Interpretation of compared output

[soeding]

I am trying to use the raw prediction matrix output from CCMpred (-r option), but could not find any documentation on how these files are structured (sorry if I missed an obvious documentation).

I figure the first L lines are the h_i parameters for the each position in the protein and the 21 lines x 21 entries blocks afterwards are the J_ij coupling coefficients for all position pairs (including a gap state).

My questions are:

• What is the ordering of the amino acids / gap state?
• How are the position pairs enumerated and which position corresponds to rows of the 21x21 matrices and which position corresponds to the columns?

Thanks a lot for your help

Joern

[soeding]

Hello Joern,

thanks for your interest in CCMpred.
The ordering of the 21 amino acid states is "ARNDCQEGHILKMFPSTWYV-"
Instead of answering the second question, I would like to point you to the python-based parsing code for both the "oldraw" and "MessagePack" raw coupling format available as part of the CCMgen sourcecode. It will parse the raw coupling files into an object with x_single and x_pair matrices available as NumPy arrays. For x_pair, the array indexing is [i, j, a, b] with a corresponding to the amino acid at position i and b corresponding to the amino acid at position b. I hope that this is a lot easier and less error-prone to use than coming up with your own parser.
Best wishes,

Stefan (Seemayer)