sequencedb.h"
 #include "mothur.h"
 
 /**********************************************************************************/
 
-class ReadClustal {
+class ReadClustal : public ReadSeqs {
 
        public:
                ReadClustal(string);
                ~ReadClustal();
                void read();
-               SequenceDB* getDB();            
-       
-       private:
-               GlobalData* globaldata;
-               string clustalFile;
-               ifstream filehandle;
-               SequenceDB sequencedb;
-               int readOk; // readOk = 0 means success, readOk = 1 means error(s).
-               
-                       
+               SequenceDB* getDB();                    
 };
 
 #endif
\ No newline at end of file

diff --git a/readfasta.cpp b/readfasta.cpp
index 347432f40c0f872e9951592de90b2363d480f25c..e1f22ce228bbb729ce24a75a10d507f243d2179c 100644 (file)
--- a/readfasta.cpp
+++ b/readfasta.cpp
@@ -12,11 +12,8 @@
 #include <fstream>
 
 /*******************************************************************************/
-ReadFasta::ReadFasta(string file) {
+ReadFasta::ReadFasta(string file) : ReadSeqs(file) {
        try {
-               openInputFile(file, filehandle);
-               fastaFile = file;
-               globaldata = GlobalData::getInstance();
        }
        catch(exception& e) {
                cout << "Standard Error: " << e.what() << " has occurred in the ReadTree class Function ReadTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
@@ -38,6 +35,7 @@ void ReadFasta::read() {
        string sequence = "";
        string temp;
        int count = 0;
+       
        while(!filehandle.eof()){
                if(count == 0)
                        filehandle >> temp;
@@ -51,10 +49,16 @@ void ReadFasta::read() {
                                count++;
                        name = temp.substr(1,temp.length()-1);
                }
-               else 
+               else {
                        sequence += temp;
+               }
                
                filehandle >> temp;
+               gobble(filehandle);
+               
+               if(filehandle.eof())
+                       sequence += temp;
+                       
        }
        Sequence newSequence(name, sequence);
        sequencedb.add(newSequence);

diff --git a/readfasta.h b/readfasta.h
index 8f5d70df46e555f22dd9d19bc69e044c7edfe659..11cdab0d3a321720e31a110c8a9d9ab206efa0b3 100644 (file)
--- a/readfasta.h
+++ b/readfasta.h
@@ -12,28 +12,20 @@
 
 using namespace std;
 
+#include "readseqs.h"
 #include "globaldata.hpp"
 #include "sequencedb.h"
 #include "mothur.h"
 
 /**********************************************************************************/
 
-class ReadFasta {
+class ReadFasta : public ReadSeqs {
 
        public:
                ReadFasta(string);
                ~ReadFasta();
                void read();
                SequenceDB* getDB();            
-       
-       private:
-               GlobalData* globaldata;
-               string fastaFile;
-               ifstream filehandle;
-               SequenceDB sequencedb;
-               int readOk; // readOk = 0 means success, readOk = 1 means error(s).
-               
-                       
 };
 
 #endif
\ No newline at end of file

diff --git a/readnexus.cpp b/readnexus.cpp
index 25444d5093f637d6b0ff8a09bc8816463745299d..5dffa42ae7c8fe72105cebfe50d4187e08cc3957 100644 (file)
--- a/readnexus.cpp
+++ b/readnexus.cpp
@@ -12,11 +12,8 @@
 #include <fstream>
 
 /*******************************************************************************/
-ReadNexus::ReadNexus(string file) {
+ReadNexus::ReadNexus(string file) : ReadSeqs(file) {
        try {
-               openInputFile(file, filehandle);
-               nexusFile = file;
-               globaldata = GlobalData::getInstance();
        }
        catch(exception& e) {
                cout << "Standard Error: " << e.what() << " has occurred in the ReadTree class Function ReadTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
@@ -56,7 +53,7 @@ void ReadNexus::read() {
                                sequencedb.add(newSeq);
                        } 
                        else 
-                               sequencedb.set(count, sequencedb.get(count).getAligned() + sequence);
+                               sequencedb.set(count, sequencedb.get(count).getUnaligned() + sequence);
                        
                        count++;
                        if(count == numSeqs) {

diff --git a/readnexus.h b/readnexus.h
index f6bfcbbef72826fb76a3dd33677fadae9f2fbe26..374b82004691af279ebcd3293110afd982fef534 100644 (file)
--- a/readnexus.h
+++ b/readnexus.h
@@ -11,28 +11,20 @@
  */
 using namespace std;
 
+#include "readseqs.h"
 #include "globaldata.hpp"
 #include "sequencedb.h"
 #include "mothur.h"
 
 /**********************************************************************************/
 
-class ReadNexus {
+class ReadNexus : public ReadSeqs {
 
        public:
                ReadNexus(string);
                ~ReadNexus();
                void read();
-               SequenceDB* getDB();            
-       
-       private:
-               GlobalData* globaldata;
-               string nexusFile;
-               ifstream filehandle;
-               SequenceDB sequencedb;
-               int readOk; // readOk = 0 means success, readOk = 1 means error(s).
-               
-                       
+               SequenceDB* getDB();                            
 };
 
 #endif
\ No newline at end of file

diff --git a/readseqs.cpp b/readseqs.cpp
new file mode 100644 (file)
index 0000000..00a301d
--- /dev/null
+++ b/readseqs.cpp
@@ -0,0 +1,43 @@
+/*
+ *  readseqs.cpp
+ *  Mothur
+ *
+ *  Created by Thomas Ryabin on 5/11/09.
+ *  Copyright 2009 __MyCompanyName__. All rights reserved.
+ *
+ */
+
+#include "readseqs.h"
+#include <iostream>
+#include <fstream>
+
+
+/*******************************************************************************/
+ReadSeqs::ReadSeqs(string file) {
+       try {
+               openInputFile(file, filehandle);
+               seqFile = file;
+               globaldata = GlobalData::getInstance();
+       }
+       catch(exception& e) {
+               cout << "Standard Error: " << e.what() << " has occurred in the ReadTree class Function ReadTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+       catch(...) {
+               cout << "An unknown error has occurred in the ReadTree class function ReadTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }               
+}
+/*******************************************************************************/
+ReadSeqs::~ReadSeqs(){
+       //for(int i = 0; i < sequencedb.getNumSeqs(); i++)
+               //delete sequencedb.get(i);
+}
+/*******************************************************************************/
+void ReadSeqs::read() {
+}
+
+/*********************************************************************************/
+SequenceDB* ReadSeqs::getDB() {
+       return &sequencedb;
+}

diff --git a/readseqs.h b/readseqs.h
new file mode 100644 (file)
index 0000000..0bbc297
--- /dev/null
+++ b/readseqs.h
@@ -0,0 +1,37 @@
+#ifndef READSEQS_H
+#define READSEQS_H
+
+/*
+ *  readseqs.h
+ *  Mothur
+ *
+ *  Created by Thomas Ryabin on 5/11/09.
+ *  Copyright 2009 __MyCompanyName__. All rights reserved.
+ *
+ */
+
+using namespace std;
+
+#include "globaldata.hpp"
+#include "sequencedb.h"
+#include "mothur.h"
+
+/**********************************************************************************/
+
+class ReadSeqs {
+
+       public:
+               ReadSeqs(string);
+               ~ReadSeqs();
+               virtual void read();
+               virtual SequenceDB* getDB();            
+       
+       protected:
+               GlobalData* globaldata;
+               string seqFile;
+               ifstream filehandle;
+               SequenceDB sequencedb;
+               int readOk; // readOk = 0 means success, readOk = 1 means error(s).             
+};
+
+#endif

diff --git a/readseqscommand.cpp b/readseqscommand.cpp
deleted file mode 100644 (file)
index dee09f5..0000000
--- a/readseqscommand.cpp
+++ /dev/null
@@ -1,100 +0,0 @@
-/*
- *  readseqscommand.cpp
- *  Mothur
- *
- *  Created by Thomas Ryabin on 4/13/09.
- *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
- *
- */
-
-#include "readseqscommand.h"
-
-//**********************************************************************************************************************
-ReadSeqsCommand::ReadSeqsCommand(){
-       try {
-               globaldata = GlobalData::getInstance();
-               filename = globaldata->inputFileName;
-               if(globaldata->getFastaFile().compare("") != 0) {
-                       readFasta = new ReadFasta(filename);
-                       readFasta->read();
-                       globaldata->gSequenceDB = readFasta->getDB();
-               }
-               else if(globaldata->getNexusFile().compare("") != 0) {
-                       readNexus = new ReadNexus(filename);
-                       readNexus->read();
-                       globaldata->gSequenceDB = readNexus->getDB();
-               }
-               else if(globaldata->getClustalFile().compare("") != 0) {
-                       readClustal = new ReadClustal(filename);
-                       readClustal->read();
-                       globaldata->gSequenceDB = readClustal->getDB();
-               }
-               else if(globaldata->getPhylipFile().compare("") != 0) {
-                       readPhylip = new ReadPhylip(filename);
-                       readPhylip->read();
-                       globaldata->gSequenceDB = readPhylip->getDB();
-               }
-       }
-       catch(exception& e) {
-               cout << "Standard Error: " << e.what() << " has occurred in the ReadSeqsCommand class Function ReadSeqsCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
-               exit(1);
-       }
-       catch(...) {
-               cout << "An unknown error has occurred in the ReadSeqsCommand class function ReadSeqsCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
-               exit(1);
-       }
-}
-
-//**********************************************************************************************************************
-
-ReadSeqsCommand::~ReadSeqsCommand(){
-       //delete readFasta->getDB();
-//     delete readNexus->getDB();
-//     delete readClustal->getDB();
-//     delete readPhylip->getDB();
-}
-
-//**********************************************************************************************************************
-
-int ReadSeqsCommand::execute(){
-       try {
-               filebuf fb;
-               
-               //fb.open ("fasta.txt",ios::out);
-               //readFasta->read();
-//             SequenceDB* db = readFasta->getDB();
-               
-               //fb.open("nexus.txt",ios::out);
-//             readNexus->read();
-//             SequenceDB* db = readNexus->getDB();    
-
-               //fb.open("clustal.txt",ios::out);
-//             readClustal->read();
-//             SequenceDB* db = readClustal->getDB();
-               
-               //fb.open("phylip.txt",ios::out);
-//             readPhylip->read();
-//             SequenceDB* db = readPhylip->getDB();
-
-               
-               
-               //for(int i = 0; i < db->size(); i++) {
-//                     cout << db->get(i).getLength() << "\n" << db->get(i).getName() << ": " << db->get(i).getAligned() << "\n\n";
-//             }
-
-               //ostream os(&fb);
-//             db->print(os);
-//             fb.close();
-
-               return 0;
-       }
-       catch(exception& e) {
-               cout << "Standard Error: " << e.what() << " has occurred in the ReadOtuCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
-               exit(1);
-       }
-       catch(...) {
-               cout << "An unknown error has occurred in the ReadOtuCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
-               exit(1);
-       }
-}
-//**********************************************************************************************************************

diff --git a/readseqscommand.h b/readseqscommand.h
deleted file mode 100644 (file)
index 52c8a7c..0000000
--- a/readseqscommand.h
+++ /dev/null
@@ -1,35 +0,0 @@
-#ifndef READSEQSCOMMAND_H
-#define READSEQSCOMMAND_H
-/*
- *  readseqscommand.h
- *  Mothur
- *
- *  Created by Thomas Ryabin on 4/13/09.
- *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
- *
- */
-
-#include "command.hpp"
-#include "readfasta.h"
-#include "readnexus.h"
-#include "readclustal.h"
-#include "readseqsphylip.h"
-
-class GlobalData;
-
-class ReadSeqsCommand : public Command {
-public:
-       ReadSeqsCommand();
-       ~ReadSeqsCommand();
-       int execute();
-       
-private:
-       GlobalData* globaldata;
-       ReadFasta* readFasta;
-       ReadNexus* readNexus;
-       ReadClustal* readClustal;
-       ReadPhylip* readPhylip;
-       string filename;
-};
-
-#endif

diff --git a/readseqsphylip.cpp b/readseqsphylip.cpp
index 40f56a0c7efe29962418b4cb056ef51374517867..fbb99c9b0d44daddf431e1fc11be168bdcc69393 100644 (file)
--- a/readseqsphylip.cpp
+++ b/readseqsphylip.cpp
@@ -31,11 +31,8 @@ bool ReadPhylip::isSeq(string seq) {
 }
 
 /*******************************************************************************/
-ReadPhylip::ReadPhylip(string file) {
+ReadPhylip::ReadPhylip(string file) : ReadSeqs(file) {
        try {
-               openInputFile(file, filehandle);
-               phylipFile = file;
-               globaldata = GlobalData::getInstance();
        }
        catch(exception& e) {
                cout << "Standard Error: " << e.what() << " has occurred in the ReadTree class Function ReadTree. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
@@ -107,7 +104,7 @@ void ReadPhylip::read() {
                                }
                        }
                        if(!(last && count == 0))
-                               sequencedb.set(count, sequencedb.get(count).getAligned() + sequence);
+                               sequencedb.set(count, sequencedb.get(count).getUnaligned() + sequence);
                        if(letterCount == numLetters && count == 0)
                                last = true;
                }

diff --git a/readseqsphylip.h b/readseqsphylip.h
index 13c0202dd48eba1c692658e4219abe70d0525eea..be4168d7d7a14fb98254ca9d70826d860bebd08e 100644 (file)
--- a/readseqsphylip.h
+++ b/readseqsphylip.h
@@ -11,13 +11,14 @@
  */
 using namespace std;
 
+#include "readseqs.h"
 #include "globaldata.hpp"
 #include "sequencedb.h"
 #include "mothur.h"
 
 /**********************************************************************************/
 
-class ReadPhylip {
+class ReadPhylip : public ReadSeqs {
 
        public:
                ReadPhylip(string);
@@ -26,12 +27,6 @@ class ReadPhylip {
                SequenceDB* getDB();            
        
        private:
-               GlobalData* globaldata;
-               string phylipFile;
-               ifstream filehandle;
-               SequenceDB sequencedb;
-               int readOk; // readOk = 0 means success, readOk = 1 means error(s).
-               
                bool isSeq(string);
 };
 

diff --git a/sequence.cpp b/sequence.cpp
index 5b3b01d445dde50d485ca9c22ff2af843a2d79ce..b59363e13cc92f969145434d1ba8c0fa64573778 100644 (file)
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -114,7 +114,7 @@ int Sequence::getLength(){
 
 //********************************************************************************************************************
 
-void Sequence::printSequence(ofstream& out){
+void Sequence::printSequence(ostream& out){
        string toPrint = unaligned;
        if(aligned.length() > unaligned.length())
                toPrint = aligned;

diff --git a/sequence.hpp b/sequence.hpp
index dea06bdca0bdcd5844c02c8e94694a3b9b8d28e4..03cbab7ffddce5531e272cf6116069edfa0ebade 100644 (file)
--- a/sequence.hpp
+++ b/sequence.hpp
@@ -33,7 +33,7 @@ public:
        string getPairwise();
        string getUnaligned();
        int getLength();
-       void printSequence(ofstream&);
+       void printSequence(ostream&);
        
 private:
        string name;

diff --git a/sequencedb.cpp b/sequencedb.cpp
index 657573e8b3d24ee40d5f7ac10270f72a792f69e0..f948bf355c84f18f7e67573ab77952d0f1e0cbf3 100644 (file)
--- a/sequencedb.cpp
+++ b/sequencedb.cpp
@@ -72,7 +72,7 @@ int SequenceDB::size() {
 
 /***********************************************************************/
 
-void SequenceDB::print(ofstream& out) {
+void SequenceDB::print(ostream& out) {
        for(int i = 0; i < data.size(); i++)
                data[i].printSequence(out);
 }

diff --git a/sequencedb.h b/sequencedb.h
index 5dcf9bd9ffb721378a2523cc447702d887dd22a8..eb89454e0235117131194e61924a91f6487c4188 100644 (file)
--- a/sequencedb.h
+++ b/sequencedb.h
@@ -36,7 +36,7 @@ public:
        void changeSize(int);      //resizes data
        void clear();              //clears data - remeber to loop through and delete the sequences inside or you will have a memory leak
        int size();                //returns datas size
-       void print(ofstream&);      //loops through data using sequence class print
+       void print(ostream&);      //loops through data using sequence class print
                
 private:
        vector<Sequence> data;

diff --git a/validparameter.cpp b/validparameter.cpp
index 3e6a9799353a78e4e87a7cd72ebc4d78a7199aa5..e97df8b0fc8e0efc9f02f6c61987f59ea1c75b89 100644 (file)
--- a/validparameter.cpp
+++ b/validparameter.cpp
@@ -220,9 +220,6 @@ void ValidParameters::initCommandParameters() {
                string readtreeArray[] = {"tree","group"};
                commandParameters["read.tree"] = addParameters(readtreeArray, sizeof(readtreeArray)/sizeof(string));
                
-               string readseqsArray[] = {"fasta","phylip","clustal","nexus","line"};
-               commandParameters["read.seqs"] = addParameters(readseqsArray, sizeof(readseqsArray)/sizeof(string));
-               
                string clusterArray[] =  {"cutoff","precision","method"};
                commandParameters["cluster"] = addParameters(clusterArray, sizeof(clusterArray)/sizeof(string));
                
@@ -271,7 +268,7 @@ void ValidParameters::initCommandParameters() {
                string heatmapArray[] =  {"groups","line","label","sorted","scale"};
                commandParameters["heatmap"] = addParameters(heatmapArray, sizeof(heatmapArray)/sizeof(string));
                
-               string filterseqsArray[] =  {"trump", "soft", "filter"};
+               string filterseqsArray[] =  {"fasta","phylip","clustal","nexus", "trump", "soft", "filter"};
                commandParameters["filter.seqs"] = addParameters(filterseqsArray, sizeof(filterseqsArray)/sizeof(string));
 
                string vennArray[] =  {"groups","line","label","calc"};
@@ -292,7 +289,7 @@ void ValidParameters::initCommandParameters() {
                string concensusArray[] =  {};
                commandParameters["concensus"] = addParameters(concensusArray, sizeof(concensusArray)/sizeof(string));
                
-               string distanceArray[] =  {"calc", "ends", "cutoff", "processors"};
+               string distanceArray[] =  {"fasta","phylip","clustal","nexus", "calc", "ends", "cutoff", "processors"};
                commandParameters["distance"] = addParameters(distanceArray, sizeof(distanceArray)/sizeof(string));
                
                string quitArray[] = {};