Improving accuracy of protein-protein interaction prediction by considering the converse problem for sequence representation

Table 1 Features of various representation schemes of protein sequence according to our evaluation model

	K	Q	P	QP	KQP
f	Counts	Positions	Counts	Counts and positions	Counts and positions
g	Eulerian trails, reversible	Simple filling, reversible	Simple filling, reversible	Simple filling, reversible	Eulerian trails and filling, reversible
dim(V)	20^k	20*q	20*p	20(1+q)p	20^k(1+q)p

K: k-mer based methods; P: segmentation based methods; Q: our quantile-position based methods; QP: combination of Q and P; KQP: combination of K, Q and P. f: mapping from sequences to vectors; g: mapping from vectors to sequences; dim(V): dimension of vector V.

ISSN: 1471-2105