Re: [Biopython] Markov modelling

Thank you very much Chris--I will read the numerical recipes.

I should have explained more about the data.

I extracted unique 5 mer patterns of amino acids from a database and look
at their secondary structures.

I want to know is there a relationship between the amino acid sequence and
the secondary structure.

This is the reason I thought Markov modelling might give me some ideas.

On Sat, Aug 11, 2012 at 6:32 PM, Chris Mitchell <chris.mit7 <at> gmail.com>wrote:

> Hi George,
>
> A few things I would do:
> 1) Make a spare dataset by randomly selecting a subset of your data to
> visualize.  Just blindly fitting the data points to any distribution
> usually doesn't end well.
> 2) Find a copy of Numerical Recipes and read up on fitting data/modeling.
>  There is a good section on Markov Models and other fitting routines.
> 3) Think about what you are trying to fit.  Is the process generating the
> data a Poisson Process?  What sort of equation should be used to model it,
> and what do the extracted parameters mean in relation to your data?  Start
> with the simplest model you can explain (linear) and add/change parameters
> only if you can justify it.
>
> Chris
>