How to compare few DNA multi-alignment algorithms?

Question

[MichaelTrev]

Hello, I have a homework project and I need help from someone who knows that field.

I have to compare the efficiency of multi-alignment algorithms based on BAliBASE. And... what is the best way to do that? I need two value: job time and score. Compared programs are T-coffee and clustalOmega, and it should be done on Linux in text commands. I know these commands, and installation process but I need help with the methodology.

Can someone tell me how can I do that properly step by step for one alignment? Then I will know how to do it on a larger number of alignments. And then I will be able to calculate statistics for individual programs in python.

Thank you in advance for your help!

P.S. Now I pick one file from mdsa_all balibase and just use t-coffee and clustalO. Score in t-coffee what i got is 400, and ClustaO have 200 000. And i don't know what is next step, coz t_coffe give me "consistency score" and Clustal "SP score "

[kloetzl]

Take a look at the “alignathon”.

Answer 1

[Joe]

The 'scores' that each tool give you are not equivalent calculations, so you will need some singular objective scoring function you can apply to the multiple alignment after you create it. This could be something as simple as (total mismatches + total gaps) / number of sequences or something to that effect. Bear in mind that each program also has customisable alignment parameters that can be tweaked, so you need to decide on a 'base case' to start from (probably program defaults would be fine).

Time is easy, you can simply use the time unix tool.

[MichaelTrev]

Thank you very much for valuable advice. I have one more question. I received information that my results are to be: a comparison between a model match and a test match. As I understand it, balibase_mdsa files are a model (model). What should I compare the model's result to be able to achieve such a result? One sentence: what is this test match?

[Joe]

I'm afraid I don't know the answer to that. That terminology doesn't mean much to me either. Not sure if thats something specific to BaliBase or not (which I'm also not massively familiar with).

[MichaelTrev]

OK, thank u soooo much for help! :)