#include "sharedutilities.h"
//**********************************************************************************************************************
-vector<string> SubSampleCommand::getValidParameters(){
- try {
- string Array[] = {"fasta", "group", "list","shared","rabund", "name","sabund","size","groups","label","outputdir","inputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+vector<string> SubSampleCommand::setParameters(){
+ try {
+ CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pfasta);
+ CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
+ CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
+ CommandParameter plist("list", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(plist);
+ CommandParameter pshared("shared", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(pshared);
+ CommandParameter prabund("rabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(prabund);
+ CommandParameter psabund("sabund", "InputTypes", "", "", "none", "FLSSR", "none",false,false); parameters.push_back(psabund);
+ CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
+ CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
+ CommandParameter psize("size", "Number", "", "0", "", "", "",false,false); parameters.push_back(psize);
+ CommandParameter ppersample("persample", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(ppersample);
+ 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, "SubSampleCommand", "getValidParameters");
+ m->errorOut(e, "SubSampleCommand", "setParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+string SubSampleCommand::getHelpString(){
+ try {
+ string helpString = "";
+ helpString += "The sub.sample command is designed to be used as a way to normalize your data, or create a smaller set from your original set.\n";
+ helpString += "The sub.sample command parameters are fasta, name, list, group, rabund, sabund, shared, groups, size, persample and label. You must provide a fasta, list, sabund, rabund or shared file as an input file.\n";
+ helpString += "The namefile is only used with the fasta file, not with the listfile, because the list file should contain all sequences.\n";
+ helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included. The group names are separated by dashes.\n";
+ helpString += "The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n";
+ helpString += "The size parameter allows you indicate the size of your subsample.\n";
+ helpString += "The persample parameter allows you indicate you want to select subsample of the same size from each of your groups, default=false. It is only used with the list and fasta files if a groupfile is given.\n";
+ helpString += "persample=false will select a random set of sequences of the size you select, but the number of seqs from each group may differ.\n";
+ helpString += "The size parameter is not set: with shared file size=number of seqs in smallest sample, with all other files if a groupfile is given and persample=true, then size=number of seqs in smallest sample, otherwise size=10% of number of seqs.\n";
+ helpString += "The sub.sample command should be in the following format: sub.sample(list=yourListFile, group=yourGroupFile, groups=yourGroups, label=yourLabels).\n";
+ helpString += "Example sub.sample(list=abrecovery.fn.list, group=abrecovery.groups, groups=B-C, size=20).\n";
+ helpString += "The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n";
+ helpString += "The sub.sample command outputs a .subsample file.\n";
+ helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
+ return helpString;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SubSampleCommand", "getHelpString");
exit(1);
}
}
//**********************************************************************************************************************
SubSampleCommand::SubSampleCommand(){
try {
- abort = true;
- //initialize outputTypes
+ abort = true; calledHelp = true;
+ setParameters();
vector<string> tempOutNames;
outputTypes["shared"] = tempOutNames;
outputTypes["list"] = tempOutNames;
}
}
//**********************************************************************************************************************
-vector<string> SubSampleCommand::getRequiredParameters(){
- try {
- string Array[] = {"fasta","list","shared","rabund", "sabund","or"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
- return myArray;
- }
- catch(exception& e) {
- m->errorOut(e, "SubSampleCommand", "getRequiredParameters");
- exit(1);
- }
-}
-//**********************************************************************************************************************
-vector<string> SubSampleCommand::getRequiredFiles(){
- try {
- vector<string> myArray;
- return myArray;
- }
- catch(exception& e) {
- m->errorOut(e, "SubSampleCommand", "getRequiredFiles");
- exit(1);
- }
-}
-//**********************************************************************************************************************
SubSampleCommand::SubSampleCommand(string option) {
try {
- globaldata = GlobalData::getInstance();
- abort = false;
+ abort = false; calledHelp = false;
allLines = 1;
- labels.clear();
//allow user to run help
- if(option == "help") { help(); abort = true; }
+ if(option == "help") { help(); abort = true; calledHelp = true; }
else {
- //valid paramters for this command
- string Array[] = {"fasta", "group", "list","shared","rabund", "sabund","name","size","groups","label","outputdir","inputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ vector<string> myArray = setParameters();
OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
else {
pickedGroups = true;
m->splitAtDash(groups, Groups);
- globaldata->Groups = Groups;
+ m->Groups = Groups;
}
string temp = validParameter.validFile(parameters, "size", false); if (temp == "not found"){ temp = "0"; }
convert(temp, size);
+ temp = validParameter.validFile(parameters, "persample", false); if (temp == "not found"){ temp = "f"; }
+ persample = m->isTrue(temp);
+
+ if (groupfile == "") { persample = false; }
+
if ((namefile != "") && (fastafile == "")) { m->mothurOut("You may only use a namefile with a fastafile."); m->mothurOutEndLine(); abort = true; }
if ((fastafile == "") && (listfile == "") && (sabundfile == "") && (rabundfile == "") && (sharedfile == "")) {
exit(1);
}
}
-
-//**********************************************************************************************************************
-
-void SubSampleCommand::help(){
- try {
- m->mothurOut("The sub.sample command is designed to be used as a way to normalize your data, or create a smaller set from your original set.\n");
- m->mothurOut("The sub.sample command parameters are fasta, name, list, group, rabund, sabund, shared, groups, size and label. You must provide a fasta, list, sabund, rabund or shared file as an input file.\n");
- m->mothurOut("The namefile is only used with the fasta file, not with the listfile, because the list file should contain all sequences.\n");
- m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like included. The group names are separated by dashes.\n");
- m->mothurOut("The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n");
- m->mothurOut("The size parameter allows you indicate the size of your subsample.\n");
- m->mothurOut("The size parameter is not set: with shared file size=number of seqs in smallest sample, with all other files, 10% of number of seqs.\n");
- m->mothurOut("The sub.sample command should be in the following format: sub.sample(list=yourListFile, group=yourGroupFile, groups=yourGroups, label=yourLabels).\n");
- m->mothurOut("Example sub.sample(list=abrecovery.fn.list, group=abrecovery.groups, groups=B-C, size=20).\n");
- m->mothurOut("The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n");
- m->mothurOut("The sub.sample command outputs a .subsample file.\n");
- m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
-
- }
- catch(exception& e) {
- m->errorOut(e, "SubSampleCommand", "help");
- exit(1);
- }
-}
-
-//**********************************************************************************************************************
-
-SubSampleCommand::~SubSampleCommand(){}
-
//**********************************************************************************************************************
int SubSampleCommand::execute(){
try {
- if (abort == true) { return 0; }
+ if (abort == true) { if (calledHelp) { return 0; } return 2; }
if (sharedfile != "") { getSubSampleShared(); }
if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); return 0; } }
if (fastafile != "") { getSubSampleFasta(); }
if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); return 0; } }
-
+ //set fasta file as new current fastafile
+ string current = "";
+ itTypes = outputTypes.find("fasta");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
+ }
+
+ itTypes = outputTypes.find("name");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
+ }
+
+ itTypes = outputTypes.find("group");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
+ }
+
+ itTypes = outputTypes.find("list");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
+ }
+
+ itTypes = outputTypes.find("shared");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
+ }
+
+ itTypes = outputTypes.find("rabund");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
+ }
+
+ itTypes = outputTypes.find("sabund");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
+ }
+
+
m->mothurOutEndLine();
m->mothurOut("Output File Names: "); m->mothurOutEndLine();
for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
outputTypes["fasta"].push_back(outputFileName); outputNames.push_back(outputFileName);
//make sure that if your picked groups size is not too big
-
- if (pickedGroups) {
- int total = 0;
- for(int i = 0; i < Groups.size(); i++) {
- total += groupMap->getNumSeqs(Groups[i]);
+ int thisSize = names.size();
+ if (persample) {
+ if (size == 0) { //user has not set size, set size = smallest samples size
+ size = groupMap->getNumSeqs(Groups[0]);
+ for (int i = 1; i < Groups.size(); i++) {
+ int thisSize = groupMap->getNumSeqs(Groups[i]);
+
+ if (thisSize < size) { size = thisSize; }
+ }
+ }else { //make sure size is not too large
+ int smallestSize = groupMap->getNumSeqs(Groups[0]);
+ for (int i = 1; i < Groups.size(); i++) {
+ int thisSize = groupMap->getNumSeqs(Groups[i]);
+
+ if (thisSize < smallestSize) { smallestSize = thisSize; }
+ }
+ if (smallestSize < size) { size = smallestSize; m->mothurOut("You have selected a size that is larger than your smallest sample, using your samllest sample size, " + toString(smallestSize) + "."); m->mothurOutEndLine(); }
+ }
+
+ m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
+ }else {
+ if (pickedGroups) {
+ int total = 0;
+ for(int i = 0; i < Groups.size(); i++) {
+ total += groupMap->getNumSeqs(Groups[i]);
+ }
+
+ if (size == 0) { //user has not set size, set size = 10% samples size
+ size = int (total * 0.10);
+ }
+
+ if (total < size) {
+ if (size != 0) {
+ m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
+ }
+ size = int (total * 0.10);
+ }
+
+ m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
}
if (size == 0) { //user has not set size, set size = 10% samples size
- size = int (total * 0.10);
+ size = int (names.size() * 0.10);
}
- if (total < size) {
- if (size != 0) {
- m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
- }
- size = int (total * 0.10);
+ if (size > thisSize) { m->mothurOut("Your fasta file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
+ size = thisSize;
}
- m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
- }
-
- if (size == 0) { //user has not set size, set size = 10% samples size
- size = int (names.size() * 0.10);
- }
-
- int thisSize = names.size();
- if (size > thisSize) { m->mothurOut("Your fasta file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
- size = thisSize;
+ if (!pickedGroups) { m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine(); }
+
}
-
random_shuffle(names.begin(), names.end());
- if (!pickedGroups) { m->mothurOut("Sampling " + toString(size) + " from " + toString(thisSize) + "."); m->mothurOutEndLine(); }
-
- //randomly select a subset of those names to include in the subsample
set<string> subset; //dont want repeat sequence names added
- for (int j = 0; j < size; j++) {
-
- if (m->control_pressed) { return 0; }
+ if (persample) {
+ for (int i = 0; i < Groups.size(); i++) {
+
+ //randomly select a subset of those names from this group to include in the subsample
+ for (int j = 0; j < size; j++) {
+
+ if (m->control_pressed) { return 0; }
+
+ //get random sequence to add, making sure we have not already added it
+ bool done = false;
+ int myrand;
+ while (!done) {
+ myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
+
+ if (subset.count(names[myrand]) == 0) {
+
+ string group = groupMap->getGroup(names[myrand]);
+ if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
+
+ if (group == Groups[i]) { subset.insert(names[myrand]); break; }
+ }
+ }
+ }
+ }
+ }else {
- //get random sequence to add, making sure we have not already added it
- bool done = false;
- int myrand;
- while (!done) {
- myrand = (int)((float)(rand()) / (RAND_MAX / (thisSize-1) + 1));
+ //randomly select a subset of those names to include in the subsample
+ for (int j = 0; j < size; j++) {
+
+ if (m->control_pressed) { return 0; }
- if (subset.count(names[myrand]) == 0) {
+ //get random sequence to add, making sure we have not already added it
+ bool done = false;
+ int myrand;
+ while (!done) {
+ myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
- if (groupfile != "") { //if there is a groupfile given fill in group info
- string group = groupMap->getGroup(names[myrand]);
- if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
+ if (subset.count(names[myrand]) == 0) {
- if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
- if (m->inUsersGroups(group, Groups)) {
+ if (groupfile != "") { //if there is a groupfile given fill in group info
+ string group = groupMap->getGroup(names[myrand]);
+ if (group == "not found") { m->mothurOut("[ERROR]: " + names[myrand] + " is not in your groupfile. please correct."); m->mothurOutEndLine(); group = "NOTFOUND"; }
+
+ if (pickedGroups) { //if hte user picked groups, we only want to keep the names of sequences from those groups
+ if (m->inUsersGroups(group, Groups)) {
+ subset.insert(names[myrand]); break;
+ }
+ }else{
subset.insert(names[myrand]); break;
}
- }else{
+ }else{ //save everyone, group
subset.insert(names[myrand]); break;
- }
- }else{ //save everyone, group
- subset.insert(names[myrand]); break;
- }
+ }
+ }
}
- }
- }
-
+ }
+ }
//read through fasta file outputting only the names on the subsample list
ifstream in;
m->openInputFile(fastafile, in);
}
in.close();
+ return 0;
+
}
catch(exception& e) {
m->errorOut(e, "SubSampleCommand", "getNames");
}
in.close();
+ return 0;
+
}
catch(exception& e) {
m->errorOut(e, "SubSampleCommand", "readNames");
if (m->control_pressed) { delete order; return 0; }
//get random number to sample from order between 0 and thisSize-1.
- int myrand = (int)((float)(rand()) / (RAND_MAX / (thisSize-1) + 1));
+ int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
int bin = order->get(myrand);
}
//make sure that if your picked groups size is not too big
- if (pickedGroups) {
- int total = 0;
- for(int i = 0; i < Groups.size(); i++) {
- total += groupMap->getNumSeqs(Groups[i]);
- }
-
- if (size == 0) { //user has not set size, set size = 10% samples size
- size = int (total * 0.10);
- }
-
- if (total < size) {
- m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
- size = int (total * 0.10);
+ if (persample) {
+ if (size == 0) { //user has not set size, set size = smallest samples size
+ size = groupMap->getNumSeqs(Groups[0]);
+ for (int i = 1; i < Groups.size(); i++) {
+ int thisSize = groupMap->getNumSeqs(Groups[i]);
+
+ if (thisSize < size) { size = thisSize; }
+ }
+ }else { //make sure size is not too large
+ int smallestSize = groupMap->getNumSeqs(Groups[0]);
+ for (int i = 1; i < Groups.size(); i++) {
+ int thisSize = groupMap->getNumSeqs(Groups[i]);
+
+ if (thisSize < smallestSize) { smallestSize = thisSize; }
+ }
+ if (smallestSize < size) { size = smallestSize; m->mothurOut("You have selected a size that is larger than your smallest sample, using your samllest sample size, " + toString(smallestSize) + "."); m->mothurOutEndLine(); }
}
- m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
+ m->mothurOut("Sampling " + toString(size) + " from each group."); m->mothurOutEndLine();
}else{
-
- if (size == 0) { //user has not set size, set size = 10% samples size
- size = int (list->getNumSeqs() * 0.10);
- }
-
- int thisSize = list->getNumSeqs();
- if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
- size = thisSize;
+ if (pickedGroups) {
+ int total = 0;
+ for(int i = 0; i < Groups.size(); i++) {
+ total += groupMap->getNumSeqs(Groups[i]);
+ }
+
+ if (size == 0) { //user has not set size, set size = 10% samples size
+ size = int (total * 0.10);
+ }
+
+ if (total < size) {
+ m->mothurOut("Your size is too large for the number of groups you selected. Adjusting to " + toString(int (total * 0.10)) + "."); m->mothurOutEndLine();
+ size = int (total * 0.10);
+ }
+
+ m->mothurOut("Sampling " + toString(size) + " from " + toString(total) + "."); m->mothurOutEndLine();
+ }else{
+
+ if (size == 0) { //user has not set size, set size = 10% samples size
+ size = int (list->getNumSeqs() * 0.10);
+ }
+
+ int thisSize = list->getNumSeqs();
+ if (size > thisSize) { m->mothurOut("Your list file only contains " + toString(thisSize) + " sequences. Setting size to " + toString(thisSize) + "."); m->mothurOutEndLine();
+ size = thisSize;
+ }
+
+ m->mothurOut("Sampling " + toString(size) + " from " + toString(list->getNumSeqs()) + "."); m->mothurOutEndLine();
}
-
- m->mothurOut("Sampling " + toString(size) + " from " + toString(list->getNumSeqs()) + "."); m->mothurOutEndLine();
}
+ //fill names
for (int i = 0; i < list->getNumBins(); i++) {
string binnames = list->get(i);
}
random_shuffle(names.begin(), names.end());
-
+
//randomly select a subset of those names to include in the subsample
set<string> subset; //dont want repeat sequence names added
- for (int j = 0; j < size; j++) {
-
- if (m->control_pressed) { break; }
-
- //get random sequence to add, making sure we have not already added it
- bool done = false;
- int myrand;
- while (!done) {
- myrand = (int)((float)(rand()) / (RAND_MAX / (names.size()-1) + 1));
+ if (persample) {
+ for (int i = 0; i < Groups.size(); i++) {
- if (subset.count(names[myrand]) == 0) { subset.insert(names[myrand]); break; }
+ for (int j = 0; j < size; j++) {
+
+ if (m->control_pressed) { break; }
+
+ //get random sequence to add, making sure we have not already added it
+ bool done = false;
+ int myrand;
+ while (!done) {
+ myrand = int((float)(names.size()) * (float)(rand()) / ((float)RAND_MAX+1.0));
+
+ if (subset.count(names[myrand]) == 0) { //you are not already added
+ if (groupMap->getGroup(names[myrand]) == Groups[i]) { subset.insert(names[myrand]); break; }
+ }
+ }
+ }
}
- }
+ }else{
+ for (int j = 0; j < size; j++) {
+
+ if (m->control_pressed) { break; }
+
+ //get random sequence to add, making sure we have not already added it
+ bool done = false;
+ int myrand;
+ while (!done) {
+ myrand = int((float)(names.size()) * (float)(rand()) / ((float)RAND_MAX+1.0));
+
+ if (subset.count(names[myrand]) == 0) { subset.insert(names[myrand]); break; }
+ }
+ }
+ }
if (groupfile != "") {
//write out new groupfile
if (m->control_pressed) { delete order; return 0; }
//get random number to sample from order between 0 and thisSize-1.
- int myrand = (int)((float)(rand()) / (RAND_MAX / (thisSize-1) + 1));
+ int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
int bin = order->get(myrand);
if (m->control_pressed) { delete order; return 0; }
//get random number to sample from order between 0 and thisSize-1.
- int myrand = (int)((float)(rand()) / (RAND_MAX / (thisSize-1) + 1));
+ int myrand = int((float)(thisSize) * (float)(rand()) / ((float)RAND_MAX+1.0));
int bin = order->get(myrand);
projects / mothur.git / blobdiff