void parseFastq(SeqStream input_stream, string filename, SeqSet &data) {
data.filename = filename;
char ch;
string temp = "";
string nm = "";
unsigned size_guess = 150; // Seems like it might speed things up
unsigned line_num = 0;
bool linebreak = false;
bool name = false;
while(!input_stream.eof()) {
// Check if stream is okay and read a character
if(!input_stream.good()) {
throw("Problem reading file");
}
input_stream.get(ch);
// Check for linebreaks. Treat multiple linebreak characters
// as one linebreak. Also, count the number of lines, and when
// four lines have been reach, construct a SeqRecord and reset.
if(ch == '\n' || ch == '\r') {
if(!linebreak) {
line_num += 1;
if(line_num == 4) {
line_num = 0;
SeqRecord rec;
rec.setName(nm);
rec.append(temp);
data.append(rec);
size_guess = rec.getSeq().size();
nm = "";
temp = "";
temp.reserve(size_guess);
name = false;
}
}
linebreak = true;
continue;
}
// If this far, not a linebreak
linebreak = false;
// For each line of the fastq file
if(line_num == 0) {
// Name
if(!name and ch != '@') {
throw("Not in fastq format");
}
if(name) {
nm += ch;
}
name = true;
} else if(line_num == 1) {
// Sequence
temp += ch;
} else if(line_num == 2) {
// Plus line - Ignore
continue;
} else if(line_num == 3) {
// Quality scores - ignore
continue;
}
}
}
void parseFasta(SeqStream input_stream, string filename, SeqSet &data) {
data.filename = filename;
char ch;
string temp = "";
string nm;
unsigned size_guess = 10000; // Seems like it might speed things up
// Enclose all of this in a while loop that goes to EOF:
input_stream.get(ch);
if(ch != '>') {
throw("Not in FASTA format");
}
bool inseq = false;
bool linebreak = false;
while(!input_stream.eof()) {
SeqRecord rec;
rec.reserve(size_guess);
nm = "";
while (true && !inseq) {
if(!input_stream.good()) {
throw("Problem reading file");
}
input_stream.get(ch);
if (ch == '\n' || ch == '\r')
inseq = true;
nm += ch;
}
rec.setName(nm);
temp = "";
while(inseq){
input_stream.get(ch);
if(input_stream.eof())
break;
// ">" after a linebreak means a new name
if(ch == '>' && linebreak) {
inseq = false;
linebreak = false;
continue;
}
// Ignore, but note linebreaks
linebreak = false;
if(ch == '\n' || ch == '\r') {
linebreak = true;
continue;
}
// Ignore whitespace
if(ch == ' ' || ch == '\t') {
continue;
}
temp += ch;
}
rec.append(temp);
data.append(rec);
size_guess = rec.getSeq().size();
}
}
void parseFasta(string filename, SeqSet &data) {
//try {
ifstream input(filename.c_str(), ifstream::in);
data.filename = filename;
char ch;
string temp = "";
string nm;
// Enclose all of this in a while loop that goes to EOF:
input.get(ch);
if(ch != '>') {
throw("Not in FASTA format");
}
bool inseq = false;
bool linebreak = false;
while(!input.eof()) {
SeqRecord rec;
nm = "";
while (true && !inseq) {
input.get(ch);
if (ch == '\n' || ch == '\r')
inseq = true;
nm += ch;
}
rec.setName(nm);
temp = "";
while(inseq){
input.get(ch);
if(input.eof())
break;
// ">" after a linebreak means a new name
if(ch == '>' && linebreak) {
inseq = false;
linebreak = false;
continue;
}
// Ignore, but note linebreaks
linebreak = false;
if(ch == '\n' || ch == '\r') {
linebreak = true;
continue;
}
// Ignore whitespace
if(ch == ' ' || ch == '\t') {
continue;
}
temp += ch;
}
rec.append(temp);
data.append(rec);
}
//} catch (...) {
// throw("Problem parsing file");
//}
}
