X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=preRef.cpp;h=7518ec41d184be27a637af8f756e7c6d46f6e32f;hb=0f8fbc39759d8a8a4fa755903b8bde403b0da6c9;hp=65e034b8c17e44e0e7716ca5f387dad9443acbf5;hpb=a95154919f950f86de9104b2b9dcf1f0c7e83387;p=rsem.git

diff --git a/preRef.cpp b/preRef.cpp
index 65e034b..7518ec4 100644
--- a/preRef.cpp
+++ b/preRef.cpp
@@ -23,7 +23,7 @@ PolyARules rules;
 Refs refs;
 
 ofstream fout;
-char refF[STRLEN], alignerFastaF[STRLEN];
+char refF[STRLEN], alignerFastaF[STRLEN], transF[STRLEN];
 
 int polyAChoice, polyALen;
 char exceptionF[STRLEN];
@@ -33,54 +33,61 @@ bool quiet; // verbose = !quiet;
 // always generate references for aligners, default convert all N into G
 int main(int argc, char* argv[]) {
 
-  if (argc < 4) {
-    printf("USAGE : rsem-preref refFastaF polyAChoice refName [-l polyALen] [-f exceptionF] [--no-ntog] [-q]\n\n");
-    printf("  refFastaF: a FASTA format file contains all reference transcripts\n");
-    printf("  polyAChoice: choice for polyA tail padding.It is a number from {0,1,2}\n");
-    printf("    0: pad polyA tail\n");
-    printf("    1: do not pad polyA tail at all\n");
-    printf("    2: pad polyA tail for all references but those in exceptionF\n");
-    printf("  -l: polyALen: specify the length of polyA tail you want to pad. Default is 100\n");
-    printf("  -f: exceptionF: file contains a list of exception reference ids. IDs starts from 1. Must set if polyAChoice = 2\n");
-    printf("  --no-ntog: do not convert N in references into G\n");
-    printf("  -q: quiet\n");
-    exit(-1);
-  }
-
-
-  polyAChoice = atoi(argv[2]);
-
-  polyALen = 125;
-  ntog = true;
-  quiet = false;
-  memset(exceptionF, 0, sizeof(exceptionF));
-
-  for (int i = 4; i < argc; i++) {
-	  if (!strcmp(argv[i], "-l")) { polyALen = atoi(argv[i + 1]); }
-	  if (!strcmp(argv[i], "-f")) { strcpy(exceptionF, argv[i + 1]); }
-	  if (!strcmp(argv[i], "--no-ntog")) { ntog = false; }
-	  if (!strcmp(argv[i], "-q")) { quiet = true; }
-  }
-
-  verbose = !quiet;
-
-  //make references
-  rules = PolyARules(polyAChoice, polyALen, exceptionF);
-  refs.makeRefs(argv[1], refp, rules);
-  M = refs.getM();
-
-  //save references
-  sprintf(refF, "%s.seq", argv[3]);
-  refs.saveRefs(refF);
-
-  sprintf(alignerFastaF, "%s.idx.fa", argv[3]);
-  fout.open(alignerFastaF);
-  for (int i = 1; i <= M; i++) {
-    fout<<">"<<refs.getRef(i).getName()<<endl<<(ntog ? aligner_refp.convert(refs.getRef(i).getSeq()) : refs.getRef(i).getSeq())<<endl;
-  }
-  fout.close();
-  
-  if (verbose) { printf("IDX.fa generated!\n"); }
-
-  return 0;
+	if (argc < 4) {
+		printf("USAGE : rsem-preref refFastaF polyAChoice refName [-l polyALen] [-f exceptionF] [--no-ntog] [-q]\n\n");
+		printf("  refFastaF: a FASTA format file contains all reference transcripts\n");
+		printf("  polyAChoice: choice for polyA tail padding.It is a number from {0,1,2}\n");
+		printf("    0: pad polyA tail\n");
+		printf("    1: do not pad polyA tail at all\n");
+		printf("    2: pad polyA tail for all references but those in exceptionF\n");
+		printf("  -l: polyALen: specify the length of polyA tail you want to pad. Default is 100\n");
+		printf("  -f: exceptionF: file contains a list of exception reference ids. IDs starts from 1. Must set if polyAChoice = 2\n");
+		printf("  --no-ntog: do not convert N in references into G\n");
+		printf("  -q: quiet\n");
+		printf("  This program will generate a file named \"refName.transcripts.fa\", which may rewrite an existing file (e.g. refFastaF).\n");
+		exit(-1);
+	}
+
+	polyAChoice = atoi(argv[2]);
+
+	polyALen = 125;
+	ntog = true;
+	quiet = false;
+	memset(exceptionF, 0, sizeof(exceptionF));
+
+	for (int i = 4; i < argc; i++) {
+		if (!strcmp(argv[i], "-l")) { polyALen = atoi(argv[i + 1]); }
+		if (!strcmp(argv[i], "-f")) { strcpy(exceptionF, argv[i + 1]); }
+		if (!strcmp(argv[i], "--no-ntog")) { ntog = false; }
+		if (!strcmp(argv[i], "-q")) { quiet = true; }
+	}
+
+	verbose = !quiet;
+
+	//make references
+	rules = PolyARules(polyAChoice, polyALen, exceptionF);
+	refs.makeRefs(argv[1], refp, rules);
+	M = refs.getM();
+
+	//save references
+	sprintf(refF, "%s.seq", argv[3]);
+	refs.saveRefs(refF);
+
+	sprintf(transF, "%s.transcripts.fa", argv[3]);
+	fout.open(transF);
+	for (int i = 1; i <= M; i++) {
+		fout<< ">"<< refs.getRef(i).getName()<< endl<< refs.getRef(i).getSeq()<< endl;
+	}
+	fout.close();
+	if (verbose) printf("%s is generated!\n", transF);
+
+	sprintf(alignerFastaF, "%s.idx.fa", argv[3]);
+	fout.open(alignerFastaF);
+	for (int i = 1; i <= M; i++) {
+		fout<<">"<<refs.getRef(i).getName()<<endl<<(ntog ? aligner_refp.convert(refs.getRef(i).getSeq()) : refs.getRef(i).getSeq())<<endl;
+	}
+	fout.close();
+	if (verbose) printf("%s is generated!\n", alignerFastaF);
+
+	return 0;
 }