0
$\begingroup$

I have a set of nucleotide sequences for which I have aligned using Clustal Omega. In particular, I performed a full alignment, and obtained a full distance matrix.

The distance matrix scores range between 0 and 1. I am looking to use this score to back-compute the number of different positions present in the alignment. Is this possible? If possible, I'm looking to avoid using code (my own or otherwise) to re-compute the number of positions differing between each pair of segments, and instead compute it directly from the distance score.

Here is a toy example of what I am receiving from ClustalOmega:

Sequence    1    2    3    4
1           0    0.1  0.06 0.1
2                0    0.4  0.23
3                     0    0.05
4                          0

The numbers are the "distances" as calculated by ClustalOmega. According to the README file, they are computed by the k-tuple measure. I tried parsing the original paper (published in 1983 in PNAS), but I could not figure out how k-tuple distances are computed, and I could not figure out how the distance metric (as reported like above) is computed from k-tuple distances.

I would like to convert those numbers into the number of positions that differ between each pair of sequences when the two are aligned. This includes substitutions, insertions, deletions. I am currently doing this for 520 sets of virus sequences. Is this possible?

$\endgroup$
6
  • $\begingroup$ What exactly do you mean by "positions" in your alignment? How many sequences are we talking about? Can you show an example of the matrix and your desired result? $\endgroup$
    – terdon
    Commented Mar 24, 2014 at 19:56
  • $\begingroup$ Sure, let me update the original question. $\endgroup$
    – ericmjl
    Commented Mar 24, 2014 at 22:36
  • 3
    $\begingroup$ It uses a gonet matrix to compare each of these two sequences. Since you you could have insertions and extensions as well as substitutions, it becomes a 3 parameter problem. 1*open + 6*penalty + substituion_penalty = X. X can be solved by a linear combination of subsituion_penalties, extensions, and insertions. So I think this will be really really hard. $\endgroup$ Commented Mar 25, 2014 at 8:21
  • $\begingroup$ with that being said. Clustal spits out the multiple sequence alightment. Why don't you just look at sequence 1 and sequence 2 and see what the insertions and subtitutions are! $\endgroup$ Commented Mar 25, 2014 at 8:22
  • $\begingroup$ Haha, it looks like I will have to do that in the end. I was trying to see if I could take the lazy route and not write a few lines of code. Thank you for all of your input nonetheless! $\endgroup$
    – ericmjl
    Commented Mar 25, 2014 at 13:00

1 Answer 1

3
$\begingroup$

It uses a gonet matrix to compare each of these two sequences. Since you you could have insertions and extensions as well as substitutions, it becomes a 3 parameter problem. 1*open + 6*penalty + substitution_penalty = X. X can be solved by a linear combination of substitution_penalties, extensions, and insertions. So I think this will be really really hard

with that being said. Clustal outputs the multiple sequence alignment. Why don't you just look at sequence 1 and sequence 2 and see what the insertions and substitutions are!

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .