Skip to content

Commit

Permalink
Fixing Issue #9 and other minor changes
Browse files Browse the repository at this point in the history
  • Loading branch information
@biocoder
biocoder committed Feb 14, 2018
1 parent dd65469 commit 262bfba
Show file tree
Hide file tree
Showing 3 changed files with 50 additions and 23 deletions.
10 changes: 8 additions & 2 deletions NGS-Utils/README.pod
View file
Original file line number Diff line number Diff line change
Expand Up @@ -737,12 +737,18 @@ features from C<< --cat-ncRNAs >> module.

Run the C<< CPC >> module. No options are required.

=item --blast-strand (Optional)

Mention the query strand to search against the database. 1 = plus strand, 2 = minus strand, 3 = both.

Default: 1

=item --skip-cpc

Skip runnning C<< CPC >> altogether. Use this option, if you think CPC is flagging a lot of
your transcripts as B<I<coding>> and instead rely only on Infernal search results.

=item --skip-blastall-core
=item --skip-cpc-core

Skip runnning core C<< CPC >> process once you know you have output from C<< CPC >>.
This option can be used when lncRNApipe fails for some reason after C<< blastall >>
Expand Down Expand Up @@ -831,6 +837,6 @@ This program is distributed under the Artistic License 2.0.

=head1 DATE

Feb-12-2018
Feb-14-2018

=cut
10 changes: 5 additions & 5 deletions NGS-Utils/categorize_ncRNAs.pl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@
use Fcntl qw / :flock SEEK_END /;

my ($LASTCHANGEDBY) = q$LastChangedBy: konganti $ =~ m/.+?\:(.+)/;
my ($LASTCHANGEDDATE) = q$LastChangedDate: 2016-01-26 12:11:27 -0500 (Tue, 26 Jan 2016) $ =~ m/.+?\:(.+)/;
my ($LASTCHANGEDDATE) = q$LastChangedDate: 2018-02-14 08:47:27 -0500 (Wed, 14 Feb 2018) $ =~ m/.+?\:(.+)/;
my ($VERSION) = q$LastChangedRevision: 1043 $ =~ m/.+?(\d+)/;
my $AUTHORFULLNAME = 'Kranti Konganti';

Expand Down Expand Up @@ -987,8 +987,8 @@ sub check_gtf_attributes {
qq/chr3\tCufflinks\ttranscript\t30549662\t30551349\t1000\t-\t.\tgene_id "CUFF.22498"; transcript_id "CUFF.22498.1"; FPKM "2.5052666329"; frac "1.000000"; conf_lo "1.676755"; conf_hi "3.353509"; cov "4.749121";\n/ .
qq/chr3\tCufflinks\texon\t30549662\t30550273\t1000\t-\t.\tgene_id "CUFF.22498"; transcript_id "CUFF.22498.1"; exon_number "1"; FPKM "2.5052666329"; frac "1.000000"; conf_lo "1.676755"; conf_hi "3.353509"; cov "4.749121";\n/ .
qq/chr3\tCufflinks\texon\t30551033\t30551349\t1000\t-\t.\tgene_id "CUFF.22498"; transcript_id "CUFF.22498.1"; exon_number "2"; FPKM "2.5052666329"; frac "1.000000"; conf_lo "1.676755"; conf_hi "3.353509"; cov "4.749121";\n/ .
qq/\nYour File:\n----------\n/ . $io->execute_get_sys_cmd_output("head -n 3 $file"), 'INFO!');
$io->warning('Sit back and relax. We got it covered ... Formatting the GTF file to process with lncRNApipe.', 'INFO!');
qq/\nYour File:\n----------\n/ . $io->execute_get_sys_cmd_output("head -n 3 $file"), 'INFO!!');
$io->warning('Sit back and relax. We got it covered ... Formatting the GTF file to process with lncRNApipe.', 'INFO!!');
$file = format_gtf($file, $label);
}
return $file;
Expand Down Expand Up @@ -1039,7 +1039,7 @@ sub format_gtf {
sub check_for_zero_class {
for (0 .. $#ARGV) {
if (-e $get_putative_ncRNAs_chkpt && !-e $p_file_names_gtf->[$_]) {
$io->warning("We could not find any requested class codes [ in $cuffcmp ] ...\nBailing out!");
#$io->warning("We could not find any requested class codes [ in $cuffcmp ] ...\nBailing out!");
$io->execute_system_command("touch $cat_class_zero");
exit 0;
}
Expand Down Expand Up @@ -1293,6 +1293,6 @@ =head1 COPYRIGHT
=head1 DATE
Jan-26-2016
Feb-14-2018
=cut
53 changes: 37 additions & 16 deletions NGS-Utils/lncRNApipe
View file
Original file line number Diff line number Diff line change
Expand Up @@ -12,8 +12,8 @@ use Data::Dumper;
Getopt::Long::Configure('prefix=--');

my ($LASTCHANGEDBY) = q$LastChangedBy: konganti $ =~ m/.+?\:(.+)/;
my ($LASTCHANGEDDATE) = q$LastChangedDate: 2018-02-12 09:27:27 -0500 (Mon, 12 February 2018) $ =~ m/.+?\:(.+)/;
my ($VERSION) = q$LastChangedRevision: 1.2.2 $ =~ m/.+?\:\s*(.*)\s*.*/;
my ($LASTCHANGEDDATE) = q$LastChangedDate: 2018-02-14 11:27:27 -0500 (Mon, 14 Feb 2018) $ =~ m/.+?\:(.+)/;
my ($VERSION) = q$LastChangedRevision: 1.2.3 $ =~ m/.+?\:\s*(.*)\s*.*/;
my $AUTHORFULLNAME = 'Kranti Konganti';

my ($help, $quiet, $setup, $get_uq_sc_opts,
Expand All @@ -25,7 +25,7 @@ my ($help, $quiet, $setup, $get_uq_sc_opts,
$lncRNApipe_ver_info, $no_update_check,
$skip_get_uq, $lncRNApipe_succ, $num_cpu_by_2,
$skip_cpc, $debug, $sp_list, $pipe_config,
$conf_run_id, $setup_compiler,
$conf_run_id, $setup_compiler, $blast_strand,
$cpanm_args, $donot_wait4jobIDs);

my $kill_jobs_thru_conf = my $show_lncrna_counts = [];
Expand All @@ -38,11 +38,12 @@ my $is_valid_option = GetOptions('help:s' => \$help,
'fetch-seq:s' => \$fetch_sc_opts,
'get-uq-feat=s' => \$get_uq_sc_opts,
'cpc' => \$start_cpc,
'blast-strand:i' => \$blast_strand,
'rnafold:s' => \$start_rnafold,
'rm-int-plots' => \$rm_int_plots,
'infernal:s' => \$start_infernal,
'cpu=i' => \$num_cpu,
'skip-blastall-core' => \$skip_cpc_core,
'skip-cpc-core' => \$skip_cpc_core,
'skip-rnafold-core' => \$skip_rnafold_core,
'skip-cmscan-core' => \$skip_cmscan_core,
'coverage-infernal|cov-inf=f' => \$inf_cov,
Expand Down Expand Up @@ -208,6 +209,20 @@ sub run_lncRNApipe {

my $dep_tools_fh = $io->open_file('<', $USER_HOME . '/.lncRNApipe.depconf');
get_deps($dep_tools_fh);

# Check blast strand options for CPC.
if (defined $blast_strand && $blast_strand !~ m/^[123]$/) {
$io->error("The binary blastall takes only 1, 2, or 3 for -S argument, wherein\n" .
"1 means search the plus strand of the query sequence\n" .
"2 means search the minus strand of the query sequence\n" .
"3 means earch both strands of the query sequence\n\n" .
"You entered: $blast_strand !");
}
elsif (!defined $blast_strand) {
#$io->warning("Will run CPC using blastall [ with default option: -S 1 ]." .
# "\nUse --blast-strand option with lncRNApipe to change this.", 'INFO!!');
$blast_strand = 1;
}

# User has requested to run the pipeline from assembly stage. Handle it here.
run_lncRNApipe_from_conf($io, $shall_I_run, $s_time) if (ref($shall_I_run) eq 'HASH');
Expand Down Expand Up @@ -317,7 +332,7 @@ sub run_lncRNApipe {
$gtfToGenePred_failed =~ s/[\s\r\n]+//g;

if ($gtfToGenePred_failed =~ m/exonFramesfieldisbeingadded/i) {
$io->warning('Trying without -genePredExt ...', 'INFO!');
$io->warning('Trying without -genePredExt ...', 'INFO!!');
$gtfToGenePred_failed = $is_failed = $io->execute_get_sys_cmd_output($deps->{'bin-gtfToGenePred'} . ' -geneNameAsName2 ' .
$annot . ' ' . $cat_dir . '/' . $io->file_basename($annot) . '.txt',
"Command call:\n" .
Expand Down Expand Up @@ -367,7 +382,7 @@ sub run_lncRNApipe {

# Rare case exit code. When categorize_ncRNAs.pl could not find cuffcompare class codes requested by user or pipeline.
if (-e $cat_class_zero) {
$io->warning("Could not find any cuffcompare class codes as requested [ in $cuffcmp_dir/lncRNApipe_cuffcmp.tracking ] ...", 'INFO!');
$io->warning("We could not find any cuffcompare class codes as requested [ in $cuffcmp_dir/lncRNApipe_cuffcmp.tracking ] ...", 'INFO!!');
$lncRNApipe_succ = 1;

undef $fetch_sc_opts;
Expand Down Expand Up @@ -598,13 +613,13 @@ sub run_lncRNApipe {
$cpc_cpu = $num_cpu_by_2 if (defined $num_cpu);

if (!defined $skip_cpc_core) {
$io->execute_system_command('NUM_CPU=' . $cpc_cpu . ' ' . $deps->{'cpc'} .
' "' . $lncRNApipe_unique_seq_file . ' ' . $CPC_out_file . ' ' . $cpc_work_dir . ' ' . $cpc_work_dir . '" ' .
$blastall_path->($deps->{'blastall'}),
$io->execute_system_command('BLAST_STRAND=' . $blast_strand . ' NUM_CPU=' . $cpc_cpu . ' ' . $deps->{'cpc'} .
' "' . $lncRNApipe_unique_seq_file . ' ' . $CPC_out_file . ' ' . $cpc_work_dir . ' ' .
$cpc_work_dir . '" ' . $blastall_path->($deps->{'blastall'}),
"Command call:\n" .
"-------------\n" . 'NUM_CPU=' . $cpc_cpu . ' ' . $deps->{'cpc'} .
' "' . $lncRNApipe_unique_seq_file . ' ' . $CPC_out_file . ' ' . $cpc_work_dir . ' ' . $cpc_work_dir . '" ' .
$blastall_path->($deps->{'blastall'}));
"-------------\n" . 'BLAST_STRAND=' . $blast_strand . ' NUM_CPU=' . $cpc_cpu . ' ' . $deps->{'cpc'} .
' "' . $lncRNApipe_unique_seq_file . ' ' . $CPC_out_file . ' ' . $cpc_work_dir . ' ' .
$cpc_work_dir . '" ' . $blastall_path->($deps->{'blastall'}));
}
} if (defined $start_cpc);

Expand Down Expand Up @@ -798,7 +813,7 @@ sub run_lncRNApipe {
"Truncated output from $cat_dir/*.CPC.predict.txt");

$io->error("Could not get final putative lncRNA count.\n" .
"Try re-running the final module. Take a look at --skip-blastall-core, --skip-rnafold-core and --skip-cmscan-core options." .
"Try re-running the final module. Take a look at --skip-cpc-core, --skip-rnafold-core and --skip-cmscan-core options." .
"\nThis may also mean that the putative lncRNAs that were used with CPC may have all been flagged as \"coding\".");
}
else {
Expand Down Expand Up @@ -2708,7 +2723,7 @@ sub get_job_dependency_chain {
"sleep 10; else break; fi; done;";
}
elsif ($file =~ m/job_IDs_cufflinks/i) {
$job_deps_str = "while [ true ]; do if [ -e \"$check_file_presence\" ] && [ -s \"$check_file_presence\" ]; then " .
$job_deps_str = "while [ true ]; do if [ -e \"$check_file_presence\" ] && [ -s \"$check_file_presence\" ] && [ ! -z \"\$(ps aux | grep -P 'cufflinks.+?($job_deps_str)' | grep -v grep)\" ]; then " .
"break; else sleep 10; fi; done;";
}
}
Expand Down Expand Up @@ -3773,12 +3788,18 @@ features from C<< --cat-ncRNAs >> module.
Run the C<< CPC >> module. No options are required.
=item --blast-strand (Optional)
Mention the query strand to search against the database. 1 = plus strand, 2 = minus strand, 3 = both.
Default: 1
=item --skip-cpc
Skip runnning C<< CPC >> altogether. Use this option, if you think CPC is flagging a lot of
your transcripts as B<I<coding>> and instead rely only on Infernal search results.
=item --skip-blastall-core
=item --skip-cpc-core
Skip runnning core C<< CPC >> process once you know you have output from C<< CPC >>.
This option can be used when lncRNApipe fails for some reason after C<< blastall >>
Expand Down Expand Up @@ -3867,6 +3888,6 @@ This program is distributed under the Artistic License 2.0.
=head1 DATE
Feb-12-2018
Feb-14-2018
=cut

0 comments on commit 262bfba

Please sign in to comment.