Generated two index multi-Fasta files instead, one without 'N' converted to 'G' and...

diff --git a/preRef.cpp b/preRef.cpp
index 7518ec41d184be27a637af8f756e7c6d46f6e32f..da6c84e3122730e2f72e1c699e5629d8ed27aed2 100644 (file)
--- a/preRef.cpp
+++ b/preRef.cpp
@@ -23,18 +23,17 @@ PolyARules rules;
 Refs refs;
 
 ofstream fout;
-char refF[STRLEN], alignerFastaF[STRLEN], transF[STRLEN];
+char refF[STRLEN], idxF[STRLEN], n2g_idxF[STRLEN];
 
 int polyAChoice, polyALen;
 char exceptionF[STRLEN];
-bool ntog; // true , change N into G; false do not change. Default is true. 
 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("USAGE : rsem-preref refFastaF polyAChoice refName [-l polyALen] [-f exceptionF] [-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");
@@ -42,9 +41,7 @@ int main(int argc, char* argv[]) {
                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);
        }
 
@@ -58,7 +55,6 @@ int main(int argc, char* argv[]) {
        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; }
        }
 
@@ -73,21 +69,21 @@ int main(int argc, char* argv[]) {
        sprintf(refF, "%s.seq", argv[3]);
        refs.saveRefs(refF);
 
-       sprintf(transF, "%s.transcripts.fa", argv[3]);
-       fout.open(transF);
+       sprintf(idxF, "%s.idx.fa", argv[3]);
+       fout.open(idxF);
        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);
+       if (verbose) printf("%s is generated!\n", idxF);
 
-       sprintf(alignerFastaF, "%s.idx.fa", argv[3]);
-       fout.open(alignerFastaF);
+       sprintf(n2g_idxF, "%s.n2g.idx.fa", argv[3]);
+       fout.open(ng2_idxF);
        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<< ">"<< refs.getRef(i).getName()<< endl<< aligner_refp.convert(refs.getRef(i).getSeq())<< endl;
        }
        fout.close();
-       if (verbose) printf("%s is generated!\n", alignerFastaF);
+       if (verbose) printf("%s is generated!\n", n2g_idxF);
 
        return 0;
 }

diff --git a/rsem-prepare-reference b/rsem-prepare-reference
index a74d97d194fce454a99b8714284d67fbf2d3dcfc..177cc43c83e35fa8e3f4bc3dbd475cc964a6016f 100755 (executable)
--- a/rsem-prepare-reference
+++ b/rsem-prepare-reference
@@ -19,9 +19,9 @@ my $polyAChoice = 0; # 0, default, pad polyA tails for all isoforms; 1, --no-pol
 my $no_polyA = 0; # for option --no-polyA 
 my $subsetFile = "";
 my $polyALen = 125;
+my $ntog = 0; 
+my $bowtie = 0;
 my $bowtie_path = "";
-my $no_bowtie = 0;
-my $no_ntog = 0; 
 my $bowtie2 = 0;
 my $bowtie2_path = "";
 my $quiet = 0;
@@ -35,9 +35,9 @@ GetOptions("gtf=s" => \$gtfF,
           "no-polyA" => \$no_polyA, 
           "no-polyA-subset=s" => \$subsetFile,
           "polyA-length=i" => \$polyALen,
+          "ntog" => \$no_ntog,
+          "bowtie" => \$bowtie,
           "bowtie-path=s" => \$bowtie_path,
-          "no-bowtie" => \$no_bowtie,
-          "no-ntog" => \$no_ntog,
           "bowtie2" => \$bowtie2,
           "bowtie2-path=s" => \$bowtie2_path,
           "q|quiet" => \$quiet,
@@ -49,12 +49,8 @@ pod2usage(-msg => "--transcript-to-gene-map and --allele-to-gene-map are mutuall
 pod2usage(-msg => "--gtf and --allele-to-gene-map are mutually exclusive!", -exitval => 2, -verbose => 2) if (($gtfF ne "") && ($alleleMappingF ne ""));
 pod2usage(-msg => "Invalid number of arguments!", -exitval => 2, -verbose => 2) if (scalar(@ARGV) != 2);
 
-if ($bowtie2) { $no_bowtie = 1; $no_ntog = 1; }
-
-pod2usage(-msg => "If bowtie is used, --no-ntog cannot be set!", -exitval => 2, -verbose => 2) if (!$no_bowtie && $no_ntog);
-
-if ($no_bowtie && ($bowtie_path ne "")) { print "Warning: If bowtie is not used, no need to set --bowtie-path option!\n"; }
-if (!$bowtie2 && ($bowtie2_path ne "")) { print "Warning: If bowtie2 is not used, no need to set --bowtie2-path option!\n"; }
+if (!$bowtie && ($bowtie_path ne "")) { print "Warning: If Bowtie is not used, no need to set --bowtie-path option!\n"; }
+if (!$bowtie2 && ($bowtie2_path ne "")) { print "Warning: If Bowtie 2 is not used, no need to set --bowtie2-path option!\n"; }
  
 my $type;