Biochemical constraint compatible address design for fuzzy retrieval of images in DNA Storage

Abstract

With the exponential growth of digital data, DNA molecules have shown its attractive application potential as information storage medium. As the scale of data stored by DNA increases, achieving random data access has become crucial. Traditional methods such as key-based data encoding require accurate index sequence primer to access the target data, and the efficiency of data retrieval is limited. To achieve fuzzy retrieval in DNA storage, primer sequences need to not only have important features that satisfies retrieval efficiency, but also meet the requirements of biochemical compatibility. Here, we introduced generative adversarial network (GAN) into the primer design step, and designed feasible primers which both meet the fuzzy retrieval requirements and biochemical constraints during DNA synthesis and sequencing process. To provide easy-to-use software package, we developed DNAfr, which covers essential in silico modules of fuzzy retrieval in DNA storage.