TarHunter

TarHunter pipeline is implemented in four steps (workflow). For miRBase miRNAs, step 1 is omitted. For eTM and noncoding target prediction, step 2 is omitted. If the cross-species conservation filter is not desired (option -N), step 4 is omitted.

Dependencies include CD-HIT (step 1) , GNU Parallel and MCL (step 2), FASTA or RNAhybrid (step 3). Install appropriate tools for your job.

Simple Test

Under the test folder, run the following command:

perl ../TarHunter.pl -q mir.txt -s spe1.txt -b cds_db/ -o out

Examples

A. CDS target prediction (ortho_mode, one step)

perl ../TarHunter.pl -q mir.txt -s spe1.txt -b cds_db/ -o out1

B. CDS target prediction (ortho_mode, two steps, recommended)
B1. Produce orthologous gene alignment file

perl ../TarHunter.pl -q mir.txt -s spe1.txt -b cds_db/ -p 0 -o ortho

B2. Search conserved miRNA targets

perl ../TarHunter.pl -q mir.txt -s spe1.txt -b cds_db/ -a ortho/ortho_aln_MUSCLE.afa -o out2

C. CDS target prediction (ortho_mode) using varied parameters
Run B1 command before running the following commands.

C1. Set score cutoff to 5

perl ../TarHunter.pl -q mir.txt -s spe1.txt -b cds_db/ -a ortho/ortho_aln_MUSCLE.afa -f 5 -o out3

(note: C1 captures miR172-MYB84 that is not detected in B2, as the default score cutoff is 4)

C2. Use total mispair cutoff

perl ../TarHunter.pl -q mir.txt -s spe1.txt -b cds_db/ -a ortho/ortho_aln_MUSCLE.afa -M 4 -o out4

C3. Use RNAhybrid

perl ../TarHunter.pl -q mir.txt -s spe1.txt -b cds_db/ -a ortho/ortho_aln_MUSCLE.afa -p 2 -o out5

C4. Without conservation filter

perl ../TarHunter.pl -q mir.txt -s spe1.txt -b cds_db/ -N -o out6

C5. Without orthologous miRNA search

perl ../TarHunter.pl -q mir.txt -s spe1.txt -b cds_db/ -a ortho/ortho_aln_MUSCLE.afa -R -o out7

D. CDS target prediction (homo_mode) using varied parameters
D1. Search conserved miRNA targets

perl ../TarHunter.pl -q mir.txt  -s spe1.txt  -b cds_db/ -G -o out8

D2. Set identity cutoff to 50%

perl ../TarHunter.pl -q mir.txt  -s spe1.txt  -b cds_db/ -G -c 0.5 -o out9

(note: a loose identity cutoff (-c 0.5) groups AP2,TOE,SMZ together)

E. Target mimics prediction

perl ../TarHunter.pl -q mir.txt -s spe2.txt -b nc_db/ -I -o out10

F. noncoding target prediction (homo_mode)

 perl ../TarHunter.pl -q mir.txt  -s spe3.txt  -b nc_db/ -G -f 5 -c 0.5 -o out11

(note: F use moderately relaxed settings (-f 5 -c 0.5) to capture miR390-TAS3)

G. prediction of the targets of non-miRBase miRNAs

 perl ../TarHunter.pl -q mir.txt -n ../miRs/sit_mir.fa -s spe4.txt -b cds_db/ -o out12

(note: G predicts the targets of S. italica miRNAs (Chavez et al 2014 Nat Commun) that are not collected in miRBase)

Notes

1.    The names of the sequence files should be species name abbreviation followed
      by dot fa, e.g., ath.fa.
2.    miRNA list can either contain miRNA IDs from multiple species, or contain miRNA
      IDs from a single species (TarHunter can automatically search for orthologous
      miRNAs in other species.
3.    Species name abbreviations and miRNA IDs should follow miRBase nomenclature.
4.    Make sure the TarHunter script and miRs folder are under the same directory.
5.    Coding sequences should be in +1 frame, otherwise will be neglected.
6.    It is recommended to use FASTA (default search engine) in target searching,
          as FASTA is much faster than RNAhybrid.
7.    The script has been tested on a Fedora (release 16) server with Perl v5.14.3,
          and on a UBUNTU (release 14.04) PC with Perl v5.18.2.

TarHunterL is a light weight TarHunter that does not have conservation filter. It only requires FASTA excutable in PATH.

Examples

A. Target prediction with score cutoff of 4

 perl ../TarHunterL.pl -q mir.fa -b targ.fa -o out1 -f 4

B. eTM prediction

 perl ../TarHunterL.pl -q mir.fa -b targ.fa -o out2 -I

Contact

If you have questions or report bugs, please send email to Xuan Ma (skyxma AT tjnu.edu.cn)