Function buildTranscript

extractRef.cpp:89–130

}

bool buildTranscript(int sp, int ep) {
	int cur_s, cur_e; // current_start, current_end
	vector<Interval> vec;

	string transcript_id = items[sp].getTranscriptID();
	string gene_id = items[sp].getGeneID();
	string gene_name = "", transcript_name = "";
	
	char strand = items[sp].getStrand();
	string seqname = items[sp].getSeqName();
	string left = items[sp].getLeft();

	vec.clear();
	cur_s = cur_e = -1;
	for (int i = sp; i <= ep; i++) {
		int start = items[i].getStart();
		int end = items[i].getEnd();

		general_assert(strand == items[i].getStrand(), "According to the GTF file given, transcript " + transcript_id + " has exons from different orientations!");
		general_assert(seqname == items[i].getSeqName(), "According to the GTF file given, transcript " + transcript_id + " has exons on multiple chromosomes!");

		if (items[i].getGeneName() != "") {
		  if (gene_name == "") gene_name = items[i].getGeneName();
		  else general_assert(gene_name == items[i].getGeneName(), "Transcript " + transcript_id + " is associated with multiple gene names!");
		}
		if (items[i].getTranscriptName() != "") {
		  if (transcript_name == "") transcript_name = items[i].getTranscriptName();
		  else general_assert(transcript_name == items[i].getTranscriptName(), "Transcript " + transcript_id + " is associated with multiple transcript names!");
		}

		if (cur_e + 1 < start) {
			if (cur_s > 0) vec.push_back(Interval(cur_s, cur_e));
			cur_s = start;
		}
		cur_e = (cur_e < end ? end : cur_e);
	}
	if (cur_s > 0) vec.push_back(Interval(cur_s, cur_e));

	transcripts.add(Transcript(transcript_id, gene_id, seqname, strand, vec, left, transcript_name, gene_name));

	return true;
}

void parse_gtf_file(char* gtfF) {
	ifstream fin(gtfF);