Any tool to generate protein similarity matrix

Question

[Zealseeker]

Hi,

I want to calculate the similarities between each protein pairwise and then implement random walk to predict interactions. A paper mentions "J. Mol. Biol. (1981) 147, 195-197" to compare two proteins, but I am afraid that may be obsolete.

I'd like to use some popular tools like BLAST/FASTA..

Is there any good toolkit that can generate such matrix? Just similarity is enough.

Example    Protein1  Protein2  Protein3
Protein1   1         0.9       0.4
Protein2   0.9       1         0.7
Protein3   0.4       0.7       1

As far as I know, BLAST can only tell the identy between two proteins, and FASTA can also tell the similairy (though I don't know why). And FASTA is more quicker, I think.

Then another problem emerges. How to use FASTA by python? the fasta35.exe (windows OS) is interactive like following.

>>fasta35 q.fasta lib.fasta
>>Enter filename for results []
>>output.txt
>>How many scores would you like to see? [20]

I'd prefer that it could output the whole consequent without any stdin. And then I can mine the results via python.

Answer 1

[Joseph Hughes]

I would use clustalo with the following command:

clustalo -i my-in-seqs.fa -o my-out-seqs.fa -v --distmat-out=output_mat.txt

The --distmat-out will give you a pairwise distance matrix for each protein sequence that you have in your input fasta file.

Hope this helps,

Joseph

[Zealseeker]

It's a good toolkit and now I'm learning to use it.

The distances ranging from 0,meaning the same, to 1, meaning totally different. Most distances are larger than 0.8...