[EMBOSS] A bug in einverted?

[Francis Tang]

Dear EMBOSS users,

I believe I have found a bug in einverted - at least in the tfm
documentation.

In "tfm einverted", under section "Notes", there is the statement

     "(Anything can be considered to be a repeat if you set the score
     threshold low enough!)"

I don't think that this is true.  For example:

Here's a sample sequence, stored as pkb00003.fasta :

     >pkb00003 Pseudobase #3
     AAAAAACGGGAAGCAAGGGGCUCAAGGGAGGCCCCAGAAACAAACUUUCCCGAU

Here's a transcript of a session where I choose a gap penalty of 3:

     $ einverted pkb00003.fasta -stdout -auto -gap 3 -mat 3 -mis -5 
-thresh 10

     Score 18: 8/8 (100%) matches, 2 gaps
            6 a-cggg-aag 13
              | |||| |||
           54 tagccctttc 45

However, when I set the gap penalty to 2, I seem to lose the previous
palindrome:

     $ einverted pkb00003.fasta -stdout -auto -gap 2 -mat 3 -mis -5 
-thresh 10

     Score 19: 7/7 (100%) matches, 1 gaps
           17 gggg-ctc 23
              |||| |||
           35 ccccggag 28

     Score 14: 8/8 (100%) matches, 5 gaps
           14 -caa-gggg-ctc 23
               |   |||| |||
           38 ag--accccggag 28

The other palindrome should have been found because by reducing the
gap penalty, it would only increase the scores of existing
palindromes.

Has anyone else noticed this before?  Should this be reported as a bug?

BTW, I am using EMBOSS 2.8.0 distribution compiled from source on
Fedora Core 2.

Thanks.

Francis.

-- 
Francis TANG, Post-Doctoral Research Fellow
Bioinformatics Institute, BMSI, A-STAR, Singapore.
Tel: +65 64788282  Fax: +65 64789048  Email: francis at bii.a-star.edu.sg
Add: Matrix L7, Biopolis   WWW: http://www.bii.a-star.edu.sg/~francis/