static hashtable* ReadReference(const char* const refname)
{
hashtable* reference = new_hashtable(12);
sequence* sp = read_fasta_sequence(refname);
while(sp != NULL){
// allocate a coverage array for the sequence
chrcoverage* cov = ckallocz(sizeof(chrcoverage));
cov->length = strlen((char*)sp->sequence);
cov->map = ckallocz(strlen((char*)sp->sequence));
cov->cov = ckallocz(strlen((char*)sp->sequence));
cov->seq = ckallocz(strlen((char*)sp->sequence)+1);
memcpy(cov->seq, sp->sequence, cov->length);
// if the name of the sequence has more than one tokens, just use the
// first token in the name
int i = 0;
while((sp->header[i] != '\n') &&
(sp->header[i] != 0) &&
(sp->header[i] != '\t') &&
(sp->header[i] != 32)) i++;
sp->header[i] = 0;
add_hashtable(reference,(char*)sp->header,strlen((char*)sp->header),cov);
sp = get_next_sequence(sp);
}
close_fasta_sequence(sp);
return reference;
}
static void PrintBinInfo(const char* const refName,
hashtable* const reference,
const int binSize)
{
int gc;
sequence* sp = read_fasta_sequence(refName);
while(sp != NULL){
int i = 0;
while((sp->header[i] != '\n') &&
(sp->header[i] != 0) &&
(sp->header[i] != '\t') &&
(sp->header[i] != 32)) i++;
sp->header[i] = 0;
int binSum = 0;
int binIndex = 0;
u64 start = 0;
chrcoverage* chrcov = (chrcoverage*)must_find_hashtable(reference,
sp->header, strlen(sp->header));
for (u64 j = 0; j < chrcov->length; j++) {
if (chrcov->map[j] == '1') {
binSum += chrcov->cov[j];
binIndex += 1;
if (binIndex == binSize) {
gc = CalculateGC(chrcov, start, j+1);
printf("%s\t%"PRIu64"\t%"PRIu64"\t%d\t%d\t%d\n",
sp->header, start, j+1, binSum, binSize, gc);
binIndex = 0;
binSum = 0;
start = j + 1;
}
}
}
if (binIndex > 0) {
gc = CalculateGC(chrcov, start, chrcov->length);
printf("%s\t%"PRIu64"\t%"PRIu64"\t%d\t%d\t%d\n",
sp->header, start, chrcov->length, binSum, binIndex, gc);
}
sp = get_next_sequence(sp);
}
close_fasta_sequence(sp);
}
bool reads_parsing::get_next_sequence(std::string &seq_read, int &len)
{
bool success = get_next_seq_from_file(seq_read,len,current_file_indx);
if (success)
return true;
if ( current_file_indx < nb_files-1 )
{
close_file(current_file_indx);
current_file_indx++;
open_file(current_file_indx);
return get_next_sequence(seq_read,len);
}
return false;
}