test_pipeline.pl

#!/usr/bin/perl -w
use strict;

my $PIPELINE = "./centroid_homfold.pl";

# test (1)
print "================================================================================\n";
print "1) Secondary structure prediction by using pre-compiled databse\n";
print "================================================================================\n";
my $test1 = "$PIPELINE --log --tmpdir ./tmp/test1 -D ./db/Rfam.seed.99.db -S ./db/Rfam.seed.99.fasta -g 8 -e 0.01 -n 30 ./test/1.fa";
print "$test1\n";
system ($test1);

# test (2)
print "================================================================================\n";
print "2) Secondary structure prediction by using user-specified RNA sequence (NOT using pre-compiled database)\n";
print "================================================================================\n";
my $test2 = "$PIPELINE --log --tmpdir ./tmp/test2 -S ./db/Rfam.seed.99.fasta -g 8 -e 0.01 -n 30 ./test/1.fa";
print "$test2\n";
system ($test2);

# test (3)
print "================================================================================\n";
print "3) Secondary structure prediction by using user-specified homologous sequence (like CentroidHomfold does)\n";
print "================================================================================\n";
my $test3 = "$PIPELINE --log --tmpdir ./tmp/test3 -H ./test/homologous.fna -g 8 -e 0.01 -n 30 ./test/target.fna";
print "$test3\n";
system ($test3);

Output

================================================================================
1) Secondary structure prediction by using pre-compiled databse
================================================================================
./centroid_homfold.pl --log --tmpdir ./tmp/test1 -D ./db/Rfam.seed.99.db -S ./db/Rfam.seed.99.fasta -g 8 -e 0.01 -n 30 ./test/1.fa
Input: ./test/1.fa, Length of sequence: 117
/db01/mhamada/centroidhomfold-last-v1.0/scripts/U2T.pl ./test/1.fa > ./tmp/test1/1.fa.dna
/db01/mhamada/centroidhomfold-last-v1.0/external/lastdb -x  ./tmp/test1/1.fa ./tmp/test1/1.fa.dna
/db01/mhamada/centroidhomfold-last-v1.0/external/lastex -s1 -E 0.01 ./tmp/test1/1.fa.prj ./db/Rfam.seed.99.db.prj >& ./tmp/test1/1.fa.lastex
Lastal score: 21
/db01/mhamada/centroidhomfold-last-v1.0/external/lastal -s1 -e 21 -o ./tmp/test1/1.fa.lastal ./db/Rfam.seed.99.db ./tmp/test1/1.fa.dna >& ./tmp/test1/1.fa.lastal.log
lastal end: 0.497855 sec
/db01/mhamada/centroidhomfold-last-v1.0/scripts/select_homologous.pl ./tmp/test1/1.fa.lastal ./db/Rfam.seed.99.fasta /db01/mhamada/centroidhomfold-last-v1.0/scripts | /db01/mhamada/centroidhomfold-last-v1.0/scripts/T2U.pl > ./tmp/test1/1.fa.homol.all
select_homologous end: 1.886235 sec
== Remove redundancy ==
/db01/mhamada/centroidhomfold-last-v1.0/scripts/remove_redundancy.pl ./tmp/test1/1.fa.homol.all > ./tmp/test1/1.fa.homol.red
== Select 30 sequences from top scores ==
/db01/mhamada/centroidhomfold-last-v1.0/scripts/getSeqFromTop.pl 30 ./tmp/test1/1.fa.homol.red > ./tmp/test1/1.fa.homol
reduce_homologous end: 0.317214 sec
/db01/mhamada/centroidhomfold-last-v1.0/external/centroid_homfold -g 8 -H ./tmp/test1/1.fa.homol ./test/1.fa --engine_a CONTRAlign --engine_s McCaskill --param /db01/mhamada/centroidhomfold-last-v1.0/param/parameters_BLstar_Vienna.txt | /db01/mhamada/centroidhomfold-last-v1.0/scripts/remove_ncbp.pl > ./tmp/test1/1.fa.output
cat ./tmp/test1/1.fa.output
> RF00001_A
UUCGGUGGUUAUAGCGGUGGGGAAACACCCGGUCCCAUUCCGAACCCGGUAGUUAAGCCCGCCAGCGCCGAUGGUACUGCACUGGUGACGGUGUGGGAGAGUAGGUCGCCGCCGGAC
(((((((((....((((((((.....((((((((.......))..))))..))....)))))).)).((.((....((((((((....))))))))....)).))..))))))))). (g=8,th=0.111111,e=-23.8)
centroid_homfold end: 1.407878 sec
Fnish: 4.250788 sec
================================================================================
2) Secondary structure prediction by using user-specified RNA sequence (NOT using pre-compiled database)
================================================================================
./centroid_homfold.pl --log --tmpdir ./tmp/test2 -S ./db/Rfam.seed.99.fasta -g 8 -e 0.01 -n 30 ./test/1.fa
Input: ./test/1.fa, Length of sequence: 117
/db01/mhamada/centroidhomfold-last-v1.0/scripts/U2T.pl ./test/1.fa > ./tmp/test2/1.fa.dna
== Generate LAST database (it might take a few minutes) ==
== You can specify pre-computed database with -D option ==
== for skipping this step                               ==
/db01/mhamada/centroidhomfold-last-v1.0/scripts/U2T.pl ./db/Rfam.seed.99.fasta > ./tmp/test2/Rfam.seed.99.fasta.dna
/db01/mhamada/centroidhomfold-last-v1.0/external/lastdb ./tmp/test2/Rfam.seed.99.fasta.db ./tmp/test2/Rfam.seed.99.fasta.dna
/db01/mhamada/centroidhomfold-last-v1.0/scripts/mkidx.pl ./db/Rfam.seed.99.fasta > ./tmp/test2/Rfam.seed.99.fasta.ary
sh: /db01/mhamada/centroidhomfold-last-v1.0/scripts/mkidx.pl: No such file or directory
/db01/mhamada/centroidhomfold-last-v1.0/external/lastdb -x  ./tmp/test2/1.fa ./tmp/test2/1.fa.dna
/db01/mhamada/centroidhomfold-last-v1.0/external/lastex -s1 -E 0.01 ./tmp/test2/1.fa.prj ./tmp/test2/Rfam.seed.99.fasta.db.prj >& ./tmp/test2/1.fa.lastex
Lastal score: 21
/db01/mhamada/centroidhomfold-last-v1.0/external/lastal -s1 -e 21 -o ./tmp/test2/1.fa.lastal ./tmp/test2/Rfam.seed.99.fasta.db ./tmp/test2/1.fa.dna >& ./tmp/test2/1.fa.lastal.log
lastal end: 0.469974 sec
/db01/mhamada/centroidhomfold-last-v1.0/scripts/select_homologous.pl ./tmp/test2/1.fa.lastal ./db/Rfam.seed.99.fasta /db01/mhamada/centroidhomfold-last-v1.0/scripts | /db01/mhamada/centroidhomfold-last-v1.0/scripts/T2U.pl > ./tmp/test2/1.fa.homol.all
select_homologous end: 1.783726 sec
== Remove redundancy ==
/db01/mhamada/centroidhomfold-last-v1.0/scripts/remove_redundancy.pl ./tmp/test2/1.fa.homol.all > ./tmp/test2/1.fa.homol.red
== Select 30 sequences from top scores ==
/db01/mhamada/centroidhomfold-last-v1.0/scripts/getSeqFromTop.pl 30 ./tmp/test2/1.fa.homol.red > ./tmp/test2/1.fa.homol
reduce_homologous end: 0.234011 sec
/db01/mhamada/centroidhomfold-last-v1.0/external/centroid_homfold -g 8 -H ./tmp/test2/1.fa.homol ./test/1.fa --engine_a CONTRAlign --engine_s McCaskill --param /db01/mhamada/centroidhomfold-last-v1.0/param/parameters_BLstar_Vienna.txt | /db01/mhamada/centroidhomfold-last-v1.0/scripts/remove_ncbp.pl > ./tmp/test2/1.fa.output
cat ./tmp/test2/1.fa.output
> RF00001_A
UUCGGUGGUUAUAGCGGUGGGGAAACACCCGGUCCCAUUCCGAACCCGGUAGUUAAGCCCGCCAGCGCCGAUGGUACUGCACUGGUGACGGUGUGGGAGAGUAGGUCGCCGCCGGAC
(((((((((....((((((((.....((((((((.......))..))))..))....)))))).)).((.((....((((((((....))))))))....)).))..))))))))). (g=8,th=0.111111,e=-23.8)
centroid_homfold end: 1.414908 sec
Fnish: 9.446598 sec
================================================================================
3) Secondary structure prediction by using user-specified homologous sequence (like CentroidHomfold does)
================================================================================
./centroid_homfold.pl --log --tmpdir ./tmp/test3 -H ./test/homologous.fna -g 8 -e 0.01 -n 30 ./test/target.fna
Input: ./test/target.fna, Length of sequence: 68
== Homologous sequences: ./test/homologous.fna is specified ==
/db01/mhamada/centroidhomfold-last-v1.0/external/centroid_homfold -g 8 -H ./test/homologous.fna ./test/target.fna | /db01/mhamada/centroidhomfold-last-v1.0/scripts/remove_ncbp.pl
>X52392.1/16775-16708
GUUCCCGUAGUUGAGAACAACAAUGGCUUUUCAAGCCGUAGUCCUUGGGGUCUAACCAAGCGGGAAUA
((((((((.((((....)))).((((((.....)))))).....((((.......)))))))))))). (g=8,th=0.111111,e=-22.9)