CRISPR repeat identification tool comparison

Question

[kspata]

Hi I am using two bioinformatics tools for identifying CRISPR repeats. These tools are pilercr and CRT. How to compare the output of these two tools especially how to know which one is better? My thoughts on this is to run my sample reads and reference genomes on both tools with and then find the matches between ref and sample reads or 2nd run blast using sample and ref and find out exact matches between the blast output and the tools output.. which approach is correct? or is there any other way I can decide which tool is better compared to other?

[WouterDeCoster]

It's unclear what the output of those tools is, and what the desired comparison would be. Probably a trade-off between specificity and sensitivity, but you will need to add more information.

In addition, your thread would benefit from a more specific title.

[kspata]

the output of both tools is number of putative crispr arrays present in an organism.

Pilercr output
Array 1
>NODE_2_length_616862_cov_34.5047

       Pos  Repeat     %id  Spacer  Left flank    Repeat                           Spacer
==========  ======  ======  ======  ==========    =============================    ======
    164881      29   100.0      32  GAACGCCCCT    .............................    GCGAAATAGTGGGGAAAAACCCCTGGTTAACC
    164942      29    96.6      32  CTGGTTAACC    .......................A.....    TAGGCCTTGATACCATCGCTCGCACCTCGTCA
    165003      29   100.0      32  CACCTCGTCA    .............................    GTTTATTACTGCTTAGTTAATTAATGGGTTGC
    165064      29   100.0      32  AATGGGTTGC    .............................    TCTTTTGATTTTGCTGCGATGTTATAACCAGA
    165125      29   100.0          TATAACCAGA    .............................    TATCCACATAT
==========  ======  ======  ======  ==========    =============================
         5      29              32                CGGTTTATCCCCGCTGGCGCGGGGAACAC

[kspata]

I compared both tools on multiple reference samples and compared the output for exact matches with crispr database. I then discarded the tool which gave over-prediction of the putative crispr arrays. I found piler-CR to be better at crispr prediction than crt for my sample.