*
*/
-using namespace std;
-
#include "sequence.hpp"
/***********************************************************************/
-
Sequence::Sequence(){
+ m = MothurOut::getInstance();
initialize();
}
-
/***********************************************************************/
-
Sequence::Sequence(string newName, string sequence) {
-
- initialize();
- name = newName;
- if(sequence.find_first_of('-') != string::npos) {
+ try {
+ m = MothurOut::getInstance();
+ initialize();
+ name = newName;
+
+ //setUnaligned removes any gap characters for us
+ setUnaligned(sequence);
setAligned(sequence);
- isAligned = 1;
}
- setUnaligned(sequence);
+ catch(exception& e) {
+ m->errorOut(e, "Sequence", "Sequence");
+ exit(1);
+ }
+}
+/***********************************************************************/
+Sequence::Sequence(string newName, string sequence, string justUnAligned) {
+ try {
+ m = MothurOut::getInstance();
+ initialize();
+ name = newName;
+
+ //setUnaligned removes any gap characters for us
+ setUnaligned(sequence);
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Sequence", "Sequence");
+ exit(1);
+ }
+}
+
+//********************************************************************************************************************
+//this function will jump over commented out sequences, but if the last sequence in a file is commented out it makes a blank seq
+Sequence::Sequence(istringstream& fastaString){
+ try {
+ m = MothurOut::getInstance();
+ initialize();
+ fastaString >> name;
+
+ if (name.length() != 0) {
+
+ name = name.substr(1);
+ string sequence;
+
+ //read comments
+ while ((name[0] == '#') && fastaString) {
+ while (!fastaString.eof()) { char c = fastaString.get(); if (c == 10 || c == 13){ break; } } // get rest of line if there's any crap there
+ sequence = getCommentString(fastaString);
+
+ if (fastaString) {
+ fastaString >> name;
+ name = name.substr(1);
+ }else {
+ name = "";
+ break;
+ }
+ }
+
+ while (!fastaString.eof()) { char c = fastaString.get(); if (c == 10 || c == 13){ break; } } // get rest of line if there's any crap there
+
+ int numAmbig = 0;
+ sequence = getSequenceString(fastaString, numAmbig);
+
+ setAligned(sequence);
+ //setUnaligned removes any gap characters for us
+ setUnaligned(sequence);
+
+ if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
+
+ }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaString.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Sequence", "Sequence");
+ exit(1);
+ }
}
//********************************************************************************************************************
+//this function will jump over commented out sequences, but if the last sequence in a file is commented out it makes a blank seq
+Sequence::Sequence(istringstream& fastaString, string JustUnaligned){
+ try {
+ m = MothurOut::getInstance();
+
+ initialize();
+ fastaString >> name;
+
+ if (name.length() != 0) {
+
+ name = name.substr(1);
+ string sequence;
+
+ //read comments
+ while ((name[0] == '#') && fastaString) {
+ while (!fastaString.eof()) { char c = fastaString.get(); if (c == 10 || c == 13){ break; } } // get rest of line if there's any crap there
+ sequence = getCommentString(fastaString);
+
+ if (fastaString) {
+ fastaString >> name;
+ name = name.substr(1);
+ }else {
+ name = "";
+ break;
+ }
+ }
+
+ while (!fastaString.eof()) { char c = fastaString.get(); if (c == 10 || c == 13){ break; } } // get rest of line if there's any crap there
+
+ int numAmbig = 0;
+ sequence = getSequenceString(fastaString, numAmbig);
+
+ //setUnaligned removes any gap characters for us
+ setUnaligned(sequence);
+
+ if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
+
+ }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaString.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Sequence", "Sequence");
+ exit(1);
+ }
+}
+
+//********************************************************************************************************************
+//this function will jump over commented out sequences, but if the last sequence in a file is commented out it makes a blank seq
Sequence::Sequence(ifstream& fastaFile){
-
- string accession; // provided a file handle to a fasta-formatted sequence file, read in the next
- fastaFile >> accession; // accession number and sequence we find...
- setName(accession);
+ try {
+ m = MothurOut::getInstance();
+ initialize();
+ fastaFile >> name;
+
+ if (name.length() != 0) {
+
+ name = name.substr(1);
+
+ string sequence;
+
+ //read comments
+ while ((name[0] == '#') && fastaFile) {
+ while (!fastaFile.eof()) { char c = fastaFile.get(); if (c == 10 || c == 13){ break; } } // get rest of line if there's any crap there
+ sequence = getCommentString(fastaFile);
+
+ if (fastaFile) {
+ fastaFile >> name;
+ name = name.substr(1);
+ }else {
+ name = "";
+ break;
+ }
+ }
+
+ //read real sequence
+ while (!fastaFile.eof()) { char c = fastaFile.get(); if (c == 10 || c == 13){ break; } } // get rest of line if there's any crap there
+
+ int numAmbig = 0;
+ sequence = getSequenceString(fastaFile, numAmbig);
+
+ setAligned(sequence);
+ //setUnaligned removes any gap characters for us
+ setUnaligned(sequence);
+
+ if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
+
+ }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
- char letter;
- string sequence;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Sequence", "Sequence");
+ exit(1);
+ }
+}
+//********************************************************************************************************************
+//this function will jump over commented out sequences, but if the last sequence in a file is commented out it makes a blank seq
+Sequence::Sequence(ifstream& fastaFile, string JustUnaligned){
+ try {
+ m = MothurOut::getInstance();
+ initialize();
+ fastaFile >> name;
+
+ if (name.length() != 0) {
+ name = name.substr(1);
+ string sequence;
+
+ //read comments
+ while ((name[0] == '#') && fastaFile) {
+ while (!fastaFile.eof()) { char c = fastaFile.get(); if (c == 10 || c == 13){ break; } } // get rest of line if there's any crap there
+ sequence = getCommentString(fastaFile);
+
+ if (fastaFile) {
+ fastaFile >> name;
+ name = name.substr(1);
+ }else {
+ name = "";
+ break;
+ }
+ }
+
+ //read real sequence
+ while (!fastaFile.eof()) { char c = fastaFile.get(); if (c == 10 || c == 13){ break; } } // get rest of line if there's any crap there
+
+ int numAmbig = 0;
+ sequence = getSequenceString(fastaFile, numAmbig);
+
+ //setUnaligned removes any gap characters for us
+ setUnaligned(sequence);
+
+ if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
+
+ }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Sequence", "Sequence");
+ exit(1);
+ }
+}
+
+//********************************************************************************************************************
+string Sequence::getSequenceString(ifstream& fastaFile, int& numAmbig) {
+ try {
+ char letter;
+ string sequence = "";
+ numAmbig = 0;
+
+ while(fastaFile){
+ letter= fastaFile.get();
+ if(letter == '>'){
+ fastaFile.putback(letter);
+ break;
+ }
+ else if(isprint(letter)){
+ letter = toupper(letter);
+ if(letter == 'U'){letter = 'T';}
+ if(letter != '.' && letter != '-' && letter != 'A' && letter != 'T' && letter != 'G' && letter != 'C' && letter != 'N'){
+ letter = 'N';
+ numAmbig++;
+ }
+ sequence += letter;
+ }
+ }
+
+ return sequence;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Sequence", "getSequenceString");
+ exit(1);
+ }
+}
+//********************************************************************************************************************
+//comment can contain '>' so we need to account for that
+string Sequence::getCommentString(ifstream& fastaFile) {
+ try {
+ char letter;
+ string sequence = "";
+
+ while(fastaFile){
+ letter=fastaFile.get();
+ if((letter == '\r') || (letter == '\n')){
+ m->gobble(fastaFile); //in case its a \r\n situation
+ break;
+ }
+ }
+
+ return sequence;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Sequence", "getCommentString");
+ exit(1);
+ }
+}
+//********************************************************************************************************************
+string Sequence::getSequenceString(istringstream& fastaFile, int& numAmbig) {
+ try {
+ char letter;
+ string sequence = "";
+ numAmbig = 0;
+
+ while(!fastaFile.eof()){
+ letter= fastaFile.get();
- while(fastaFile){
- letter= fastaFile.get();
- if(letter == '>'){
- fastaFile.putback(letter);
- break;
+ if(letter == '>'){
+ fastaFile.putback(letter);
+ break;
+ }
+ else if(isprint(letter)){
+ letter = toupper(letter);
+ if(letter == 'U'){letter = 'T';}
+ if(letter != '.' && letter != '-' && letter != 'A' && letter != 'T' && letter != 'G' && letter != 'C' && letter != 'N'){
+ letter = 'N';
+ numAmbig++;
+ }
+ sequence += letter;
+ }
}
- else if(isprint(letter)){
- letter = toupper(letter);
- if(letter == 'U'){letter = 'T';}
- sequence += letter;
+
+ return sequence;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Sequence", "getSequenceString");
+ exit(1);
+ }
+}
+//********************************************************************************************************************
+//comment can contain '>' so we need to account for that
+string Sequence::getCommentString(istringstream& fastaFile) {
+ try {
+ char letter;
+ string sequence = "";
+
+ while(fastaFile){
+ letter=fastaFile.get();
+ if((letter == '\r') || (letter == '\n')){
+ m->gobble(fastaFile); //in case its a \r\n situation
+ break;
+ }
}
+ return sequence;
}
-
- if(sequence.find_first_of('-') != string::npos){ // if there are any gaps in the sequence, assume that it is
- setAligned(sequence); // an alignment file
+ catch(exception& e) {
+ m->errorOut(e, "Sequence", "getCommentString");
+ exit(1);
}
- setUnaligned(sequence); // also set the unaligned sequence file
}
-
//********************************************************************************************************************
void Sequence::initialize(){
void Sequence::setUnaligned(string sequence){
- if(sequence.find_first_of('-') != string::npos) {
+ if(sequence.find_first_of('.') != string::npos || sequence.find_first_of('-') != string::npos) {
string temp = "";
for(int j=0;j<sequence.length();j++) {
if(isalpha(sequence[j])) { temp += sequence[j]; }
//if the alignment starts or ends with a gap, replace it with a period to indicate missing data
aligned = sequence;
alignmentLength = aligned.length();
+ setUnaligned(sequence);
if(aligned[0] == '-'){
for(int i=0;i<alignmentLength;i++){
}
}
}
-
+ isAligned = 1;
}
//********************************************************************************************************************
//********************************************************************************************************************
string Sequence::getAligned(){
- return aligned;
+ if(isAligned == 0) { return unaligned; }
+ else { return aligned; }
+}
+
+//********************************************************************************************************************
+
+string Sequence::getInlineSeq(){
+ return name + '\t' + aligned;
}
+
//********************************************************************************************************************
string Sequence::getPairwise(){
//********************************************************************************************************************
+void Sequence::removeAmbigBases(){
+
+ for(int j=0;j<alignmentLength;j++){
+ if(aligned[j] != 'A' && aligned[j] != 'T' && aligned[j] != 'G' && aligned[j] != 'C'){
+ aligned[j] = '-';
+ }
+ }
+ setUnaligned(aligned);
+}
+
+//********************************************************************************************************************
+
int Sequence::getLongHomoPolymer(){
if(longHomoPolymer == -1){
longHomoPolymer = 1;
//********************************************************************************************************************
int Sequence::getStartPos(){
- if(endPos == -1){
+ if(startPos == -1){
for(int j = 0; j < alignmentLength; j++) {
if(aligned[j] != '.'){
- startPos = j;
+ startPos = j + 1;
break;
}
}
- }
+ }
+ if(isAligned == 0){ startPos = 1; }
+
return startPos;
}
//********************************************************************************************************************
+void Sequence::padToPos(int start){
+
+ for(int j = startPos-1; j < start-1; j++) {
+ aligned[j] = '.';
+ }
+ startPos = start;
+
+}
+
+//********************************************************************************************************************
+
int Sequence::getEndPos(){
if(endPos == -1){
for(int j=alignmentLength-1;j>=0;j--){
if(aligned[j] != '.'){
- endPos = j;
+ endPos = j + 1;
break;
}
}
}
+ if(isAligned == 0){ endPos = numBases; }
+
return endPos;
}
//********************************************************************************************************************
+void Sequence::padFromPos(int end){
+
+ for(int j = end; j < endPos; j++) {
+ aligned[j] = '.';
+ }
+ endPos = end;
+
+}
+
+//********************************************************************************************************************
+
bool Sequence::getIsAligned(){
return isAligned;
}
+//********************************************************************************************************************
+
+void Sequence::reverseComplement(){
+
+ string temp;
+ for(int i=numBases-1;i>=0;i--){
+ if(unaligned[i] == 'A') { temp += 'T'; }
+ else if(unaligned[i] == 'T'){ temp += 'A'; }
+ else if(unaligned[i] == 'G'){ temp += 'C'; }
+ else if(unaligned[i] == 'C'){ temp += 'G'; }
+ else { temp += 'N'; }
+ }
+ unaligned = temp;
+ aligned = temp;
+
+}
//********************************************************************************************************************
+
+void Sequence::trim(int length){
+
+ if(numBases > length){
+ unaligned = unaligned.substr(0,length);
+ numBases = length;
+ }
+
+}
+
+///**************************************************************************************************/
projects / mothur.git / blobdiff