Integration of residue attributes for sequence diversity characterization of terpenoid enzymes

Progress in the "omics" fields such as genomics, transcriptomics, proteomics, and metabolomics has engendered a need for innovative analytical techniques to derive meaningful information from the ever increasing molecular data. KNApSAcK motorcycle DB is a popular database for enzymes related to secondary metabolic pathways in plants. One of the challenges in analyses of protein sequence data in such repositories is the standard notation of sequences as strings of alphabetical characters. This has created lack of a natural underlying metric that eases amenability to computation. In view of this requirement, we applied novel integration of selected biochemical and physical attributes of amino acids derived from the amino acid index and quantified in numerical scale, to examine diversity of peptide sequences of terpenoid synthases accumulated in KNApSAcK motorcycle DB. We initially generated a reduced amino acid index table. This is a set of biochemical and physical properties obtained by random forest feature selection of important indices from the amino acid index. Principal component analysis was then applied for characterization of enzymes involved in synthesis of terpenoids. The variance explained was increased by incorporation of residue attributes for analyses.