Prediction of Origin of Replication in Genome Using DNA Sequence Based Features

Origin of replication (OriC) within prokaryotic and eukaryotic genomes are single or several sequences in the genome where DNA double helix strands start unwinding and the replication starts by providing access to the DNA polymerase. Knowledge about OriC is of great importance due to its effect in DNA replication, which is the most important task during cell division. In this paper, we propose a novel method for prediction of OriC in genome sequences. Our proposed supervised learning based method uses simple k-mer features derived from DNA sequences and Support Vector Machine as classification algorithm. Our method is trained using a standard benchmark dataset and on cross fold validation it achieves 84.36% accuracy which is significantly improved than the state-of-the-art methods. It also performs well on other metrics such as Mathew's Correlation Coefficient (MCC), Sensitivity and area under receiver operating characteristic curve (auROC) whose values respectively are 0.69, 0.8621 and 0.8907 and thus is superior in performance compared to the previous state-of-the-art methods.