else {
//valid paramters for this command
- string Array[] = {"fasta", "output", "calc", "countends", "cutoff", "processors", "outputdir","inputdir"};
+ string Array[] = {"fasta","oldfasta","column", "output", "calc", "countends", "cutoff", "processors", "outputdir","inputdir"};
vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
OptionParser parser(option);
map<string, string> parameters = parser.getParameters();
- ValidParameters validParameter;
+ ValidParameters validParameter("dist.seqs");
map<string, string>::iterator it2;
//check to make sure all parameters are valid for command
it2 = parameters.find("fasta");
//user has given a template file
if(it2 != parameters.end()){
- path = hasPath(it2->second);
+ path = m->hasPath(it2->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["fasta"] = inputDir + it2->second; }
}
+
+ it2 = parameters.find("oldfasta");
+ //user has given a template file
+ if(it2 != parameters.end()){
+ path = m->hasPath(it2->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["oldfasta"] = inputDir + it2->second; }
+ }
+
+ it2 = parameters.find("column");
+ //user has given a template file
+ if(it2 != parameters.end()){
+ path = m->hasPath(it2->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["column"] = inputDir + it2->second; }
+ }
}
//check for required parameters
else if (fastafile == "not open") { abort = true; }
else{
ifstream inFASTA;
- openInputFile(fastafile, inFASTA);
+ m->openInputFile(fastafile, inFASTA);
alignDB = SequenceDB(inFASTA);
inFASTA.close();
}
+ oldfastafile = validParameter.validFile(parameters, "oldfasta", true);
+ if (oldfastafile == "not found") { oldfastafile = ""; }
+ else if (oldfastafile == "not open") { abort = true; }
+
+ column = validParameter.validFile(parameters, "column", true);
+ if (column == "not found") { column = ""; }
+ else if (column == "not open") { abort = true; }
+
//if the user changes the output directory command factory will send this info to us in the output parameter
outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
outputDir = "";
- outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
+ outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
}
//check for optional parameter and set defaults
else {
if (calc == "default") { calc = "onegap"; }
}
- splitAtDash(calc, Estimators);
+ m->splitAtDash(calc, Estimators);
string temp;
temp = validParameter.validFile(parameters, "countends", false); if(temp == "not found"){ temp = "T"; }
output = validParameter.validFile(parameters, "output", false); if(output == "not found"){ output = "column"; }
+ if (((column != "") && (oldfastafile == "")) || ((column == "") && (oldfastafile != ""))) { m->mothurOut("If you provide column or oldfasta, you must provide both."); m->mothurOutEndLine(); abort=true; }
+
+ if ((column != "") && (oldfastafile != "") && (output != "column")) { m->mothurOut("You have provided column and oldfasta, indicating you want to append distances to your column file. Your output must be in column format to do so."); m->mothurOutEndLine(); abort=true; }
+
if ((output != "column") && (output != "lt") && (output != "square")) { m->mothurOut(output + " is not a valid output form. Options are column, lt and square. I will use column."); m->mothurOutEndLine(); output = "column"; }
ValidCalculators validCalculator;
- if (isTrue(countends) == true) {
+ if (m->isTrue(countends) == true) {
for (int i=0; i<Estimators.size(); i++) {
if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
if (Estimators[i] == "nogaps") { distCalculator = new ignoreGaps(); }
void DistanceCommand::help(){
try {
m->mothurOut("The dist.seqs command reads a file containing sequences and creates a distance file.\n");
- m->mothurOut("The dist.seqs command parameters are fasta, calc, countends, output, cutoff and processors. \n");
+ m->mothurOut("The dist.seqs command parameters are fasta, oldfasta, column, calc, countends, output, cutoff and processors. \n");
m->mothurOut("The fasta parameter is required.\n");
+ m->mothurOut("The oldfasta and column parameters allow you to append the distances calculated to the column file.\n");
m->mothurOut("The calc parameter allows you to specify the method of calculating the distances. Your options are: nogaps, onegap or eachgap. The default is onegap.\n");
m->mothurOut("The countends parameter allows you to specify whether to include terminal gaps in distance. Your options are: T or F. The default is T.\n");
m->mothurOut("The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n");
if (abort == true) { return 0; }
+ int startTime = time(NULL);
+
+ //save number of new sequence
+ numNewFasta = alignDB.getNumSeqs();
+
+ //sanity check the oldfasta and column file as well as add oldfasta sequences to alignDB
+ if ((oldfastafile != "") && (column != "")) { if (!(sanityCheck())) { return 0; } }
+
+ if (m->control_pressed) { return 0; }
+
int numSeqs = alignDB.getNumSeqs();
cutoff += 0.005;
string outputFile;
if (output == "lt") { //does the user want lower triangle phylip formatted file
- outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "phylip.dist";
+ outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "phylip.dist";
remove(outputFile.c_str());
//output numSeqs to phylip formatted dist file
}else if (output == "column") { //user wants column format
- outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "dist";
+ outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "dist";
+
+ //so we don't accidentally overwrite
+ if (outputFile == column) {
+ string tempcolumn = column + ".old";
+ rename(column.c_str(), tempcolumn.c_str());
+ }
+
remove(outputFile.c_str());
}else { //assume square
- outputFile = outputDir + getRootName(getSimpleName(fastafile)) + "square.dist";
+ outputFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "square.dist";
remove(outputFile.c_str());
}
//each process gets where it should start and stop in the file
start = int (sqrt(float(pid)/float(processors)) * numSeqs);
end = int (sqrt(float(pid+1)/float(processors)) * numSeqs);
-
- MPI_File outMPI;
- int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
-
- char filename[outputFile.length()];
- strcpy(filename, outputFile.c_str());
-
- MPI_File_open(MPI_COMM_WORLD, filename, amode, MPI_INFO_NULL, &outMPI);
- if (pid == 0) { //you are the root process
-
- //do your part
- string outputMyPart;
- driverMPI(start, end, outMPI, cutoff);
+ if (output != "lt") {
+ MPI_File outMPI;
+ int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
+
+ //char* filename = new char[outputFile.length()];
+ //memcpy(filename, outputFile.c_str(), outputFile.length());
+
+ char filename[1024];
+ strcpy(filename, outputFile.c_str());
+
+ MPI_File_open(MPI_COMM_WORLD, filename, amode, MPI_INFO_NULL, &outMPI);
+ //delete filename;
+
+ if (pid == 0) { //you are the root process
+
+ //do your part
+ string outputMyPart;
+ driverMPI(start, end, outMPI, cutoff);
+
+ if (m->control_pressed) { MPI_File_close(&outMPI); delete distCalculator; return 0; }
+
+ //wait on chidren
+ for(int i = 1; i < processors; i++) {
+ if (m->control_pressed) { MPI_File_close(&outMPI); delete distCalculator; return 0; }
+
+ char buf[4];
+ MPI_Recv(buf, 4, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
+ }
+ }else { //you are a child process
+ //do your part
+ driverMPI(start, end, outMPI, cutoff);
+
+ if (m->control_pressed) { MPI_File_close(&outMPI); delete distCalculator; return 0; }
- //wait on chidren
- for(int i = 1; i < processors; i++) {
char buf[4];
- MPI_Recv(buf, 4, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
+ strcpy(buf, "done");
+ //tell parent you are done.
+ MPI_Send(buf, 4, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
}
- if (output == "lt") {
- convertToLowerTriangle(outputFile);
- }
+ MPI_File_close(&outMPI);
- }else { //you are a child process
- //do your part
- driverMPI(start, end, outMPI, cutoff);
-
- char buf[4];
- strcpy(buf, "done");
+ }else { //lower triangle format
+ if (pid == 0) { //you are the root process
- //tell parent you are done.
- MPI_Send(buf, 4, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
+ //do your part
+ string outputMyPart;
+ long mySize;
+ driverMPI(start, end, outputFile, mySize);
+
+ if (m->control_pressed) { delete distCalculator; return 0; }
+
+ int amode=MPI_MODE_APPEND|MPI_MODE_WRONLY|MPI_MODE_CREATE; //
+ MPI_File outMPI;
+ MPI_File inMPI;
+
+ //char* filename = new char[outputFile.length()];
+ //memcpy(filename, outputFile.c_str(), outputFile.length());
+
+ char filename[1024];
+ strcpy(filename, outputFile.c_str());
+
+ MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
+ //delete filename;
+
+ //wait on chidren
+ for(int b = 1; b < processors; b++) {
+ long fileSize;
+
+ if (m->control_pressed) { MPI_File_close(&outMPI); delete distCalculator; return 0; }
+
+ MPI_Recv(&fileSize, 1, MPI_LONG, b, tag, MPI_COMM_WORLD, &status);
+
+ string outTemp = outputFile + toString(b) + ".temp";
+
+ char* buf = new char[outTemp.length()];
+ memcpy(buf, outTemp.c_str(), outTemp.length());
+
+ MPI_File_open(MPI_COMM_SELF, buf, MPI_MODE_DELETE_ON_CLOSE|MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI);
+ delete buf;
+
+ int count = 0;
+ while (count < fileSize) {
+ char buf2[1];
+ MPI_File_read(inMPI, buf2, 1, MPI_CHAR, &status);
+ MPI_File_write(outMPI, buf2, 1, MPI_CHAR, &status);
+ count += 1;
+ }
+
+ MPI_File_close(&inMPI); //deleted on close
+ }
+
+ MPI_File_close(&outMPI);
+ }else { //you are a child process
+ //do your part
+ long size;
+ driverMPI(start, end, (outputFile + toString(pid) + ".temp"), size);
+
+ if (m->control_pressed) { delete distCalculator; return 0; }
+
+ //tell parent you are done.
+ MPI_Send(&size, 1, MPI_LONG, 0, tag, MPI_COMM_WORLD);
+ }
}
-
- MPI_File_close(&outMPI);
+ MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
//append and remove temp files
for (; it != processIDS.end(); it++) {
- appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
+ m->appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
remove((outputFile + toString(it->second) + ".temp").c_str());
}
}
if (output == "square") { convertMatrix(outputFile); }
+ ifstream fileHandle;
+ fileHandle.open(outputFile.c_str());
+ if(fileHandle) {
+ m->gobble(fileHandle);
+ if (fileHandle.eof()) { m->mothurOut(outputFile + " is blank. This can result if there are no distances below your cutoff."); m->mothurOutEndLine(); }
+ }
+
+ //append the old column file to the new one
+ if ((oldfastafile != "") && (column != "")) {
+ //we had to rename the column file so we didnt overwrite above, but we want to keep old name
+ if (outputFile == column) {
+ string tempcolumn = column + ".old";
+ m->appendFiles(tempcolumn, outputFile);
+ remove(tempcolumn.c_str());
+ }else{
+ m->appendFiles(outputFile, column);
+ remove(outputFile.c_str());
+ outputFile = column;
+ }
+ }
+
+
#ifdef USE_MPI
}
#endif
m->mothurOut("Output File Name: "); m->mothurOutEndLine();
m->mothurOut(outputFile); m->mothurOutEndLine();
m->mothurOutEndLine();
-
+ m->mothurOut("It took " + toString(time(NULL) - startTime) + " to calculate the distances for " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
return 0;
}
if (m->control_pressed) { outFile.close(); return 0; }
+ //if there was a column file given and we are appending, we don't want to calculate the distances that are already in the column file
+ //the alignDB contains the new sequences and then the old, so if i an oldsequence and j is an old sequence then break out of this loop
+ if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
+
distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
double dist = distCalculator->getDist();
exit(1);
}
}
+#ifdef USE_MPI
/**************************************************************************************************/
/////// need to fix to work with calcs and sequencedb
int DistanceCommand::driverMPI(int startLine, int endLine, MPI_File& outMPI, float cutoff){
if (m->control_pressed) { return 0; }
+ //if there was a column file given and we are appending, we don't want to calculate the distances that are already in the column file
+ //the alignDB contains the new sequences and then the old, so if i an oldsequence and j is an old sequence then break out of this loop
+ if ((i >= numNewFasta) && (j >= numNewFasta)) { break; }
+
distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
double dist = distCalculator->getDist();
if (output == "column") { outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n'); }
}
- if ((output == "square") || (output == "lt")){ //make a square column you can convert to square phylip
+ if (output == "square") { //make a square column you can convert to square phylip
outputString += (alignDB.get(i).getName() + ' ' + alignDB.get(j).getName() + ' ' + toString(dist) + '\n');
outputString += (alignDB.get(j).getName() + ' ' + alignDB.get(i).getName() + ' ' + toString(dist) + '\n');
}
-
}
if(i % 100 == 0){
- m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ cout << i << '\t' << (time(NULL) - startTime) << endl;
}
- if(i % 10 == 0){ //output to file
- //send results to parent
- int length = outputString.length();
- char buf[length];
- strcpy(buf, outputString.c_str());
+
+ //send results to parent
+ int length = outputString.length();
+
+ char* buf = new char[length];
+ memcpy(buf, outputString.c_str(), length);
+
+ MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
+ outputString = "";
+ delete buf;
+
+ }
+
+ //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
+ return 1;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "DistanceCommand", "driverMPI");
+ exit(1);
+ }
+}
+/**************************************************************************************************/
+/////// need to fix to work with calcs and sequencedb
+int DistanceCommand::driverMPI(int startLine, int endLine, string file, long& size){
+ try {
+ MPI_Status status;
+
+ MPI_File outMPI;
+ int amode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
+
+ //char* filename = new char[file.length()];
+ //memcpy(filename, file.c_str(), file.length());
+
+ char filename[1024];
+ strcpy(filename, file.c_str());
+
+ MPI_File_open(MPI_COMM_SELF, filename, amode, MPI_INFO_NULL, &outMPI);
+ //delete filename;
+
+ int startTime = time(NULL);
+
+ string outputString = "";
+ size = 0;
+
+ if((output == "lt") && startLine == 0){ outputString += toString(alignDB.getNumSeqs()) + "\n"; }
+
+ for(int i=startLine;i<endLine;i++){
+ if(output == "lt") {
+ string name = alignDB.get(i).getName();
+ if (name.length() < 10) { //pad with spaces to make compatible
+ while (name.length() < 10) { name += " "; }
+ }
+ outputString += name + "\t";
+ }
+ for(int j=0;j<i;j++){
+
+ if (m->control_pressed) { return 0; }
+
+ distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
+ double dist = distCalculator->getDist();
- MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
- outputString = "";
+ if (output == "lt") { outputString += toString(dist) + "\t"; }
}
- }
+ if (output == "lt") { outputString += "\n"; }
+
- m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
- if(outputString != ""){ //output to file
- //send results to parent
- int length = outputString.length();
- char buf[length];
- strcpy(buf, outputString.c_str());
-
- MPI_File_write_shared(outMPI, buf, length, MPI_CHAR, &status);
- outputString = "";
+ if(i % 100 == 0){
+ //m->mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ cout << i << '\t' << (time(NULL) - startTime) << endl;
+ }
+
+
+ //send results to parent
+ int length = outputString.length();
+ char* buf = new char[length];
+ memcpy(buf, outputString.c_str(), length);
+
+ MPI_File_write(outMPI, buf, length, MPI_CHAR, &status);
+ size += outputString.length();
+ outputString = "";
+ delete buf;
}
+ //m->mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); m->mothurOutEndLine();
+ cout << (endLine-1) << '\t' << (time(NULL) - startTime) << endl;
+ MPI_File_close(&outMPI);
+
return 1;
}
catch(exception& e) {
- m->errorOut(e, "DistanceCommand", "driver");
+ m->errorOut(e, "DistanceCommand", "driverMPI");
exit(1);
}
}
-
+#endif
/**************************************************************************************************/
int DistanceCommand::convertMatrix(string outputFile) {
try{
//sort file by first column so the distances for each row are together
- string outfile = getRootName(outputFile) + "sorted.dist.temp";
+ string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
//use the unix sort
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
//output to new file distance for each row and save positions in file where new row begins
ifstream in;
- openInputFile(outfile, in);
+ m->openInputFile(outfile, in);
ofstream out;
- openOutputFile(outputFile, out);
+ m->openOutputFile(outputFile, out);
out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
rowDists[first] = 0.00; //distance to yourself is 0.0
in.seekg(0);
- //openInputFile(outfile, in);
+ //m->openInputFile(outfile, in);
while(!in.eof()) {
if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
- in >> first >> second >> dist; gobble(in);
+ in >> first >> second >> dist; m->gobble(in);
if (first != currentRow) {
//print out last row
try{
//sort file by first column so the distances for each row are together
- string outfile = getRootName(outputFile) + "sorted.dist.temp";
+ string outfile = m->getRootName(outputFile) + "sorted.dist.temp";
//use the unix sort
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
//output to new file distance for each row and save positions in file where new row begins
ifstream in;
- openInputFile(outfile, in);
+ m->openInputFile(outfile, in);
ofstream out;
- openOutputFile(outputFile, out);
+ m->openOutputFile(outputFile, out);
out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
rowDists[first] = 0.00; //distance to yourself is 0.0
in.seekg(0);
- //openInputFile(outfile, in);
+ //m->openInputFile(outfile, in);
while(!in.eof()) {
if (m->control_pressed) { in.close(); remove(outfile.c_str()); out.close(); return 0; }
- in >> first >> second >> dist; gobble(in);
+ in >> first >> second >> dist; m->gobble(in);
if (first != currentRow) {
//print out last row
exit(1);
}
}
+/**************************************************************************************************/
+//its okay if the column file does not contain all the names in the fasta file, since some distance may have been above a cutoff,
+//but no sequences can be in the column file that are not in oldfasta. also, if a distance is above the cutoff given then remove it.
+//also check to make sure the 2 files have the same alignment length.
+bool DistanceCommand::sanityCheck() {
+ try{
+ bool good = true;
+
+ //make sure the 2 fasta files have the same alignment length
+ ifstream in;
+ m->openInputFile(fastafile, in);
+ int fastaAlignLength = 0;
+ if (in) {
+ Sequence tempIn(in);
+ fastaAlignLength = tempIn.getAligned().length();
+ }
+ in.close();
+
+ ifstream in2;
+ m->openInputFile(oldfastafile, in2);
+ int oldfastaAlignLength = 0;
+ if (in2) {
+ Sequence tempIn2(in2);
+ oldfastaAlignLength = tempIn2.getAligned().length();
+ }
+ in2.close();
+
+ if (fastaAlignLength != oldfastaAlignLength) { m->mothurOut("fasta files do not have the same alignment length."); m->mothurOutEndLine(); return false; }
+
+ //read fasta file and save names as well as adding them to the alignDB
+ set<string> namesOldFasta;
+
+ ifstream inFasta;
+ m->openInputFile(oldfastafile, inFasta);
+
+ while (!inFasta.eof()) {
+ if (m->control_pressed) { inFasta.close(); return good; }
+
+ Sequence temp(inFasta);
+
+ if (temp.getName() != "") {
+ namesOldFasta.insert(temp.getName()); //save name
+ alignDB.push_back(temp); //add to DB
+ }
+
+ m->gobble(inFasta);
+ }
+
+ inFasta.close();
+
+ //read through the column file checking names and removing distances above the cutoff
+ ifstream inDist;
+ m->openInputFile(column, inDist);
+
+ ofstream outDist;
+ string outputFile = column + ".temp";
+ m->openOutputFile(outputFile, outDist);
+
+ string name1, name2;
+ float dist;
+ while (!inDist.eof()) {
+ if (m->control_pressed) { inDist.close(); outDist.close(); remove(outputFile.c_str()); return good; }
+
+ inDist >> name1 >> name2 >> dist; m->gobble(inDist);
+
+ //both names are in fasta file and distance is below cutoff
+ if ((namesOldFasta.count(name1) == 0) || (namesOldFasta.count(name2) == 0)) { good = false; break; }
+ else{
+ if (dist <= cutoff) {
+ outDist << name1 << '\t' << name2 << '\t' << dist << endl;
+ }
+ }
+ }
+
+ inDist.close();
+ outDist.close();
+
+ if (good) {
+ remove(column.c_str());
+ rename(outputFile.c_str(), column.c_str());
+ }else{
+ remove(outputFile.c_str()); //temp file is bad because file mismatch above
+ }
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "DistanceCommand", "m->appendFiles");
+ exit(1);
+ }
+}
+
/**************************************************************************************************
-void DistanceCommand::appendFiles(string temp, string filename) {
+void DistanceCommand::m->appendFiles(string temp, string filename) {
try{
ofstream output;
ifstream input;
//open output file in append mode
- openOutputFileAppend(filename, output);
- openInputFile(temp, input);
+ m->openOutputFileAppend(filename, output);
+ m->openInputFile(temp, input);
while(char c = input.get()){
if(input.eof()) { break; }
output.close();
}
catch(exception& e) {
- m->errorOut(e, "DistanceCommand", "appendFiles");
+ m->errorOut(e, "DistanceCommand", "m->appendFiles");
exit(1);
}
}
projects / mothur.git / blobdiff