void SamHeader::parse_text(const char* text) {
std::istringstream strin(text);
std::string line, tag;
std::map<std::string, std::string> dict;
std::map<std::string, std::string>::iterator dict_iter;
HDstr = SQstr = RGstr = PGstr = COstr = other = "";
pids.clear();
while (getline(strin, line)) {
if (line[0] != '@') continue;
tag = line.substr(1, 2);
if (tag == "HD") {
general_assert(HDstr == "", "@HD tag can only present once!");
HDstr = line;
HDstr += "\n";
}
else if (tag == "SQ") {
SQstr += line;
SQstr += "\n";
}
else if (tag == "RG") {
RGstr += line;
RGstr += "\n";
}
else if (tag == "PG") {
dict = parse_line(line);
dict_iter = dict.find("ID");
general_assert(dict_iter != dict.end(), "\"" + line + "\" does not contain an ID!" );
general_assert(pids.find(dict_iter->second) == pids.end(), "Program record identifier " + dict_iter->second + " is not unique!");
pids.insert(dict_iter->second);
PGstr += line;
PGstr += "\n";
}
else if (tag == "CO") {
COstr += line;
COstr += "\n";
}
else {
other += line;
line += "\n";
}
}
}
void SamHeader::replaceSQ(const char* faiF) {
std::ifstream fin(faiF);
general_assert(fin.is_open(), "Cannot open " + cstrtos(faiF) + "! It may not exist.");
std::string line;
size_t pos;
SQstr = "";
while (getline(fin, line)) {
pos = line.find_first_of('\t');
assert(pos != std::string::npos && pos > 0 && pos + 1 < line.length() && line[pos + 1] != '\t');
SQstr += "@SQ\tSN:" + line.substr(0, pos) + "\tLN:" + line.substr(pos + 1, line.find_first_of('\t', pos + 1)) + "\n";
}
fin.close();
}
void build_wiggles(const std::string& bam_filename,
WiggleProcessor& processor) {
samfile_t *bam_in = samopen(bam_filename.c_str(), "r", NULL);
general_assert(bam_in != NULL, "Cannot open " + bam_filename + "!");
bam_hdr_t *header = bam_in->header;
bool *used = new bool[header->n_targets];
memset(used, 0, sizeof(bool) * header->n_targets);
int cur_tid = -1; //current tid;
HIT_INT_TYPE cnt = 0;
bam1_t *b = bam_init1();
Wiggle wiggle;
while (samread(bam_in, b) >= 0) {
if (bam_is_unmapped(b)) continue;
if (b->core.tid != cur_tid) {
if (cur_tid >= 0) { used[cur_tid] = true; processor.process(wiggle); }
cur_tid = b->core.tid;
wiggle.name = header->target_name[cur_tid];
wiggle.length = header->target_len[cur_tid];
wiggle.read_depth.assign(wiggle.length, 0.0);
}
add_bam_record_to_wiggle(b, wiggle);
++cnt;
if (cnt % 1000000 == 0) std::cout<< cnt<< std::endl;
}
if (cur_tid >= 0) { used[cur_tid] = true; processor.process(wiggle); }
for (int32_t i = 0; i < header->n_targets; i++)
if (!used[i]) {
wiggle.name = header->target_name[i];
wiggle.length = header->target_len[i];
wiggle.read_depth.clear();
processor.process(wiggle);
}
bam_destroy1(b);
samclose(bam_in);
delete[] used;
}