Alternatively, de Lichtenberg et al. found that many cell cycle regulated genes shared common protein features, thus proposed a method to predict novel ones based on protein features [ 7].
(6) where the function E is a neural network with one hidden layer, estimating the log-likelihood of state transition from and based on protein features in a local window (of size 11) centered at the two residues to be aligned.
We evaluated several types of input encoding based on protein features, evolutionary and functional information.
> The consensus phylogeny for Campylobacter genomes was obtained from 16S genes aligned with MUSCLE (Edgar 2004) and using Neighbor-Joining and maximum-likelihood (ML) methods; and from the full proteome, using an alignment-free method based on protein feature frequency profiles (Jun et al. 2010).
Various machine learning methods have been applied and different kinds of features based on protein sequence and/or structural information have been used.
Some methods to extract features are based on protein sequence, such as amino acid composition, n-order couple composition, pseudo-amino acid composition, and residue couple.
These six web servers are all based on protein sequences and they combined several features derived from sequence information, such as amino acid frequency, evolutionary profile, sequence conservation, predicted secondary structure, predicted solvent accessibility, electrostatic potential, hydrophobicity, BLOSUM62 matrix, position-specific scoring matrix (PSSM) etc [ 2, 4, 5, 7].
Until now, a number of feature extraction methods based on protein sequences have been proposed.
These methods can be categorized into three groups: 1) approaches that focus on molecular docking with known protein structures [ 1- 5]; 2) methods that predict putative interacting sites based on protein sequences [ 6- 17]; 3) methods that identify interacting sites based on the hybrid features of protein structure and sequences [ 15].
Therefore, we decided to explore methods based on other protein features for the prediction of cyclin proteins.
Therefore, we set forth to explore machine-learning based methods based on various protein features for the prediction of CDKI proteins.
