#include "seqsummarycommand.h"
#include "sequence.hpp"
+//**********************************************************************************************************************
+vector<string> SeqSummaryCommand::setParameters(){
+ try {
+ CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
+ CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
+ CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+
+ vector<string> myArray;
+ for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SeqSummaryCommand", "setParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+string SeqSummaryCommand::getHelpString(){
+ try {
+ string helpString = "";
+ helpString += "The summary.seqs command reads a fastafile and summarizes the sequences.\n";
+ helpString += "The summary.seqs command parameters are fasta, name and processors, fasta is required, unless you have a valid current fasta file.\n";
+ helpString += "The name parameter allows you to enter a name file associated with your fasta file. \n";
+ helpString += "The summary.seqs command should be in the following format: \n";
+ helpString += "summary.seqs(fasta=yourFastaFile, processors=2) \n";
+ helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
+ return helpString;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SeqSummaryCommand", "getHelpString");
+ exit(1);
+ }
+}
+
+//**********************************************************************************************************************
+SeqSummaryCommand::SeqSummaryCommand(){
+ try {
+ abort = true; calledHelp = true;
+ setParameters();
+ vector<string> tempOutNames;
+ outputTypes["summary"] = tempOutNames;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SeqSummaryCommand", "SeqSummaryCommand");
+ exit(1);
+ }
+}
//***************************************************************************************************************
SeqSummaryCommand::SeqSummaryCommand(string option) {
try {
- abort = false;
+ abort = false; calledHelp = false;
//allow user to run help
- if(option == "help") { help(); abort = true; }
+ if(option == "help") { help(); abort = true; calledHelp = true; }
+ else if(option == "citation") { citation(); abort = true; calledHelp = true;}
else {
- //valid paramters for this command
- string Array[] = {"fasta","processors","outputdir","inputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ vector<string> myArray = setParameters();
OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
- ValidParameters validParameter;
+ ValidParameters validParameter("summary.seqs");
map<string,string>::iterator it;
//check to make sure all parameters are valid for command
it = parameters.find("fasta");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["fasta"] = inputDir + it->second; }
}
+
+ it = parameters.find("name");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["name"] = inputDir + it->second; }
+ }
}
+ //initialize outputTypes
+ vector<string> tempOutNames;
+ outputTypes["summary"] = tempOutNames;
+
//check for required parameters
fastafile = validParameter.validFile(parameters, "fasta", true);
if (fastafile == "not open") { abort = true; }
- else if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the summary.seqs command."); m->mothurOutEndLine(); abort = true; }
+ else if (fastafile == "not found") {
+ fastafile = m->getFastaFile();
+ if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
+ else { m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
+ }
+
+ namefile = validParameter.validFile(parameters, "name", true);
+ if (namefile == "not open") { namefile = ""; abort = true; }
+ else if (namefile == "not found") { namefile = ""; }
//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
}
- string temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; }
- convert(temp, processors);
+ string temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
+ m->setProcessors(temp);
+ convert(temp, processors);
}
exit(1);
}
}
-//**********************************************************************************************************************
-
-void SeqSummaryCommand::help(){
- try {
- m->mothurOut("The summary.seqs command reads a fastafile and summarizes the sequences.\n");
- m->mothurOut("The summary.seqs command parameters are fasta and processors, fasta is required.\n");
- m->mothurOut("The summary.seqs command should be in the following format: \n");
- m->mothurOut("summary.seqs(fasta=yourFastaFile, processors=2) \n");
- m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
- }
- catch(exception& e) {
- m->errorOut(e, "SeqSummaryCommand", "help");
- exit(1);
- }
-}
-
-//***************************************************************************************************************
-
-SeqSummaryCommand::~SeqSummaryCommand(){ /* do nothing */ }
-
//***************************************************************************************************************
int SeqSummaryCommand::execute(){
try{
- if (abort == true) { return 0; }
+ if (abort == true) { if (calledHelp) { return 0; } return 2; }
+
+ //set current fasta to fastafile
+ m->setFastaFile(fastafile);
- string summaryFile = outputDir + getSimpleName(fastafile) + ".summary";
+ string summaryFile = outputDir + m->getSimpleName(fastafile) + ".summary";
int numSeqs = 0;
vector<int> seqLength;
vector<int> ambigBases;
vector<int> longHomoPolymer;
+
+ if (namefile != "") { nameMap = m->readNames(namefile); }
+
+ if (m->control_pressed) { return 0; }
#ifdef USE_MPI
int pid, numSeqsPerProcessor;
int tag = 2001;
int startTag = 1; int endTag = 2; int lengthTag = 3; int baseTag = 4; int lhomoTag = 5;
int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
- vector<long> MPIPos;
+ vector<unsigned long int> MPIPos;
MPI_Status status;
MPI_Status statusOut;
if (pid == 0) { //you are the root process
//print header
- string outputString = "seqname\tstart\tend\tnbases\tambigs\tpolymer\n";
+ string outputString = "seqname\tstart\tend\tnbases\tambigs\tpolymer\tnumSeqs\n";
int length = outputString.length();
char* buf2 = new char[length];
memcpy(buf2, outputString.c_str(), length);
MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &statusOut);
delete buf2;
- MPIPos = setFilePosFasta(fastafile, numSeqs); //fills MPIPos, returns numSeqs
+ MPIPos = m->setFilePosFasta(fastafile, numSeqs); //fills MPIPos, returns numSeqs
for(int i = 1; i < processors; i++) {
MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
+ vector<unsigned long int> positions = m->divideFile(fastafile, processors);
+
+ for (int i = 0; i < (positions.size()-1); i++) {
+ lines.push_back(new linePair(positions[i], positions[(i+1)]));
+ }
+
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
if(processors == 1){
- ifstream inFASTA;
- openInputFile(fastafile, inFASTA);
- getNumSeqs(inFASTA, numSeqs);
- inFASTA.close();
-
- lines.push_back(new linePair(0, numSeqs));
-
- driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, summaryFile, lines[0]);
+ numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, summaryFile, lines[0]);
}else{
- numSeqs = setLines(fastafile);
- createProcessesCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, summaryFile);
+ numSeqs = createProcessesCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, summaryFile);
rename((summaryFile + toString(processIDS[0]) + ".temp").c_str(), summaryFile.c_str());
//append files
for(int i=1;i<processors;i++){
- appendFiles((summaryFile + toString(processIDS[i]) + ".temp"), summaryFile);
+ m->appendFiles((summaryFile + toString(processIDS[i]) + ".temp"), summaryFile);
remove((summaryFile + toString(processIDS[i]) + ".temp").c_str());
}
}
if (m->control_pressed) { return 0; }
#else
- ifstream inFASTA;
- openInputFile(fastafile, inFASTA);
- getNumSeqs(inFASTA, numSeqs);
- inFASTA.close();
-
- lines.push_back(new linePair(0, numSeqs));
-
- driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, summaryFile, lines[0]);
+ numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, summaryFile, lines[0]);
if (m->control_pressed) { return 0; }
#endif
#endif
sort(seqLength.begin(), seqLength.end());
sort(ambigBases.begin(), ambigBases.end());
sort(longHomoPolymer.begin(), longHomoPolymer.end());
-
- int ptile0_25 = int(numSeqs * 0.025);
- int ptile25 = int(numSeqs * 0.250);
- int ptile50 = int(numSeqs * 0.500);
- int ptile75 = int(numSeqs * 0.750);
- int ptile97_5 = int(numSeqs * 0.975);
- int ptile100 = numSeqs - 1;
+ int size = startPosition.size();
+
+ int ptile0_25 = int(size * 0.025);
+ int ptile25 = int(size * 0.250);
+ int ptile50 = int(size * 0.500);
+ int ptile75 = int(size * 0.750);
+ int ptile97_5 = int(size * 0.975);
+ int ptile100 = size - 1;
//to compensate for blank sequences that would result in startPosition and endPostion equalling -1
if (startPosition[0] == -1) { startPosition[0] = 0; }
m->mothurOut("75%-tile:\t" + toString(startPosition[ptile75]) + "\t" + toString(endPosition[ptile75]) + "\t" + toString(seqLength[ptile75]) + "\t" + toString(ambigBases[ptile75]) + "\t" + toString(longHomoPolymer[ptile75])); m->mothurOutEndLine();
m->mothurOut("97.5%-tile:\t" + toString(startPosition[ptile97_5]) + "\t" + toString(endPosition[ptile97_5]) + "\t" + toString(seqLength[ptile97_5]) + "\t" + toString(ambigBases[ptile97_5]) + "\t" + toString(longHomoPolymer[ptile97_5])); m->mothurOutEndLine();
m->mothurOut("Maximum:\t" + toString(startPosition[ptile100]) + "\t" + toString(endPosition[ptile100]) + "\t" + toString(seqLength[ptile100]) + "\t" + toString(ambigBases[ptile100]) + "\t" + toString(longHomoPolymer[ptile100])); m->mothurOutEndLine();
- m->mothurOut("# of Seqs:\t" + toString(numSeqs)); m->mothurOutEndLine();
+ if (namefile == "") { m->mothurOut("# of Seqs:\t" + toString(numSeqs)); m->mothurOutEndLine(); }
+ else { m->mothurOut("# of unique seqs:\t" + toString(numSeqs)); m->mothurOutEndLine(); m->mothurOut("total # of seqs:\t" + toString(startPosition.size())); m->mothurOutEndLine(); }
if (m->control_pressed) { remove(summaryFile.c_str()); return 0; }
m->mothurOutEndLine();
m->mothurOut("Output File Name: "); m->mothurOutEndLine();
- m->mothurOut(summaryFile); m->mothurOutEndLine();
+ m->mothurOut(summaryFile); m->mothurOutEndLine(); outputNames.push_back(summaryFile); outputTypes["summary"].push_back(summaryFile);
m->mothurOutEndLine();
#ifdef USE_MPI
}
}
/**************************************************************************************/
-int SeqSummaryCommand::driverCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename, string sumFile, linePair* line) {
+int SeqSummaryCommand::driverCreateSummary(vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, string filename, string sumFile, linePair* filePos) {
try {
ofstream outSummary;
- openOutputFile(sumFile, outSummary);
+ m->openOutputFile(sumFile, outSummary);
//print header if you are process 0
- if (line->start == 0) {
- outSummary << "seqname\tstart\tend\tnbases\tambigs\tpolymer" << endl;
+ if (filePos->start == 0) {
+ outSummary << "seqname\tstart\tend\tnbases\tambigs\tpolymer\tnumSeqs" << endl;
}
ifstream in;
- openInputFile(filename, in);
+ m->openInputFile(filename, in);
- in.seekg(line->start);
-
- for(int i=0;i<line->num;i++){
+ in.seekg(filePos->start);
+
+ bool done = false;
+ int count = 0;
+
+ while (!done) {
if (m->control_pressed) { in.close(); outSummary.close(); return 1; }
- Sequence current(in);
+ Sequence current(in); m->gobble(in);
if (current.getName() != "") {
- startPosition.push_back(current.getStartPos());
- endPosition.push_back(current.getEndPos());
- seqLength.push_back(current.getNumBases());
- ambigBases.push_back(current.getAmbigBases());
- longHomoPolymer.push_back(current.getLongHomoPolymer());
+ int num = 1;
+ if (namefile != "") {
+ //make sure this sequence is in the namefile, else error
+ map<string, int>::iterator it = nameMap.find(current.getName());
+
+ if (it == nameMap.end()) { m->mothurOut("[ERROR]: " + current.getName() + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
+ else { num = it->second; }
+ }
+
+ //for each sequence this sequence represents
+ for (int i = 0; i < num; i++) {
+ startPosition.push_back(current.getStartPos());
+ endPosition.push_back(current.getEndPos());
+ seqLength.push_back(current.getNumBases());
+ ambigBases.push_back(current.getAmbigBases());
+ longHomoPolymer.push_back(current.getLongHomoPolymer());
+ }
+
+ count++;
outSummary << current.getName() << '\t';
outSummary << current.getStartPos() << '\t' << current.getEndPos() << '\t';
outSummary << current.getNumBases() << '\t' << current.getAmbigBases() << '\t';
- outSummary << current.getLongHomoPolymer() << endl;
+ outSummary << current.getLongHomoPolymer() << '\t' << num << endl;
}
- gobble(in);
+
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ unsigned long int pos = in.tellg();
+ if ((pos == -1) || (pos >= filePos->end)) { break; }
+ #else
+ if (in.eof()) { break; }
+ #endif
+
+ //report progress
+ //if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
}
+ //report progress
+ //if((count) % 100 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
+
in.close();
- return 0;
+ return count;
}
catch(exception& e) {
m->errorOut(e, "SeqSummaryCommand", "driverCreateSummary");
}
#ifdef USE_MPI
/**************************************************************************************/
-int SeqSummaryCommand::MPICreateSummary(int start, int num, vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, MPI_File& inMPI, MPI_File& outMPI, vector<long>& MPIPos) {
+int SeqSummaryCommand::MPICreateSummary(int start, int num, vector<int>& startPosition, vector<int>& endPosition, vector<int>& seqLength, vector<int>& ambigBases, vector<int>& longHomoPolymer, MPI_File& inMPI, MPI_File& outMPI, vector<unsigned long int>& MPIPos) {
try {
int pid;
MPI_Status status;
MPI_Comm_rank(MPI_COMM_WORLD, &pid);
-
for(int i=0;i<num;i++){
if (m->control_pressed) { return 0; }
Sequence current(iss);
if (current.getName() != "") {
- startPosition.push_back(current.getStartPos());
- endPosition.push_back(current.getEndPos());
- seqLength.push_back(current.getNumBases());
- ambigBases.push_back(current.getAmbigBases());
- longHomoPolymer.push_back(current.getLongHomoPolymer());
+
+ int num = 1;
+ if (namefile != "") {
+ //make sure this sequence is in the namefile, else error
+ map<string, int>::iterator it = nameMap.find(current.getName());
+
+ if (it == nameMap.end()) { cout << "[ERROR]: " << current.getName() << " is not in your namefile, please correct." << endl; m->control_pressed = true; }
+ else { num = it->second; }
+ }
+
+ //for each sequence this sequence represents
+ for (int i = 0; i < num; i++) {
+ startPosition.push_back(current.getStartPos());
+ endPosition.push_back(current.getEndPos());
+ seqLength.push_back(current.getNumBases());
+ ambigBases.push_back(current.getAmbigBases());
+ longHomoPolymer.push_back(current.getLongHomoPolymer());
+ }
string outputString = current.getName() + "\t" + toString(current.getStartPos()) + "\t" + toString(current.getEndPos()) + "\t";
- outputString += toString(current.getNumBases()) + "\t" + toString(current.getAmbigBases()) + "\t" + toString(current.getLongHomoPolymer()) + "\n";
+ outputString += toString(current.getNumBases()) + "\t" + toString(current.getAmbigBases()) + "\t" + toString(current.getLongHomoPolymer()) + "\t" + toString(num) + "\n";
//output to file
length = outputString.length();
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
- int exitCommand = 1;
+ int num = 0;
processIDS.clear();
//loop through and create all the processes you want
while (process != processors) {
- int pid = vfork();
+ int pid = fork();
if (pid > 0) {
processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
process++;
}else if (pid == 0){
- driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, sumFile + toString(getpid()) + ".temp", lines[process]);
+ num = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, sumFile + toString(getpid()) + ".temp", lines[process]);
+
+ //pass numSeqs to parent
+ ofstream out;
+ string tempFile = fastafile + toString(getpid()) + ".num.temp";
+ m->openOutputFile(tempFile, out);
+
+ out << num << endl;
+ out << startPosition.size() << endl;
+ for (int k = 0; k < startPosition.size(); k++) { out << startPosition[k] << '\t'; } out << endl;
+ for (int k = 0; k < endPosition.size(); k++) { out << endPosition[k] << '\t'; } out << endl;
+ for (int k = 0; k < seqLength.size(); k++) { out << seqLength[k] << '\t'; } out << endl;
+ for (int k = 0; k < ambigBases.size(); k++) { out << ambigBases[k] << '\t'; } out << endl;
+ for (int k = 0; k < longHomoPolymer.size(); k++) { out << longHomoPolymer[k] << '\t'; } out << endl;
+
+ out.close();
+
exit(0);
- }else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
+ }else {
+ m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
+ for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
+ exit(0);
+ }
}
//force parent to wait until all the processes are done
wait(&temp);
}
- return exitCommand;
-#endif
- }
- catch(exception& e) {
- m->errorOut(e, "SeqSummaryCommand", "createProcessesCreateSummary");
- exit(1);
- }
-}
-/**************************************************************************************************/
-
-int SeqSummaryCommand::setLines(string filename) {
- try {
-
- vector<unsigned long int> positions;
-
- ifstream inFASTA;
- openInputFile(filename, inFASTA);
+ //parent reads in and combine Filter info
+ for (int i = 0; i < processIDS.size(); i++) {
+ string tempFilename = fastafile + toString(processIDS[i]) + ".num.temp";
+ ifstream in;
+ m->openInputFile(tempFilename, in);
- string input;
- while(!inFASTA.eof()){
- input = getline(inFASTA);
-
- if (input.length() != 0) {
- if(input[0] == '>'){ unsigned long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
- }
- }
- inFASTA.close();
-
- int numFastaSeqs = positions.size();
-
- FILE * pFile;
- unsigned long int size;
-
- //get num bytes in file
- pFile = fopen (filename.c_str(),"rb");
- if (pFile==NULL) perror ("Error opening file");
- else{
- fseek (pFile, 0, SEEK_END);
- size=ftell (pFile);
- fclose (pFile);
- }
-
- int numSeqsPerProcessor = numFastaSeqs / processors;
-
- for (int i = 0; i < processors; i++) {
-
- unsigned long int startPos = positions[ i * numSeqsPerProcessor ];
- if(i == processors - 1){
- numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
- }else{
- unsigned long int myEnd = positions[ (i+1) * numSeqsPerProcessor ];
- }
- lines.push_back(new linePair(startPos, numSeqsPerProcessor));
+ int temp, tempNum;
+ in >> tempNum; m->gobble(in); num += tempNum;
+ in >> tempNum; m->gobble(in);
+ for (int k = 0; k < tempNum; k++) { in >> temp; startPosition.push_back(temp); } m->gobble(in);
+ for (int k = 0; k < tempNum; k++) { in >> temp; endPosition.push_back(temp); } m->gobble(in);
+ for (int k = 0; k < tempNum; k++) { in >> temp; seqLength.push_back(temp); } m->gobble(in);
+ for (int k = 0; k < tempNum; k++) { in >> temp; ambigBases.push_back(temp); } m->gobble(in);
+ for (int k = 0; k < tempNum; k++) { in >> temp; longHomoPolymer.push_back(temp); } m->gobble(in);
+
+ in.close();
+ remove(tempFilename.c_str());
}
- return numFastaSeqs;
+ return num;
+#endif
}
catch(exception& e) {
- m->errorOut(e, "SeqSummaryCommand", "setLines");
+ m->errorOut(e, "SeqSummaryCommand", "createProcessesCreateSummary");
exit(1);
}
}
-//***************************************************************************************************************
+/**********************************************************************************************************************/
+
projects / mothur.git / blobdiff