This function firstly normalizes PSSM Matrix by formula:
\(P-min(P)/max(P)-min(P)\) then for any standard amino acid specifies its
position in protein sequence whereby a sub-matrix from PSSM corresponding
to these positions will be extracted, then for this sub-matrix computes
colSums
of its columns to create a vector of length 20, eventually a feature vector of
length 400 will be obtained.