DNA for cold data archiving: using machine learning for robust decoding

Abstract

This work deals with the storage of digital data into synthetic DNA. DNA data storage consists an innovative step for the digital data handling as it allows efficient long-term storage of cold data (data infrequently accessed). Unfortunately, despite the enormous advantages of DNA data storage, the main drawback is the high cost of the biological procedures of DNA synthesis (writing) and sequencing (reading). This work focuses on the reduction of the cost of DNA sequencing by studying a sequencing machine called Nanopore. This device is faster, cheaper and more user-friendly in comparison to other sequencers widely used in the biological field. However, the main disadvantage is the high error rate of the sequencing procedure. Consequently, this internship subject deals with the study of the sequencing errors of the Nanopore and the implementation of error correction algorithms to improve the quality of the reconstructed data.