Special Protein Molecules Computational Identification

• Main Author: Quan Zou (Ed.) (auth)
• Format: Electronic Book Chapter
• Language: English
• Published: MDPI - Multidisciplinary Digital Publishing Institute 2018
• Subjects: MHC binding peptide; type III secreted proteins; machine learning; oncogene; anticancer peptides; bioinformatics; Proteomics; DNA/RNA binding proteins; prediction; PseAAC features; Cell-Penetrating Peptides; protein classification; feature selection
• Online Access: DOAB: download the publication; DOAB: description of the publication

It is time consuming and costly to detect new molecules of some special proteins. These special proteins include cytokines, enzymes, cell-penetrating peptides, anticancer peptides, cancer lectins, G-protein-coupled receptors, etc. Researchers often employ computer programs to list some candidates, and to validate the candidates with molecular experiments. These computer programs are key to possible savings on wet experiment costs. Software results with high false positive will lead to high costs in the validation process. In this Special Issue, we focus on these computer program approaches and algorithms. Some "golden features" from protein primary sequences have been proposed for these problems, such as Chou's PseAAC (pseudo amino acid composition). PseAAC has been tried on nearly all kinds of protein identification, together with SVM (support vector machines, a type of classifier). However, I prefer special features, and classification methods should be proposed for special protein molecules. "Golden features" cannot work well on all kinds of proteins. I hope that submissions will focus on a type of special protein molecule, collect related data sets, obtain better prediction performance (especially low false positives), and develop user-friendly software tools or web servers.