Issue #24 | return 0 on --help
Torsten Seemann
6 years ago
| 8 | 8 | # global variables |
| 9 | 9 | |
| 10 | 10 | my $EXE = $FindBin::RealScript; |
| 11 | my $VERSION = "0.8"; | |
| 11 | my $VERSION = "0.9-dev"; | |
| 12 | 12 | my $DESC = "rapid ribosomal RNA prediction"; |
| 13 | 13 | my $AUTHOR = 'Torsten Seemann <torsten.seemann@gmail.com>'; |
| 14 | 14 | my $URL = 'https://github.com/tseemann/barrnap'; |
| 26 | 26 | # . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . |
| 27 | 27 | # command line options |
| 28 | 28 | |
| 29 | my(@Options, $quiet, $kingdom, $threads, $evalue, $lencutoff, $reject, $incseq); | |
| 29 | my(@Options, $quiet, $kingdom, $threads, $evalue, $lencutoff, $reject, $incseq, $outseq); | |
| 30 | 30 | setOptions(); |
| 31 | 31 | |
| 32 | 32 | # . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . |
| 57 | 57 | msg("Will reject genes < $reject of expected length."); |
| 58 | 58 | |
| 59 | 59 | my $kdom = $KINGDOM{ lc substr($kingdom,0,1) } or |
| 60 | err("I don't recognise --kingdom '$kingdom'. Try: bac arc euk mito"); | |
| 60 | err("I don't recognise --kingdom '$kingdom'. Try:", values(%KINGDOM) ); | |
| 61 | 61 | |
| 62 | 62 | my $hmmdb = "$DBDIR/$kdom.hmm"; |
| 63 | 63 | err("Can't find database: $hmmdb") unless -r $hmmdb; |
| 70 | 70 | $fasta && -r $fasta or err("Usage: $EXE <file.fasta>"); |
| 71 | 71 | |
| 72 | 72 | msg("Scanning $fasta for $kdom rRNA genes... please wait"); |
| 73 | my $cmd = "$NHMMER --cpu $threads -E $evalue --w_length $MAXLEN -o /dev/null --tblout /dev/stdout \Q$hmmdb\E \Q$fasta\E"; | |
| 73 | my $cmd = "$NHMMER --cpu $threads -E $evalue --w_length $MAXLEN -o /dev/null --tblout /dev/stdout \Q$hmmdb\E \Q$fasta\E"; | |
| 74 | 74 | msg("Command: $cmd"); |
| 75 | 75 | my @hits = qx($cmd 2>&1); |
| 76 | 76 | |
| 77 | 77 | # . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . |
| 78 | 78 | # process the output |
| 79 | 79 | |
| 80 | my @hitseq; | |
| 80 | 81 | my @feat; |
| 81 | 82 | HIT: |
| 82 | 83 | foreach (@hits) { |
| 139 | 140 | print while (<FASTA>); # `cat $fasta` |
| 140 | 141 | } |
| 141 | 142 | |
| 143 | if ($outseq) { | |
| 144 | msg("Saving hit sequences to '$outseq'"); | |
| 145 | open my $OUT, '>', $outseq; | |
| 146 | for my $h (@hitseq) { | |
| 147 | print $OUT ">",$h->[0],"\n",$h->[1],"\n"; | |
| 148 | } | |
| 149 | close $OUT; | |
| 150 | } | |
| 151 | ||
| 142 | 152 | #---------------------------------------------------------------------- |
| 143 | 153 | |
| 144 | 154 | sub msg { |
| 170 | 180 | |
| 171 | 181 | If you use Barrnap in your work, please cite: |
| 172 | 182 | |
| 173 | Seemann T (2013) | |
| 183 | Seemann T (2017) | |
| 174 | 184 | $EXE $VERSION : $DESC |
| 175 | 185 | $URL |
| 176 | 186 | |
| 192 | 202 | {OPT=>"help", VAR=>\&usage, DESC=>"This help"}, |
| 193 | 203 | {OPT=>"version", VAR=>\&version, DESC=>"Print version and exit"}, |
| 194 | 204 | {OPT=>"citation",VAR=>\&show_citation, DESC=>"Print citation for referencing $EXE"}, |
| 195 | {OPT=>"kingdom=s", VAR=>\$kingdom, DEFAULT=>'bac', | |
| 196 | DESC=>"Kingdom: ".join(' ', values %KINGDOM) }, | |
| 205 | {OPT=>"kingdom=s", VAR=>\$kingdom, DEFAULT=>'bac', DESC=>"Kingdom: ".join(' ', values %KINGDOM) }, | |
| 197 | 206 | {OPT=>"quiet!", VAR=>\$quiet, DEFAULT=>0, DESC=>"No screen output"}, |
| 198 | 207 | {OPT=>"threads=i", VAR=>\$threads, DEFAULT=>8, DESC=>"Number of threads/cores/CPUs to use"}, |
| 199 | 208 | {OPT=>"lencutoff=f",VAR=>\$lencutoff, DEFAULT=>0.8, DESC=>"Proportional length threshold to label as partial"}, |
| 200 | 209 | {OPT=>"reject=f",VAR=>\$reject, DEFAULT=>0.5, DESC=>"Proportional length threshold to reject prediction"}, |
| 201 | 210 | {OPT=>"evalue=f",VAR=>\$evalue, DEFAULT=>1E-6, DESC=>"Similarity e-value cut-off"}, |
| 202 | 211 | {OPT=>"incseq!", VAR=>\$incseq, DEFAULT=>0, DESC=>"Include FASTA input sequences in GFF3 output"}, |
| 212 | # {OPT=>"outseq!", VAR=>\$outseq, DEFAULT=>'', DESC=>"Save hits to this FASTA file"}, | |
| 203 | 213 | ); |
| 204 | 214 | |
| 205 | (!@ARGV) && (usage()); | |
| 206 | ||
| 207 | &GetOptions(map {$_->{OPT}, $_->{VAR}} grep { ref } @Options) || usage(); | |
| 215 | (!@ARGV) && (usage(1)); | |
| 216 | ||
| 217 | &GetOptions(map {$_->{OPT}, $_->{VAR}} grep { ref } @Options) || usage(1); | |
| 208 | 218 | |
| 209 | 219 | # Now setup default values. |
| 210 | 220 | foreach (@Options) { |
| 217 | 227 | #---------------------------------------------------------------------- |
| 218 | 228 | |
| 219 | 229 | sub usage { |
| 220 | print STDERR "Synopsis:\n $EXE $VERSION - $DESC\n"; | |
| 221 | print STDERR "Author:\n $AUTHOR\n"; | |
| 222 | print STDERR "Usage:\n $EXE [options] <chromosomes.fasta>\n"; | |
| 230 | my($exitcode) = @_; | |
| 231 | $exitcode = 0 if $exitcode eq 'help'; # what gets passed by getopt func ref | |
| 232 | $exitcode ||= 0; | |
| 233 | select STDERR if $exitcode; # write to STDERR if exitcode is error | |
| 234 | ||
| 235 | print "Synopsis:\n $EXE $VERSION - $DESC\n"; | |
| 236 | print "Author:\n $AUTHOR\n"; | |
| 237 | print "Usage:\n $EXE [options] <chromosomes.fasta>\n"; | |
| 223 | 238 | foreach (@Options) { |
| 224 | 239 | if (ref) { |
| 225 | 240 | my $def = defined($_->{DEFAULT}) ? " (default '$_->{DEFAULT}')" : ""; |
| 229 | 244 | $opt =~ s/=s$/ [X]/; |
| 230 | 245 | $opt =~ s/=i$/ [N]/; |
| 231 | 246 | $opt =~ s/=f$/ [n.n]/; |
| 232 | printf STDERR " --%-15s %s%s\n", $opt, $_->{DESC}, $def; | |
| 247 | printf " --%-15s %s%s\n", $opt, $_->{DESC}, $def; | |
| 233 | 248 | } |
| 234 | 249 | else { |
| 235 | print STDERR "$_\n"; | |
| 250 | print "$_\n"; | |
| 236 | 251 | } |
| 237 | 252 | } |
| 238 | exit(1); | |
| 239 | } | |
| 240 | ||
| 241 | #---------------------------------------------------------------------- | |
| 253 | exit($exitcode); | |
| 254 | } | |
| 255 | ||
| 256 | #---------------------------------------------------------------------- | |