This feature groups together rows that are related to the same amino acid. This is done using a vector \(SA(k)\), in which \(k\) spans \(1:400\) and \(SA(k) = avg_{i = 1, 2... L}H[i, j] \times \delta(P(i), A(z))\), in which \(H\) is the HMM matrix, \(P\) in the protein sequence, \(A\) is an ordered set of amino acids, the variables \(j, z = 1:20\), the variable \(k = j + 20 \times (z-1)\) when creating the vector, and \(\delta()\) represents Kronecker's delta.
References
Nanni, L., Lumini, A., & Brahnam, S. (2014). An Empirical Study of Different Approaches for Protein Classification. The Scientific World Journal, 2014, 1–17.
Examples
h<- hmm_Single_Average(system.file("extdata", "1DLHA2-7", package="protHMM"))