Genome Sequencing Error Correction Using Deep Convolutional Neural Network

Abstract

Advancement in technology has led to sequencing a vast number of genomic data. The genome sequencing process has gone through three main generations, and each generation improves tremendously on its predecessor. Quite recently, the human genome has also been successfully sequenced. Varied sequencing platforms have emanated through this technological advancement. Each platform uses a different sequencing procedure. This has led to the introduction of errors into sequenced data. These errors compromise the quality of the sequenced data and any inferences made about them. Many efforts and techniques have been used to try and correct sequencing errors. The issue remains prevalent primarily because the processes or methods used to correct genomic sequencing errors are overly intricate to implement, necessitating a significant amount of computational resources. The error correction process is identified as a pattern recognition problem and proposed a classification (machine learning) approach to solving the sequencing error correction problem. This involves using a deep convolutional neural network that is gentle on computational resources and very fast at correcting insertion and deletion errors. The system is tested using the genome data NA12878, which was obtained from the NBSI, and it achieved 99.5% classification accuracy