FilterSeqsCommand::FilterSeqsCommand(){
globaldata = GlobalData::getInstance();
- if(globaldata->getFastaFile() != "") { readSeqs = new ReadFasta(globaldata->inputFileName); }
- else if(globaldata->getNexusFile() != "") { readSeqs = new ReadNexus(globaldata->inputFileName); }
- else if(globaldata->getClustalFile() != "") { readSeqs = new ReadClustal(globaldata->inputFileName); }
- else if(globaldata->getPhylipFile() != "") { readSeqs = new ReadPhylip(globaldata->inputFileName); }
-
- readSeqs->read();
- db = readSeqs->getDB();
- numSeqs = db->size();
-
- alignmentLength = db->get(0).getAlignLength();
-
- filter = string(alignmentLength, '1');
+ if(globaldata->getFastaFile() == "") { cout << "You must enter a fasta formatted file" << endl; }
+ trump = globaldata->getTrump()[0];
+// vertical =
+// readSeqs->read();
+// db = readSeqs->getDB();
+// numSeqs = db->size();
+//
+// alignmentLength = db->get(0).getAlignLength();
+//
+// filter = string(alignmentLength, '1');
}
/**************************************************************************************/
else{
filter = hardFilter;
}
-
+
}
/**************************************************************************************/
-void FilterSeqsCommand::doTrump() {
+void FilterSeqsCommand::doTrump(Sequence seq) {
- char trump = globaldata->getTrump()[0];
- for(int i = 0; i < numSeqs; i++) {
- string curAligned = db->get(i).getAligned();;
-
- for(int j = 0; j < alignmentLength; j++) {
- if(curAligned[j] == trump){
- filter[j] = '0';
- }
- }
- }
+// for(int i = 0; i < numSeqs; i++) {
+// string curAligned = db->get(i).getAligned();;
+//
+// for(int j = 0; j < alignmentLength; j++) {
+// if(curAligned[j] == trump){
+// filter[j] = '0';
+// }
+// }
+// }
}
/**************************************************************************************/
-void FilterSeqsCommand::doVertical() {
-
- vector<int> counts(alignmentLength, 0);
-
- for(int i = 0; i < numSeqs; i++) {
- string curAligned = db->get(i).getAligned();;
-
- for(int j = 0; j < alignmentLength; j++) {
- if(curAligned[j] == '-' || curAligned[j] == '.'){
- counts[j]++;
- }
- }
- }
-
for(int i=0;i<alignmentLength;i++){
- if(counts[i] == numSeqs) { filter[i] = '0'; }
- }
+void FilterSeqsCommand::doVertical(Sequence seq) {
+
+// vector<int> counts(alignmentLength, 0);
+//
+// for(int i = 0; i < numSeqs; i++) {
+// string curAligned = db->get(i).getAligned();;
+//
+// for(int j = 0; j < alignmentLength; j++) {
+// if(curAligned[j] == '-' || curAligned[j] == '.'){
+// counts[j]++;
+// }
+// }
+// }
+
// for(int i=0;i<alignmentLength;i++){
+// if(counts[i] == numSeqs) { filter[i] = '0'; }
+// }
}
/**************************************************************************************/
-void FilterSeqsCommand::doSoft() {
-
- int softThreshold = numSeqs * (float)atoi(globaldata->getSoft().c_str()) / 100.0;
-
- vector<int> a(alignmentLength, 0);
- vector<int> t(alignmentLength, 0);
- vector<int> g(alignmentLength, 0);
- vector<int> c(alignmentLength, 0);
- vector<int> x(alignmentLength, 0);
-
- for(int i=0;i<numSeqs;i++){
- string curAligned = db->get(i).getAligned();;
-
-
for(int j=0;j<alignmentLength;j++){
- if(toupper(curAligned[j]) == 'A') { a[j]++; }
- else if(toupper(curAligned[j]) == 'T' || toupper(curAligned[i]) == 'U') { t[j]++; }
- else if(toupper(curAligned[j]) == 'G') { g[j]++; }
- else if(toupper(curAligned[j]) == 'C') { c[j]++; }
- }
- }
-
-
for(int i=0;i<alignmentLength;i++){
- if(a[i] < softThreshold && t[i] < softThreshold && g[i] < softThreshold && c[i] < softThreshold){
- filter[i] = '0';
- }
- }
+void FilterSeqsCommand::doSoft(Sequence seq) {
+
+// int softThreshold = numSeqs * (float)atoi(globaldata->getSoft().c_str()) / 100.0;
+//
+// vector<int> a(alignmentLength, 0);
+// vector<int> t(alignmentLength, 0);
+// vector<int> g(alignmentLength, 0);
+// vector<int> c(alignmentLength, 0);
+// vector<int> x(alignmentLength, 0);
+//
+// for(int i=0;i<numSeqs;i++){
+// string curAligned = db->get(i).getAligned();;
+//
+
// for(int j=0;j<alignmentLength;j++){
+// if(toupper(curAligned[j]) == 'A') { a[j]++; }
+// else if(toupper(curAligned[j]) == 'T' || toupper(curAligned[i]) == 'U') { t[j]++; }
+// else if(toupper(curAligned[j]) == 'G') { g[j]++; }
+// else if(toupper(curAligned[j]) == 'C') { c[j]++; }
+// }
+// }
+//
+
// for(int i=0;i<alignmentLength;i++){
+// if(a[i] < softThreshold && t[i] < softThreshold && g[i] < softThreshold && c[i] < softThreshold){
+// filter[i] = '0';
+// }
+// }
}
/**************************************************************************************/
int FilterSeqsCommand::execute() {
try {
-
- if(globaldata->getHard().compare("") != 0) { doHard(); } // has to be applied first!
- if(globaldata->getTrump().compare("") != 0) { doTrump(); }
- if(globaldata->getVertical() == "T") { doVertical(); }
- if(globaldata->getSoft().compare("") != 0) { doSoft(); }
-
+ ifstream inFASTA;
+ openInputFile(globaldata->getFastaFile(), inFASTA);
+
+ Sequence testSeq(inFASTA);
+ alignmentLength = testSeq.getAlignLength();
+ inFASTA.seekg(0);
+ int numSeqs = 0;
+
+ if(globaldata->getHard().compare("") != 0) { doHard(); }
+ else { filter = string(alignmentLength, '1'); }
+ while(!inFASTA.eof()){
+ Sequence seq(inFASTA);
+ if(globaldata->getTrump().compare("") != 0) { doTrump(seq); }
+ if(isTrue(globaldata->getVertical()) == 1) { doVertical(seq); }
+// if(globaldata->getSoft().compare("") != 0) { doSoft(seq); }
+ numSeqs++;
+ }
+
ofstream outfile;
string filterFile = getRootName(globaldata->inputFileName) + "filter";
openOutputFile(filterFile, outfile);
string filteredFasta = getRootName(globaldata->inputFileName) + "filter.fasta";
openOutputFile(filteredFasta, outfile);
- for(int i=0;i<numSeqs;i++){
- string curAligned = db->get(i).getAligned();
- outfile << '>' << db->get(i).getName() << endl;
-
for(int j=0;j<alignmentLength;j++){
- if(filter[j] == '1'){
- outfile << curAligned[j];
- }
- }
- outfile << endl;
- }
- outfile.close();
+// for(int i=0;i<numSeqs;i++){
+// string curAligned = db->get(i).getAligned();
+// outfile << '>' << db->get(i).getName() << endl;
+
// for(int j=0;j<alignmentLength;j++){
+// if(filter[j] == '1'){
+// outfile << curAligned[j];
+// }
+// }
+// outfile << endl;
+// }
+// outfile.close();
int filteredLength = 0;
for(int i=0;i<alignmentLength;i++){
projects / mothur.git / blobdiff