Markov Chain For Generating Random Protein Sequences

Question

[Perry]

I have 1000+ protein sequences. I want to generate random sequences using a Markov model based on residue transitions found my sequences. I'm told Matlab will make a Markov chain based on multiple sequences, but I would like to use a free alternative to Matlab (python, ruby, R, etc). Can anyone provide me with a library or module?

Answer 1

[Istvan Albert]

It seems Python programmers like writing Markov generators. I often see this topic pop up on Python blogs in the context of generating pseudo random text. A quick search shows a few hits:

• Markov chains
• Pseudo random text generator
• Markov chains in python
• Markov chain algorithm

I guess you would only need to change to tokenizer to split on letters rather than words.

[Perry]

Thanks. The first link helped.

[Jarretinha]

For 1st and 2nd order chains you can use Sean Eddy's Squid lib. It's written in C. But, you could add some randomizations as a control. In this case you could use uShuffle to preserve high-level orderings. Then you'll see why biology is very much local . . .

[Eli Korvigo]

All links, but the last one, are dead.

[Istvan Albert]

pretty sad state of the internet - you can't trust anything to be around