#include "splitmatrix.h"
#include "phylotree.h"
-#include "sequencedb.h"
-#include "onegapdist.h"
-#include "dist.h"
+#include "distancecommand.h"
+#include "seqsummarycommand.h"
/***********************************************************************/
}
/***********************************************************************/
-SplitMatrix::SplitMatrix(string ffile, string tax, float c, string t){
+SplitMatrix::SplitMatrix(string ffile, string name, string tax, float c, float cu, string t, int p, bool cl, string output){
m = MothurOut::getInstance();
fastafile = ffile;
+ namefile = name;
taxFile = tax;
- cutoff = c;
+ cutoff = c; //tax level cutoff
+ distCutoff = cu; //for fasta method if you are creating distance matrix you need a cutoff for that
method = t;
+ processors = p;
+ classic = cl;
+ outputDir = output;
}
/***********************************************************************/
if (large) { splitDistanceLarge(); }
else { splitDistanceRAM(); }
+
+ return 0;
}
catch(exception& e) {
int SplitMatrix::splitClassify(){
try {
cutoff = int(cutoff);
-
+
map<string, int> seqGroup;
map<string, int>::iterator it;
map<string, int>::iterator it2;
//build tree from users taxonomy file
PhyloTree* phylo = new PhyloTree();
- ifstream in;
- openInputFile(taxFile, in);
-
- //read in users taxonomy file and add sequences to tree
- string seqname, tax;
- while(!in.eof()){
- in >> seqname >> tax; gobble(in);
- phylo->addSeqToTree(seqname, tax);
- }
- in.close();
+ map<string, string> temp;
+ m->readTax(taxFile, temp);
+
+ for (map<string, string>::iterator itTemp = temp.begin(); itTemp != temp.end();) {
+ phylo->addSeqToTree(itTemp->first, itTemp->second);
+ temp.erase(itTemp++);
+ }
phylo->assignHeirarchyIDs(0);
createDistanceFilesFromTax(seqGroup, numGroups);
}
-
return 0;
}
/***********************************************************************/
int SplitMatrix::createDistanceFilesFromTax(map<string, int>& seqGroup, int numGroups){
try {
+ map<string, int> copyGroups = seqGroup;
map<string, int>::iterator it;
- map<string, int>::iterator it2;
- map<string, int> seqIndexInFasta;
+ set<string> names;
- //read fastafile
- SequenceDB alignDB;
-
- ifstream filehandle;
- openInputFile(fastafile, filehandle);
- int numSeqs = 0;
- while (!filehandle.eof()) {
- //input sequence info into sequencedb
- Sequence newSequence(filehandle);
+ for (int i = 0; i < numGroups; i++) { //remove old temp files, just in case
+ m->mothurRemove((fastafile + "." + toString(i) + ".temp"));
+ }
- if (newSequence.getName() != "") {
- alignDB.push_back(newSequence);
- seqIndexInFasta[newSequence.getName()] = numSeqs;
- numSeqs++;
+ ifstream in;
+ m->openInputFile(fastafile, in);
+
+ //parse fastafile
+ ofstream outFile;
+ while (!in.eof()) {
+ Sequence query(in); m->gobble(in);
+ if (query.getName() != "") {
+
+ it = seqGroup.find(query.getName());
+
+ //save names in case no namefile is given
+ if (namefile == "") { names.insert(query.getName()); }
+
+ if (it != seqGroup.end()) { //not singleton
+ m->openOutputFileAppend((fastafile + "." + toString(it->second) + ".temp"), outFile);
+ query.printSequence(outFile);
+ outFile.close();
+
+ copyGroups.erase(query.getName());
+ }
}
+ }
+ in.close();
+
+ //warn about sequence in groups that are not in fasta file
+ for(it = copyGroups.begin(); it != copyGroups.end(); it++) {
+ m->mothurOut("ERROR: " + it->first + " is missing from your fastafile. This could happen if your taxonomy file is not unique and your fastafile is, or it could indicate and error."); m->mothurOutEndLine();
+ exit(1);
+ }
+
+ copyGroups.clear();
+
+ //process each distance file
+ for (int i = 0; i < numGroups; i++) {
+
+ string options = "";
+ if (classic) { options = "fasta=" + (fastafile + "." + toString(i) + ".temp") + ", processors=" + toString(processors) + ", output=lt"; }
+ else { options = "fasta=" + (fastafile + "." + toString(i) + ".temp") + ", processors=" + toString(processors) + ", cutoff=" + toString(distCutoff); }
+ if (outputDir != "") { options += ", outputdir=" + outputDir; }
+
+ Command* command = new DistanceCommand(options);
+
+ command->execute();
+ delete command;
- //takes care of white space
- gobble(filehandle);
+ m->mothurRemove((fastafile + "." + toString(i) + ".temp"));
+
+ //remove old names files just in case
+ m->mothurRemove((namefile + "." + toString(i) + ".temp"));
}
- filehandle.close();
+
+ singleton = namefile + ".extra.temp";
+ ofstream remainingNames;
+ m->openOutputFile(singleton, remainingNames);
- Dist* distCalculator = new oneGapDist();
+ bool wroteExtra = false;
+
+ ifstream bigNameFile;
+ m->openInputFile(namefile, bigNameFile);
+ string name, nameList;
+ while(!bigNameFile.eof()){
+ bigNameFile >> name >> nameList; m->gobble(bigNameFile);
+
+ //did this sequence get assigned a group
+ it = seqGroup.find(name);
+
+ if (it != seqGroup.end()) {
+ m->openOutputFileAppend((namefile + "." + toString(it->second) + ".temp"), outFile);
+ outFile << name << '\t' << nameList << endl;
+ outFile.close();
+ }else{
+ wroteExtra = true;
+ remainingNames << name << '\t' << nameList << endl;
+ }
+ }
+ bigNameFile.close();
-//still not done....
+ for(int i=0;i<numGroups;i++){
+ string tempNameFile = namefile + "." + toString(i) + ".temp";
+ if (outputDir == "") { outputDir = m->hasPath(fastafile); }
+ string tempDistFile = "";
+ if (classic) { tempDistFile = outputDir + m->getRootName(m->getSimpleName((fastafile + "." + toString(i) + ".temp"))) + "phylip.dist";}
+ else { tempDistFile = outputDir + m->getRootName(m->getSimpleName((fastafile + "." + toString(i) + ".temp"))) + "dist"; }
+
+ //if there are valid distances
+ ifstream fileHandle;
+ fileHandle.open(tempDistFile.c_str());
+ if(fileHandle) {
+ m->gobble(fileHandle);
+ if (!fileHandle.eof()) { //check for blank file - this could occur if all dists in group are above cutoff
+ map<string, string> temp;
+ temp[tempDistFile] = tempNameFile;
+ dists.push_back(temp);
+ }else {
+ ifstream in;
+ m->openInputFile(tempNameFile, in);
+
+ while(!in.eof()) {
+ in >> name >> nameList; m->gobble(in);
+ wroteExtra = true;
+ remainingNames << name << '\t' << nameList << endl;
+ }
+ in.close();
+ m->mothurRemove(tempNameFile);
+ }
+ }
+ fileHandle.close();
+ }
+ remainingNames.close();
+ if (!wroteExtra) {
+ m->mothurRemove(singleton);
+ singleton = "none";
+ }
+
+ if (m->control_pressed) { for (int i = 0; i < dists.size(); i++) { m->mothurRemove((dists[i].begin()->first)); m->mothurRemove((dists[i].begin()->second)); } dists.clear(); }
return 0;
}
map<string, int>::iterator it2;
ifstream dFile;
- openInputFile(distFile, dFile);
+ m->openInputFile(distFile, dFile);
ofstream outFile;
for (int i = 0; i < numGroups; i++) { //remove old temp files, just in case
- remove((distFile + "." + toString(i) + ".temp").c_str());
+ m->mothurRemove((distFile + "." + toString(i) + ".temp"));
}
//for buffering the io to improve speed
string seqA, seqB;
float dist;
- if (m->control_pressed) { dFile.close(); for (int i = 0; i < numGroups; i++) { remove((distFile + "." + toString(i) + ".temp").c_str()); } }
+ if (m->control_pressed) { dFile.close(); for (int i = 0; i < numGroups; i++) { m->mothurRemove((distFile + "." + toString(i) + ".temp")); } }
- dFile >> seqA >> seqB >> dist; gobble(dFile);
+ dFile >> seqA >> seqB >> dist; m->gobble(dFile);
//if both sequences are in the same group then they are within the cutoff
it = seqGroup.find(seqA);
if ((it != seqGroup.end()) && (it2 != seqGroup.end())) { //they are both not singletons
if (it->second == it2->second) { //they are from the same group so add the distance
if (numOutputs[it->second] > 30) {
- openOutputFileAppend((distFile + "." + toString(it->second) + ".temp"), outFile);
+ m->openOutputFileAppend((distFile + "." + toString(it->second) + ".temp"), outFile);
outFile << outputs[it->second] << seqA << '\t' << seqB << '\t' << dist << endl;
outFile.close();
outputs[it->second] = "";
dFile.close();
for (int i = 0; i < numGroups; i++) { //remove old temp files, just in case
- remove((namefile + "." + toString(i) + ".temp").c_str());
+ m->mothurRemove((namefile + "." + toString(i) + ".temp"));
//write out any remaining buffers
if (numOutputs[i] > 0) {
- openOutputFileAppend((distFile + "." + toString(i) + ".temp"), outFile);
+ m->openOutputFileAppend((distFile + "." + toString(i) + ".temp"), outFile);
outFile << outputs[i];
outFile.close();
outputs[i] = "";
}
ifstream bigNameFile;
- openInputFile(namefile, bigNameFile);
+ m->openInputFile(namefile, bigNameFile);
singleton = namefile + ".extra.temp";
ofstream remainingNames;
- openOutputFile(singleton, remainingNames);
+ m->openOutputFile(singleton, remainingNames);
bool wroteExtra = false;
string name, nameList;
while(!bigNameFile.eof()){
- bigNameFile >> name >> nameList; gobble(bigNameFile);
+ bigNameFile >> name >> nameList; m->gobble(bigNameFile);
//did this sequence get assigned a group
it = seqGroup.find(name);
if (it != seqGroup.end()) {
- openOutputFileAppend((namefile + "." + toString(it->second) + ".temp"), outFile);
+ m->openOutputFileAppend((namefile + "." + toString(it->second) + ".temp"), outFile);
outFile << name << '\t' << nameList << endl;
outFile.close();
}else{
}
}
bigNameFile.close();
- remainingNames.close();
-
- if (!wroteExtra) {
- remove(singleton.c_str());
- singleton = "none";
- }
-
+
for(int i=0;i<numGroups;i++){
+ string tempNameFile = namefile + "." + toString(i) + ".temp";
+ string tempDistFile = distFile + "." + toString(i) + ".temp";
+
//if there are valid distances
if (validDistances[i]) {
- string tempNameFile = namefile + "." + toString(i) + ".temp";
- string tempDistFile = distFile + "." + toString(i) + ".temp";
-
map<string, string> temp;
temp[tempDistFile] = tempNameFile;
dists.push_back(temp);
+ }else{
+ ifstream in;
+ m->openInputFile(tempNameFile, in);
+
+ while(!in.eof()) {
+ in >> name >> nameList; m->gobble(in);
+ wroteExtra = true;
+ remainingNames << name << '\t' << nameList << endl;
+ }
+ in.close();
+ m->mothurRemove(tempNameFile);
}
}
+ remainingNames.close();
+
+ if (!wroteExtra) {
+ m->mothurRemove(singleton);
+ singleton = "none";
+ }
+
if (m->control_pressed) {
for (int i = 0; i < dists.size(); i++) {
- remove((dists[i].begin()->first).c_str());
- remove((dists[i].begin()->second).c_str());
+ m->mothurRemove((dists[i].begin()->first));
+ m->mothurRemove((dists[i].begin()->second));
}
dists.clear();
}
ofstream outFile;
ifstream dFile;
- openInputFile(distFile, dFile);
+ m->openInputFile(distFile, dFile);
while(dFile){
string seqA, seqB;
dFile >> seqA >> seqB >> dist;
- if (m->control_pressed) { dFile.close(); for(int i=0;i<numGroups;i++){ if(groups[i].size() > 0){ remove((distFile + "." + toString(i) + ".temp").c_str()); } } return 0; }
+ if (m->control_pressed) { dFile.close(); for(int i=0;i<numGroups;i++){ if(groups[i].size() > 0){ m->mothurRemove((distFile + "." + toString(i) + ".temp")); } } return 0; }
if(dist < cutoff){
//cout << "in cutoff: " << dist << endl;
delete memblock;
fileB.close();
- remove(fileName2.c_str());
+ m->mothurRemove(fileName2);
//write out the merged memory
if (numOutputs[groupID] > 60) {
delete memblock;
fileB.close();
- remove(fileName2.c_str());
+ m->mothurRemove(fileName2);
//write out the merged memory
if (numOutputs[groupID] > 60) {
}
}
}
- gobble(dFile);
+ m->gobble(dFile);
}
dFile.close();
while(bigNameFile){
bigNameFile >> name >> nameList;
nameMap[name] = nameList;
- gobble(bigNameFile);
+ m->gobble(bigNameFile);
}
bigNameFile.close();
if (m->control_pressed) {
for (int i = 0; i < dists.size(); i++) {
- remove((dists[i].begin()->first).c_str());
- remove((dists[i].begin()->second).c_str());
+ m->mothurRemove((dists[i].begin()->first));
+ m->mothurRemove((dists[i].begin()->second));
}
dists.clear();
}
int numGroups = 0;
ifstream dFile;
- openInputFile(distFile, dFile);
+ m->openInputFile(distFile, dFile);
while(dFile){
string seqA, seqB;
dFile >> seqA >> seqB >> dist;
- if (m->control_pressed) { dFile.close(); for(int i=0;i<numGroups;i++){ if(groups[i].size() > 0){ remove((distFile + "." + toString(i) + ".temp").c_str()); } } return 0; }
+ if (m->control_pressed) { dFile.close(); for(int i=0;i<numGroups;i++){ if(groups[i].size() > 0){ m->mothurRemove((distFile + "." + toString(i) + ".temp")); } } return 0; }
if(dist < cutoff){
//cout << "in cutoff: " << dist << endl;
}
}
}
- gobble(dFile);
+ m->gobble(dFile);
}
dFile.close();
projects / mothur.git / blobdiff