Gene Essentiality Mediated Base Substitution in Escherichia coli genome: Machine Learning Analysis

Several selection and mutation factors are known to influence evolution in the nucleotide sequence in the genome of organisms. Using the machine learning approach, we considered gene essentiality as a selection factor, and we investigated base substitution among essential and remaining gene sets in the Escherichia coli genome. RF and SVM classifiers achieved an AUC score of 75% each. The importance of the base substitution-based features differs in predicting essentiality. Features such as G to A substitutions, mutations within the amino group, and transversion mutations were found to be the most important features. In contrast, G to C, G to T, and T to C substitutions were among the least ones. This result indicates that the base substitution patterns in essential genes might differ from that in non-essential genes. The result of this manuscript will be helpful for understanding evolution at the molecular level.