void parse_gtf_file(char* gtfF) {
ifstream fin(gtfF);
string line, tid, gid;
GTFItem item;
general_assert(fin.is_open(), "Cannot open " + cstrtos(gtfF) + "! It may not exist.");
int cnt = 0;
items.clear();
while (getline(fin, line)) {
if (skip(line)) continue;
item.parse(line);
string feature = item.getFeature();
if (feature == "exon") {
if (item.getStart() > item.getEnd()) {
printf("Warning: exon's start position is larger than its end position! This exon is discarded.\n");
printf("\t%s\n\n", line.c_str());
}
else if (item.getStart() < 1) {
printf("Warning: exon's start position is less than 1! This exon is discarded.\n");
printf("\t%s\n\n", line.c_str());
}
else {
item.parseAttributes(line);
if (mappingType > 0) {
tid = item.getTranscriptID();
mi_iter = mi_table.find(tid);
general_assert(mi_iter != mi_table.end(), "Mapping Info is not correct, cannot find " + tid + "'s gene_id!");
gid = mi_iter->second;
item.setGeneID(gid);
}
items.push_back(item);
}
}
++cnt;
if (verbose && cnt % 200000 == 0) { printf("Parsed %d lines\n", cnt); }
}
fin.close();
sort(items.begin(), items.end());
int sp = 0, ep; // start pointer, end pointer
int nItems = items.size();
sn2tr.clear();
while (sp < nItems) {
tid = items[sp].getTranscriptID();
ep = sp + 1;
while (ep < nItems && items[ep].getTranscriptID() == tid) ++ep;
--ep;
buildTranscript(sp, ep);
int sid = transcripts.getM();
const Transcript& transcript = transcripts.getTranscriptAt(sid);
sn2tr_iter = sn2tr.find(transcript.getSeqName());
if (sn2tr_iter == sn2tr.end()) {
vector<int> vec(1, sid);
sn2tr[transcript.getSeqName()] = vec;
}
else {
sn2tr_iter->second.push_back(sid);
}
sp = ep + 1;
}
items.clear();
if (verbose) { printf("Parsing GTF File is done!\n"); }
}
void parse_gtf_file(char* gtfF) {
ifstream fin(gtfF);
string line, curgid, tid, gid; // curgid: current gene id;
GTFItem item;
if (!fin.is_open()) { fprintf(stderr, "Cannot open %s! It may not exist.\n", gtfF); exit(-1); }
int cnt = 0;
items.clear();
while (getline(fin, line)) {
if (line[0] == '#') continue; // if this line is comment, jump it
item.parse(line);
string feature = item.getFeature();
if (feature == "exon") {
if (item.getStart() > item.getEnd()) {
fprintf(stderr, "Warning: exon's start position is larger than its end position! This exon is discarded.\n");
fprintf(stderr, "\t%s\n\n", line.c_str());
}
else if (item.getStart() < 1) {
fprintf(stderr, "Warning: exon's start position is less than 1! This exon is discarded.\n");
fprintf(stderr, "\t%s\n\n", line.c_str());
}
else {
if (hasMappingFile) {
tid = item.getTranscriptID();
mi_iter = mi_table.find(tid);
if (mi_iter == mi_table.end()) {
fprintf(stderr, "Mapping Info is not correct, cannot find %s's gene_id!\n", tid.c_str());
exit(-1);
}
//assert(iter != table.end());
gid = mi_iter->second;
item.setGeneID(gid);
}
items.push_back(item);
}
}
++cnt;
if (verbose && cnt % 200000 == 0) { printf("Parsed %d lines\n", cnt); }
}
fin.close();
sort(items.begin(), items.end());
starts.clear();
sn2tr.clear();
curgid = "";
int sp = 0, ep; // start pointer, end pointer
int nItems = items.size();
while (sp < nItems) {
tid = items[sp].getTranscriptID();
gid = items[sp].getGeneID();
ep = sp + 1;
while (ep < nItems && items[ep].getTranscriptID() == tid) ep++;
ep--;
buildTranscript(sp, ep);
int sid = transcripts.getM();
const Transcript& transcript = transcripts.getTranscriptAt(sid);
if (curgid != gid) {
starts.push_back(sid);
curgid = gid;
}
iter = sn2tr.find(transcript.getSeqName());
if (iter == sn2tr.end()) {
vector<int> vec(1, sid);
sn2tr[transcript.getSeqName()] = vec;
}
else {
iter->second.push_back(sid);
}
sp = ep + 1;
}
M = transcripts.getM();
starts.push_back(M + 1);
items.clear();
if (M < 1) {
fprintf(stderr, "Number of transcripts in the reference is less than 1!\n");
exit(-1);
}
if (verbose) { printf("Parsing gtf File is done!\n"); }
}