/**
* Reads past the next ambiguous or unambiguous stretch of sequence
* from the given FASTA file and returns its length. Does not do
* anything with the sequence characters themselves; this is purely for
* measuring lengths.
*/
RefRecord fastaRefReadSize(FileBuf& in,
const RefReadInParams& rparms,
bool first,
BitpairOutFileBuf* bpout)
{
int c;
static int lastc = '>'; // last character seen
// RefRecord params
size_t len = 0; // 'len' counts toward total length
// 'off' counts number of ambiguous characters before first
// unambiguous character
size_t off = 0;
// Pick off the first carat and any preceding whitespace
if(first) {
assert(!in.eof());
lastc = '>';
c = in.getPastWhitespace();
if(in.eof()) {
// Got eof right away; emit warning
cerr << "Warning: Empty input file" << endl;
lastc = -1;
return RefRecord(0, 0, true);
}
assert(c == '>');
}
first = true;
// Skip to the end of the id line; if the next line is either
// another id line or a comment line, keep skipping
if(lastc == '>') {
// Skip to the end of the name line
do {
if((c = in.getPastNewline()) == -1) {
// No more input
cerr << "Warning: Encountered empty reference sequence" << endl;
lastc = -1;
return RefRecord(0, 0, true);
}
if(c == '>') {
cerr << "Warning: Encountered empty reference sequence" << endl;
}
// continue until a non-name, non-comment line
} while (c == '>');
} else {
first = false; // not the first in a sequence
off = 1; // The gap has already been consumed, so count it
if((c = in.get()) == -1) {
// Don't emit a warning, since this might legitimately be
// a gap on the end of the final sequence in the file
lastc = -1;
return RefRecord(off, len, first);
}
}
// Now skip to the first DNA character, counting gap characters
// as we go
int lc = -1; // last-DNA char variable for color conversion
while(true) {
int cat = dna4Cat[c];
if(rparms.nsToAs && cat == 2) c = 'A';
if(cat == 1) {
// This is a DNA character
if(rparms.color) {
if(lc != -1) {
// Got two consecutive unambiguous DNAs
break; // to read-in loop
}
// Keep going; we need two consecutive unambiguous DNAs
lc = charToDna5[(int)c];
// The 'if(off > 0)' takes care of the case where
// the reference is entirely unambiguous and we don't
// want to incorrectly increment off.
if(off > 0) off++;
} else {
break; // to read-in loop
}
} else if(cat == 2) {
if(lc != -1 && off == 0) off++;
lc = -1;
off++; // skip over gap character and increment
} else if(c == '>') {
if(off > 0 && lastc == '>') {
cerr << "Warning: Encountered reference sequence with only gaps" << endl;
} else if(lastc == '>') {
cerr << "Warning: Encountered empty reference sequence" << endl;
}
lastc = '>';
return RefRecord(off, 0, first);
}
c = in.get();
if(c == -1) {
// End-of-file
if(off > 0 && lastc == '>') {
cerr << "Warning: Encountered reference sequence with only gaps" << endl;
} else if(lastc == '>') {
cerr << "Warning: Encountered empty reference sequence" << endl;
}
lastc = -1;
return RefRecord(off, 0, first);
}
}
assert(!rparms.color || (lc != -1));
assert_eq(1, dna4Cat[c]); // C must be unambiguous base
if(off > 0 && rparms.color && first) {
// Handle the case where the first record has ambiguous
// characters but we're in color space; one of those counts is
// spurious
off--;
}
// in now points just past the first character of a sequence
// line, and c holds the first character
while(c != -1 && c != '>') {
if(rparms.nsToAs && dna4Cat[c] == 2) c = 'A';
uint8_t cat = dna4Cat[c];
int cc = toupper(c);
if(rparms.bisulfite && cc == 'C') c = cc = 'T';
if(cat == 1) {
// It's a DNA character
assert(cc == 'A' || cc == 'C' || cc == 'G' || cc == 'T');
// Consume it
len++;
// Output it
if(bpout != NULL) {
if(rparms.color) {
// output color
bpout->write(dinuc2color[charToDna5[(int)c]][lc]);
} else if(!rparms.color) {
// output nucleotide
bpout->write(charToDna5[c]);
}
}
lc = charToDna5[(int)c];
} else if(cat == 2) {
// It's an N or a gap
lastc = c;
assert(cc != 'A' && cc != 'C' && cc != 'G' && cc != 'T');
return RefRecord(off, len, first);
} else {
// Not DNA and not a gap, ignore it
#ifndef NDEBUG
if(!isspace(c)) {
cerr << "Unexpected character in sequence: ";
if(isprint(c)) {
cerr << ((char)c) << endl;
} else {
cerr << "(" << c << ")" << endl;
}
}
#endif
}
c = in.get();
}
lastc = c;
return RefRecord(off, len, first);
}
/**
* Parse a single quality string from fb and store qualities in r.
* Assume the next character obtained via fb.get() is the first
* character of the quality string. When returning, the next
* character returned by fb.peek() or fb.get() should be the first
* character of the following line.
*/
int parseQuals(
Read& r,
FileBuf& fb,
int firstc,
int readLen,
int trim3,
int trim5,
bool intQuals,
bool phred64,
bool solexa64)
{
int c = firstc;
assert(c != '\n' && c != '\r');
r.qual.clear();
if (intQuals) {
while (c != '\r' && c != '\n' && c != -1) {
bool neg = false;
int num = 0;
while(!isspace(c) && !fb.eof()) {
if(c == '-') {
neg = true;
assert_eq(num, 0);
} else {
if(!isdigit(c)) {
char buf[2048];
cerr << "Warning: could not parse quality line:" << endl;
fb.getPastNewline();
cerr << fb.copyLastN(buf);
buf[2047] = '\0';
cerr << buf;
throw 1;
}
assert(isdigit(c));
num *= 10;
num += (c - '0');
}
c = fb.get();
}
if(neg) num = 0;
// Phred-33 ASCII encode it and add it to the back of the
// quality string
r.qual.append('!' + num);
// Skip over next stretch of whitespace
while(c != '\r' && c != '\n' && isspace(c) && !fb.eof()) {
c = fb.get();
}
}
} else {
while (c != '\r' && c != '\n' && c != -1) {
r.qual.append(charToPhred33(c, solexa64, phred64));
c = fb.get();
while(c != '\r' && c != '\n' && isspace(c) && !fb.eof()) {
c = fb.get();
}
}
}
if ((int)r.qual.length() < readLen-1 ||
((int)r.qual.length() < readLen && !r.color))
{
tooFewQualities(r.name);
}
r.qual.trimEnd(trim3);
if(r.qual.length()-trim5 < r.patFw.length()) {
assert(gColor && r.primer != -1);
assert_gt(trim5, 0);
trim5--;
}
r.qual.trimBegin(trim5);
if(r.qual.length() <= 0) return 0;
assert_eq(r.qual.length(), r.patFw.length());
while(fb.peek() == '\n' || fb.peek() == '\r') fb.get();
return (int)r.qual.length();
}