All the methods which can handle multiple sequences without any structure input can only capture motifs whose occurrences are necessarily contiguous subsequences.
However, poor performance has often been reported on capturing motifs in generic promoters [ 27].
Exact or degenerate consensuses are in fact powerful enough to capture significant motifs given the much higher redundancy of motif instances in the sequences.
Since sequence motifs are often restricted to a particular family or sub-family, such strategies are unable to capture structural motifs spanning across several families.
Thus, our approach improves upon previous efforts to capture sequence motifs that predict substrate status.
Remarkably, all these models turn out to have comparable discriminative power, but are not sufficient to capture all motifs.
We found that this limited set of gap insertion effectively helps to capture composite motifs with multiple functional elements.
Moreover, since some amino acids play important roles in structure of solenoids, we propose statistics features to capture repeating motifs of solenoids.
Our model suggests that the periodical pattern of cell cycle genes is largely coded in their promoter regions, which can be captured by motif and transcription factor binding data.
Therefore, we propose using amino acid compositions as additional features to capture repeating motifs of solenoid proteins as follows (6) where X = [A, C, D, E, F,···, V, W, Y] is the one-letter code corresponding to each of the 20 amino acids, and N X is the number of amino acids X in a sequence, and N s is the length of that sequence.
