X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=extractRef.cpp;h=2d2b17cea4fde25be35e6c9ab6925ba404144a74;hb=58d504aaf36ae486b1dba6d03e0e9f1c25855037;hp=12b3222744db41cf1602890b9ae0fccfd2dfedca;hpb=0f1ac7d42823cf0f47d25f4f0a04620357421589;p=rsem.git

diff --git a/extractRef.cpp b/extractRef.cpp
index 12b3222..2d2b17c 100644
--- a/extractRef.cpp
+++ b/extractRef.cpp
@@ -50,24 +50,24 @@ map<string, string> mi_table; // mapping info table
 map<string, string>::iterator mi_iter; //mapping info table's iterator
 
 void loadMappingInfo(char* mappingF) {
-  ifstream fin(mappingF);
-  string line, key, value;
-
-  if (!fin.is_open()) {
-	  fprintf(stderr, "Cannot open %s! It may not exist.\n", mappingF);
-	  exit(-1);
-  }
-
-  mi_table.clear();
-  while (getline(fin, line)) {
-    line = cleanStr(line);
-    if (line[0] == '#') continue;
-    istringstream strin(line);
-    strin>>value>>key;
-    mi_table[key] = value;
-  }
-
-  fin.close();
+	ifstream fin(mappingF);
+	string line, key, value;
+
+	if (!fin.is_open()) {
+		fprintf(stderr, "Cannot open %s! It may not exist.\n", mappingF);
+		exit(-1);
+	}
+
+	mi_table.clear();
+	while (getline(fin, line)) {
+		line = cleanStr(line);
+		if (line[0] == '#') continue;
+		istringstream strin(line);
+		strin>>value>>key;
+		mi_table[key] = value;
+	}
+
+	fin.close();
 }
 
 bool buildTranscript(int sp, int ep) {
@@ -86,8 +86,14 @@ bool buildTranscript(int sp, int ep) {
 		int start = items[i].getStart();
 		int end = items[i].getEnd();
 
-		assert(strand == items[i].getStrand());
-		assert(seqname == items[i].getSeqName());
+		if (strand != items[i].getStrand()) {
+		  fprintf(stderr, "According to the GTF file given, a transcript has exons from different orientations!\n");
+		  exit(-1);
+		}
+		if (seqname != items[i].getSeqName()) {
+		  fprintf(stderr, "According to the GTF file given, a transcript has exons on multiple chromosomes!\n");
+		  exit(-1);
+		}
 
 		if (cur_e + 1 < start) {
 			if (cur_s > 0) vec.push_back(Interval(cur_s, cur_e));
@@ -128,14 +134,14 @@ void parse_gtf_file(char* gtfF) {
  			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);
+					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);
  			}
@@ -241,7 +247,7 @@ void writeResults(char* refName) {
 }
 
 int main(int argc, char* argv[]) {
-	if (argc < 6 || (hasMappingFile = atoi(argv[4])) && argc < 7) {
+  if (argc < 6 || ((hasMappingFile = atoi(argv[4])) && argc < 7)) {
 		printf("Usage: rsem-extract-reference-transcripts refName quiet gtfF hasMappingFile [mappingFile] chromosome_file_1 [chromosome_file_2 ...]\n");
 		exit(-1);
 	}
@@ -297,7 +303,9 @@ int main(int argc, char* argv[]) {
 
 	for (int i = 1; i <= M; i++) {
 		if (seqs[i] == "") {
-			fprintf(stderr, "%s's sequence is empty! You must provide all chromosome files of transcripts which are presented in the .gtf file!\n", transcripts.getTranscriptAt(i).getTranscriptID().c_str());
+			const Transcript& transcript = transcripts.getTranscriptAt(i);
+			fprintf(stderr, "Cannot extract transcript %s's sequence from chromosome %s! Loading chromosome %s's sequence is failed. Please check if 1) the chromosome directory is set correctly; 2) the list of chromosome files is complete; 3) the FASTA files containing chromosome sequences are not truncated or having wrong format.\n", \
+				transcript.getTranscriptID().c_str(), transcript.getSeqName().c_str(), transcript.getSeqName().c_str());
 			exit(-1);
 		}
 	}