CHEMGENOME 2.0

ChemGenome is an ab initio model for gene evaluation and gene prediction in prokaryotic genomes based on physico-chemical characteristics of codons calculated from molecular dynamics (MD) simulations. It uses physico-chemical properties to construct a 3D vector to answer the fundamental question, 'What is a gene?'. The protocol for gene prediction has been divided into three modules that form an automated pipeline. The methodology combines filters based on stereochemical properties of protein sequences and frequencies of occurrences of codons via their corresponding amino acids from 175000 Swissprot proteins to reduce over prediction. The methodology has been validated on 372 prokaryotic genomes with sensitivity, specificity and correlation coefficients averaged over 356208 genes and an equal number of frame-shifted genes (non-genes) as 97.5%, 97.20% & 94.25% respectively.