#include "removeseqscommand.h"
#include "sequence.hpp"
+#include "listvector.hpp"
//**********************************************************************************************************************
else {
//valid paramters for this command
- string Array[] = {"fasta","name", "group", "alignreport", "accnos" };
+ string Array[] = {"fasta","name", "group", "alignreport", "accnos", "list" };
vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
OptionParser parser(option);
if (alignfile == "not open") { abort = true; }
else if (alignfile == "not found") { alignfile = ""; }
- if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "")) { mothurOut("You must provide one of the following: fasta, name, group, alignreport."); mothurOutEndLine(); abort = true; }
+ listfile = validParameter.validFile(parameters, "list", true);
+ if (listfile == "not open") { abort = true; }
+ else if (listfile == "not found") { listfile = ""; }
- if (parameters.size() > 2) { mothurOut("You may only enter one of the following: fasta, name, group, alignreport."); mothurOutEndLine(); abort = true; }
+ if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "")) { mothurOut("You must provide one of the following: fasta, name, group, alignreport or list."); mothurOutEndLine(); abort = true; }
+
+ if (parameters.size() > 2) { mothurOut("You may only enter one of the following: fasta, name, group, alignreport, or list."); mothurOutEndLine(); abort = true; }
}
}
void RemoveSeqsCommand::help(){
try {
- mothurOut("The remove.seqs command reads an .accnos file and one of the following file types: fasta, name, group or alignreport file.\n");
+ mothurOut("The remove.seqs command reads an .accnos file and one of the following file types: fasta, name, group, list or alignreport file.\n");
mothurOut("It outputs a file containing the sequences NOT in the .accnos file.\n");
- mothurOut("The remove.seqs command parameters are accnos, fasta, name, group and alignreport. You must provide accnos and one of the other parameters.\n");
+ mothurOut("The remove.seqs command parameters are accnos, fasta, name, group, list and alignreport. You must provide accnos and one of the other parameters.\n");
mothurOut("The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
mothurOut("Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
else if (namefile != "") { readName(); }
else if (groupfile != "") { readGroup(); }
else if (alignfile != "") { readAlign(); }
+ else if (listfile != "") { readList(); }
return 0;
}
Sequence currSeq(in);
name = currSeq.getName();
- //if this name is in the accnos file
- if (names.count(name) == 0) {
- wroteSomething = true;
-
- currSeq.printSequence(out);
- }else { names.erase(name); }
-
+ if (name != "") {
+ //if this name is in the accnos file
+ if (names.count(name) == 0) {
+ wroteSomething = true;
+
+ currSeq.printSequence(out);
+ }else { names.erase(name); }
+ }
gobble(in);
}
in.close();
exit(1);
}
}
+//**********************************************************************************************************************
+void RemoveSeqsCommand::readList(){
+ try {
+ string outputFileName = getRootName(listfile) + "pick" + getExtension(listfile);
+ ofstream out;
+ openOutputFile(outputFileName, out);
+
+ ifstream in;
+ openInputFile(listfile, in);
+
+ bool wroteSomething = false;
+
+ while(!in.eof()){
+ //read in list vector
+ ListVector list(in);
+
+ //make a new list vector
+ ListVector newList;
+ newList.setLabel(list.getLabel());
+
+ //for each bin
+ for (int i = 0; i < list.getNumBins(); i++) {
+
+ //parse out names that are in accnos file
+ string binnames = list.get(i);
+
+ string newNames = "";
+ while (binnames.find_first_of(',') != -1) {
+ string name = binnames.substr(0,binnames.find_first_of(','));
+ binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
+
+ //if that name is in the .accnos file, add it
+ if (names.count(name) == 0) { newNames += name + ","; }
+ }
+
+ //get last name
+ if (names.count(binnames) == 0) { newNames += binnames; }
+ //if there are names in this bin add to new list
+ if (newNames != "") { newList.push_back(newNames); }
+ }
+
+ //print new listvector
+ if (newList.getNumBins() != 0) {
+ wroteSomething = true;
+ newList.print(out);
+ }
+
+ gobble(in);
+ }
+ in.close();
+ out.close();
+
+ if (wroteSomething == false) {
+ mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
+ remove(outputFileName.c_str());
+ }
+
+ }
+ catch(exception& e) {
+ errorOut(e, "RemoveSeqsCommand", "readList");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
void RemoveSeqsCommand::readName(){
try {
string outputFileName = getRootName(namefile) + "pick" + getExtension(namefile);
+
ofstream out;
openOutputFile(outputFileName, out);
in >> firstCol;
in >> secondCol;
-
+
vector<string> parsedNames;
//parse second column saving each name
while (secondCol.find_first_of(',') != -1) {
name = secondCol.substr(0,secondCol.find_first_of(','));
secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
parsedNames.push_back(name);
+
}
//get name after last ,
parsedNames.push_back(secondCol);
-
- vector<string> validSecond;
+
+ vector<string> validSecond; validSecond.clear();
for (int i = 0; i < parsedNames.size(); i++) {
if (names.count(parsedNames[i]) == 0) {
validSecond.push_back(parsedNames[i]);
- }else { names.erase(parsedNames[i]); }
+ }
}
-
//if the name in the first column is in the set then print it and any other names in second column also in set
if (names.count(firstCol) == 0) {
-
+
wroteSomething = true;
out << firstCol << '\t';
//make first name in set you come to first column and then add the remaining names to second column
}else {
- names.erase(firstCol);
-
+
//you want part of this row
if (validSecond.size() != 0) {
//alignreport file has a column header line then all other lines contain 16 columns. we just want the first column since that contains the name
void RemoveSeqsCommand::readAlign(){
try {
- string outputFileName = getRootName(alignfile) + "pick" + getExtension(alignfile);
+ string outputFileName = getRootName(getRootName(alignfile)) + "pick.align.report";
ofstream out;
openOutputFile(outputFileName, out);
projects / mothur.git / blobdiff