else {
//valid paramters for this command
- string Array[] = {"fasta", "name", "diffs"};
+ string Array[] = {"fasta", "name", "diffs", "outputdir","inputdir"};
vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
OptionParser parser(option);
map<string, string> parameters = parser.getParameters();
ValidParameters validParameter;
+ map<string, string>::iterator it;
//check to make sure all parameters are valid for command
for (map<string, string>::iterator it2 = parameters.begin(); it2 != parameters.end(); it2++) {
if (validParameter.isValidParameter(it2->first, myArray, it2->second) != true) { abort = true; }
}
+ //if the user changes the input directory command factory will send this info to us in the output parameter
+ string inputDir = validParameter.validFile(parameters, "inputdir", false);
+ if (inputDir == "not found"){ inputDir = ""; }
+ else {
+ string path;
+ it = parameters.find("fasta");
+ //user has given a template file
+ if(it != parameters.end()){
+ path = 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 = 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; }
+ }
+ }
+
//check for required parameters
fastafile = validParameter.validFile(parameters, "fasta", true);
if (fastafile == "not found") { mothurOut("fasta is a required parameter for the pre.cluster command."); mothurOutEndLine(); abort = true; }
else if (fastafile == "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
+ }
+
//check for optional parameter and set defaults
// ...at some point should added some additional type checking...
namefile = validParameter.validFile(parameters, "name", true);
-
if (namefile == "not found") { namefile = ""; }
else if (namefile == "not open") { abort = true; }
-// else { readNameFile(); }
+ else { readNameFile(); }
string temp = validParameter.validFile(parameters, "diffs", false); if(temp == "not found"){ temp = "1"; }
convert(temp, diffs);
if (abort == true) { return 0; }
//reads fasta file and return number of seqs
- int numSeqs = readNamesFASTA(); //fills alignSeqs and makes all seqs active
+ int numSeqs = readFASTA(); //fills alignSeqs and makes all seqs active
if (numSeqs == 0) { mothurOut("Error reading fasta file...please correct."); mothurOutEndLine(); return 0; }
if (diffs > length) { mothurOut("Error: diffs is greater than your sequence length."); mothurOutEndLine(); return 0; }
}//end if active i
if(i % 100 == 0) { cout << i << '\t' << numSeqs - count << '\t' << count << endl; }
}
-
- string newFastaFile = getRootName(fastafile) + "precluster" + getExtension(fastafile);
- string newNamesFile = getRootName(fastafile) + "precluster.names";
+
+ string fileroot = outputDir + getRootName(getSimpleName(fastafile));
+
+ string newFastaFile = fileroot + "precluster" + getExtension(fastafile);
+ string newNamesFile = fileroot + "precluster.names";
mothurOut("Total number of sequences before precluster was " + toString(alignSeqs.size()) + "."); mothurOutEndLine();
exit(1);
}
}
+
/**************************************************************************************************/
-int PreClusterCommand::readFASTA(){
- try {
-// ifstream inFasta;
-// openInputFile(fastafile, inFasta);
-// length = 0;
-//// map<string, string>::iterator it;
-//
-// while (!inFasta.eof()) {
-// Sequence temp(inFasta); //read seq
-//
-// if (temp.getName() != "") {
-// if (namefile != "") {
-// //make sure fasta and name files match
-// it = names.find(temp.getName());
-// if (it == names.end()) { mothurOut(temp.getName() + " is not in your names file, please correct."); mothurOutEndLine(); exit(1); }
-// }else { sizes[temp.getName()] = 1; }
-//
-// seqPNode tempNode(sizes[temp.getName()], temp);
-// alignSeqs.push_back(tempNode);
-// active[temp.getName()] = true;
-// }
-// gobble(inFasta);
-// }
-// inFasta.close();
-//
-// if (alignSeqs.size() != 0) { length = alignSeqs[0].seq.getAligned().length(); }
-//
- return alignSeqs.size();
- }
- catch(exception& e) {
- errorOut(e, "PreClusterCommand", "readFASTA");
- exit(1);
- }
-}
-/**************************************************************************************************/
-int PreClusterCommand::readNamesFASTA(){
+//this requires the names and fasta file to be in the same order
+
+int PreClusterCommand::readFASTA(){
try {
- ifstream inNames;
+ //ifstream inNames;
ifstream inFasta;
- openInputFile(namefile, inNames);
+ //openInputFile(namefile, inNames);
openInputFile(fastafile, inFasta);
- string firstCol, secondCol, nameString;
+ //string firstCol, secondCol, nameString;
length = 0;
- while (inFasta && inNames) {
+ while (!inFasta.eof()) {
- inNames >> firstCol >> secondCol;
- nameString = secondCol;
+ //inNames >> firstCol >> secondCol;
+ //nameString = secondCol;
- gobble(inNames);
- int size = 1;
- while (secondCol.find_first_of(',') != -1) {
- size++;
- secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
+ //gobble(inNames);
+ //int size = 1;
+ //while (secondCol.find_first_of(',') != -1) {
+ // size++;
+ // secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
+ //}
+
+ Sequence seq(inFasta); gobble(inFasta);
+
+ if (seq.getName() != "") { //can get "" if commented line is at end of fasta file
+ if (namefile != "") {
+ itSize = sizes.find(seq.getName());
+
+ if (itSize == sizes.end()) { mothurOut(seq.getName() + " is not in your names file, please correct."); mothurOutEndLine(); exit(1); }
+ else{
+ seqPNode tempNode(itSize->second, seq, names[seq.getName()]);
+ alignSeqs.push_back(tempNode);
+ if (seq.getAligned().length() > length) { length = alignSeqs[0].seq.getAligned().length(); }
+ }
+ }else { //no names file, you are identical to yourself
+ seqPNode tempNode(1, seq, seq.getName());
+ alignSeqs.push_back(tempNode);
+ if (seq.getAligned().length() > length) { length = alignSeqs[0].seq.getAligned().length(); }
+ }
}
- Sequence seq(inFasta);
- if (seq.getName() != firstCol) { mothurOut(seq.getName() + " is not in your names file, please correct."); mothurOutEndLine(); exit(1); }
- else{
- seqPNode tempNode(size, seq, nameString);
- alignSeqs.push_back(tempNode);
- if (seq.getAligned().length() > length) { length = alignSeqs[0].seq.getAligned().length(); }
- }
}
inFasta.close();
- inNames.close();
+ //inNames.close();
return alignSeqs.size();
}
catch(exception& e) {
- errorOut(e, "PreClusterCommand", "readNamesFASTA");
+ errorOut(e, "PreClusterCommand", "readFASTA");
exit(1);
}
}
/**************************************************************************************************/
+void PreClusterCommand::readNameFile(){
+ try {
+ ifstream in;
+ openInputFile(namefile, in);
+ string firstCol, secondCol;
+
+ while (!in.eof()) {
+ in >> firstCol >> secondCol; gobble(in);
+ names[firstCol] = secondCol;
+ int size = 1;
+ while (secondCol.find_first_of(',') != -1) {
+ size++;
+ secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
+ }
+ sizes[firstCol] = size;
+ }
+ in.close();
+ }
+ catch(exception& e) {
+ errorOut(e, "PreClusterCommand", "readNameFile");
+ exit(1);
+ }
+}
+
+/**************************************************************************************************/
+
projects / mothur.git / blobdiff