In
recent years, DNA-based storage systems have emerged as promising
solutions for long-term data archival due to their high data density and
durability. This talk explores the intricacies of DNA storage,
highlighting its advantages and challenges.
DNA-based storage possesses distinct characteristics that distinguish it
from conventional storage mediums. Synthesized DNA strands contain
inherent redundancy due to replication during synthesis and sequencing
processes, yet their unordered arrangement introduces complexity to data
retrieval. Additionally, synthesis and sequencing introduce errors such
as insertions, deletions, and substitutions, posing significant
challenges to data integrity.
This talk will begin with an examination of the background of DNA
storage and the challenges posed by the DNA storage channel. Throughout
the presentation, we will emphasize several open problems related to the
channel, underscoring the need for innovative solutions to advance DNA
storage technology.
To bridge theory with application, we will discuss two recent research
projects. The first focuses on the DNA coverage depth problem, which
aims to reduce both, cost and latency in DNA storage. We will
investigate how error-correcting codes and retrieval algorithms impact
the sequencing coverage depth and explore how this coverage can be
optimized. Additionally, we will explore a scalable end-to-end DNA
storage pipeline, integrating deep neural networks and tensor-product
based error-correcting codes, significantly improving speed, accuracy,
and code rate.
These insights contribute to advancing the efficiency and feasibility of
DNA storage solutions, setting the stage for future advancements.