globaldata = GlobalData::getInstance();
abort = false;
+ unique = true;
allLines = 1;
labels.clear();
else {
//valid paramters for this command
- string Array[] = {"label","groups","fasta","list","group","output"};
+ string Array[] = {"label","unique","shared","fasta","list","group","output","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 it = parameters.begin(); it != parameters.end(); it++) {
+ for (it = parameters.begin(); it != parameters.end(); it++) {
if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { 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 = ""; }
+
+ //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("list");
+ //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["list"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("group");
+ //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["group"] = inputDir + it->second; }
+ }
+ }
+
+
//check for required parameters
listfile = validParameter.validFile(parameters, "list", true);
if (listfile == "not open") { abort = true; }
output = validParameter.validFile(parameters, "output", false);
if (output == "not found") { output = ""; }
- groups = validParameter.validFile(parameters, "groups", false);
+ groups = validParameter.validFile(parameters, "unique", false);
if (groups == "not found") { groups = ""; }
else {
+ userGroups = "unique." + groups;
splitAtDash(groups, Groups);
globaldata->Groups = Groups;
+
+ }
+
+ groups = validParameter.validFile(parameters, "shared", false);
+ if (groups == "not found") { groups = ""; }
+ else {
+ userGroups = groups;
+ splitAtDash(groups, Groups);
+ globaldata->Groups = Groups;
+ unique = false;
}
fastafile = validParameter.validFile(parameters, "fasta", true);
void GetSharedOTUCommand::help(){
try {
- mothurOut("The get.sharedotu command parameters are list, group, label, groups, output and fasta. The list and group parameters are required.\n");
+ mothurOut("The get.sharedseqs command parameters are list, group, label, unique, shared, output and fasta. The list and group parameters are required.\n");
mothurOut("The label parameter allows you to select what distance levels you would like output files for, and are separated by dashes.\n");
- mothurOut("The groups parameter allows you to select groups you would like to know the shared info for, and are separated by dashes.\n");
+ mothurOut("The unique and shared parameters allow you to select groups you would like to know the shared info for, and are separated by dashes.\n");
+ mothurOut("If you enter your groups under the unique parameter mothur will return the otus that contain ONLY sequences from those groups.\n");
+ mothurOut("If you enter your groups under the shared parameter mothur will return the otus that contain sequences from those groups and may also contain sequences from other groups.\n");
+ mothurOut("If you do not enter any groups then the get.sharedseqs command will return sequences that are unique to all groups in your group file.\n");
mothurOut("The fasta parameter allows you to input a fasta file and outputs a fasta file for each distance level containing only the sequences that are in OTUs shared by the groups specified.\n");
mothurOut("The output parameter allows you to output the list of names without the group and bin number added. \n");
mothurOut("With this option you can use the names file as an input in get.seqs and remove.seqs commands. To do this enter output=accnos. \n");
- mothurOut("The get.sharedotu command outputs a .names file for each distance level containing a list of sequences in the OTUs shared by the groups specified.\n");
- mothurOut("The get.sharedotu command should be in the following format: get.sabund(label=yourLabels, groups=yourGroups, fasta=yourFastafile, output=yourOutput).\n");
- mothurOut("Example get.sharedotu(list=amazon.fn.list, label=unique-0.01, group=forest-pasture, fasta=amazon.fasta, output=accnos).\n");
+ mothurOut("The get.sharedseqs command outputs a .names file for each distance level containing a list of sequences in the OTUs shared by the groups specified.\n");
+ mothurOut("The get.sharedseqs command should be in the following format: get.sabund(label=yourLabels, groups=yourGroups, fasta=yourFastafile, output=yourOutput).\n");
+ mothurOut("Example get.sharedseqs(list=amazon.fn.list, label=unique-0.01, group=forest-pasture, fasta=amazon.fasta, output=accnos).\n");
mothurOut("The output to the screen is the distance and the number of otus at that distance for the groups you specified.\n");
mothurOut("The default value for label is all labels in your inputfile. The default for groups is all groups in your file.\n");
mothurOut("Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabel).\n\n");
if (abort == true) { return 0; }
groupMap = new GroupMap(groupfile);
- groupMap->readMap();
+ int error = groupMap->readMap();
+ if (error == 1) { delete groupMap; return 0; }
+
globaldata->gGroupmap = groupMap;
if (Groups.size() == 0) {
Groups = groupMap->namesOfGroups;
+
+ //make string for outputfile name
+ userGroups = "unique.";
+ for(int i = 0; i < Groups.size(); i++) { userGroups += Groups[i] + "-"; }
+ userGroups = userGroups.substr(0, userGroups.length()-1);
}
-
+
//put groups in map to find easier
for(int i = 0; i < Groups.size(); i++) {
groupFinder[Groups[i]] = Groups[i];
while(!inFasta.eof()) {
Sequence seq(inFasta);
- seqs.push_back(seq);
+ if (seq.getName() != "") { seqs.push_back(seq); }
}
inFasta.close();
}
map<string, string> fastaMap;
ofstream outNames;
- string outputFileNames = getRootName(listfile) + shared->getLabel() + ".names";
+ string outputFileNames;
+
+ if (outputDir == "") { outputDir += hasPath(listfile); }
+ if (output != "accnos") {
+ outputFileNames = outputDir + getRootName(getSimpleName(listfile)) + shared->getLabel() + userGroups + ".shared.seqs";
+ }else {
+ outputFileNames = outputDir + getRootName(getSimpleName(listfile)) + shared->getLabel() + userGroups + ".accnos";
+ }
openOutputFile(outputFileNames, outNames);
bool wroteSomething = false;
//go through each bin, find out if shared
for (int i = 0; i < shared->getNumBins(); i++) {
- bool sharedByAll = true;
+ bool uniqueOTU = true;
map<string, int> atLeastOne;
for (int m = 0; m < Groups.size(); m++) {
vector<string> namesOfSeqsInThisBin;
string names = shared->get(i);
- while ((names.find_first_of(',') != -1) && sharedByAll) {
+ while ((names.find_first_of(',') != -1)) {
string name = names.substr(0,names.find_first_of(','));
names = names.substr(names.find_first_of(',')+1, names.length());
-
+
//find group
string seqGroup = groupMap->getGroup(name);
if (output != "accnos") {
- namesOfSeqsInThisBin.push_back((name + "|" + seqGroup + "|" + toString(i+1)));
+ namesOfSeqsInThisBin.push_back((name + "\t" + seqGroup + "\t" + toString(i+1)));
}else { namesOfSeqsInThisBin.push_back(name); }
-
+
if (seqGroup == "not found") { mothurOut(name + " is not in your groupfile. Please correct."); mothurOutEndLine(); exit(1); }
//is this seq in one of hte groups we care about
it = groupFinder.find(seqGroup);
- if (it == groupFinder.end()) { sharedByAll = false; } //you have a sequence from a group you don't want
+ if (it == groupFinder.end()) { uniqueOTU = false; } //you have a sequence from a group you don't want
else { atLeastOne[seqGroup]++; }
}
//get last name
- //find group
- if (sharedByAll) {
- string seqGroup = groupMap->getGroup(names);
- if (output != "accnos") {
- namesOfSeqsInThisBin.push_back((names + "|" + seqGroup + "|" + toString(i+1)));
- }else { namesOfSeqsInThisBin.push_back(names); }
-
- if (seqGroup == "not found") { mothurOut(names + " is not in your groupfile. Please correct."); mothurOutEndLine(); exit(1); }
+ string seqGroup = groupMap->getGroup(names);
+ if (output != "accnos") {
+ namesOfSeqsInThisBin.push_back((names + "\t" + seqGroup + "\t" + toString(i+1)));
+ }else { namesOfSeqsInThisBin.push_back(names); }
+
+ if (seqGroup == "not found") { mothurOut(names + " is not in your groupfile. Please correct."); mothurOutEndLine(); exit(1); }
+
+ //is this seq in one of hte groups we care about
+ it = groupFinder.find(seqGroup);
+ if (it == groupFinder.end()) { uniqueOTU = false; } //you have a sequence from a group you don't want
+ else { atLeastOne[seqGroup]++; }
- //is this seq in one of hte groups we care about
- it = groupFinder.find(seqGroup);
- if (it == groupFinder.end()) { sharedByAll = false; } //you have a sequence from a group you don't want
- else { atLeastOne[seqGroup]++; }
- }
//make sure you have at least one seq from each group you want
+ bool sharedByAll = true;
map<string, int>::iterator it2;
for (it2 = atLeastOne.begin(); it2 != atLeastOne.end(); it2++) {
if (it2->second == 0) { sharedByAll = false; }
}
- //if shared, save names of seqs in that bin
- if (sharedByAll) {
+ //if the user wants unique bins and this is unique then print
+ //or this the user wants shared bins and this bin is shared then print
+ if ((unique && uniqueOTU && sharedByAll) || (!unique && sharedByAll)) {
wroteSomething = true;
num++;
if (fastafile != "") {
if (output != "accnos") {
- string seqName = namesOfSeqsInThisBin[j].substr(0,namesOfSeqsInThisBin[j].find_last_of('|'));
- seqName = seqName.substr(0,seqName.find_last_of('|'));
+ string seqName = namesOfSeqsInThisBin[j].substr(0,namesOfSeqsInThisBin[j].find_last_of('\t'));
+ seqName = seqName.substr(0,seqName.find_last_of('\t'));
fastaMap[seqName] = namesOfSeqsInThisBin[j]; //fastaMap needs to contain just the seq name for output later
}else {
fastaMap[namesOfSeqsInThisBin[j]] = namesOfSeqsInThisBin[j];
}
}
}
-
-
-
}
outNames.close();
}else { mothurOut("\t" + toString(num)); mothurOutEndLine(); }
//if fasta file provided output new fasta file
- if ((fastafile != "") && wroteSomething) {
- string outputFileFasta = getRootName(fastafile) + shared->getLabel() + ".fasta";
+ if ((fastafile != "") && wroteSomething) {
+ if (outputDir == "") { outputDir += hasPath(fastafile); }
+ string outputFileFasta = outputDir + getRootName(getSimpleName(fastafile)) + shared->getLabel() + userGroups + ".shared.fasta";
ofstream outFasta;
openOutputFile(outputFileFasta, outFasta);
}
outFasta << seqs[k].getAligned() << endl;
- }
+ }else { mothurOut(seqs[k].getName() + " is not in your fasta file. Please correct."); mothurOutEndLine(); }
}
outFasta.close();
projects / mothur.git / blobdiff