Design of Protein Segments and Peptides for Binding to Protein Targets

Table 6

An overview of some of the methods developed to predict CPP function. Note that the reported accuracies are not on similar datasets and thus cannot be compared.


Method	Training Set	Features	Result	Accuracy

CellPPD [148]: SVM, MEME/MSAT	708 experimentally validated, unique CPPs from the CPPsite database	Amino acid composition, Dipeptide Composition, Physicochemical Properties of amino acids	Detailed 120 CPP motifs identified from their dataset	97.40%
SkipCPP-Pred [150]: K-skip-n-gram model, RF classifier	1855 experimentally validated CPPs from CPPsite 2.0	Input sequences	High quality dataset CPP924 with reduced redundancy	90.6%
CPPred-FL [152]: RF classifier	Benchmark dataset CPP924 containing 924 peptide sequences	6 features based on class information and 19 features based on probabilistic information	New feature representation algorithm that includes all sequence based feature descriptors	91.2%
BChemRF-CPPred [154]: MLP architecture, SVM, Gaussian process classifier	CPPsite 2.0, C2Pred server	Structure based descriptors including molecular weight, number of rotatable bonds, topological polar surface area, number of hydrogen bond donors and acceptors, etc. Amino acid composition-based descriptors including dipeptide composition. Pseudo-amino acid composition related to hydrophobicity, hydrophilicity, side chain mass, etc	Evaluated the influence of structural and physicochemical properties in the permeability of peptides	89.62%