A Content-Addressable DNA Database with Learned Sequence Encodings

Engineering gene sequences with defined functional properties is a major goal of synthetic biology. Deep neural network models, together with gradient ascent-style optimization, show promise for sequence generation. The generated sequences can however get stuck in local minima, have low diversity and their fitness depends heavily on initialization. Here, we develop deep exploration networks (DENs), a type of generative model tailor-made for searching a sequence space to minimize the cost of a neural network fitness predictor. By making the network compete with itself to control sequence diversity during training, we obtain generators capable of sampling hundreds of thousands of high-fitness sequences. We demonstrate the power of DENs in the context of engineering RNA isoforms, including polyadenylation and cell type-specific differential splicing. Using DENs, we engineered polyadenylation signals with more than 10-fold higher selection odds than the best gradient ascent-generated patterns and identified splice regulatory elements predicted to result in highly differential splicing between cell lines.

show abstract

“…The first method has previously been demonstrated in the context of genomics (Killoran et. al., 2017;Stewart et. al.…”

Section: Pattern Representation For Genomicsmentioning

confidence: 99%

Deep exploration networks for rapid engineering of functional DNA sequences

Linder

Bogard

Rosenberg

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…In fact, our simulations find that this limitation restricts theoretical densities and capacities by over 30 fold when dense encodings are used ( Figure 2D), with some densities not even achievable without encountering primer conflicts within the data payload region; 3) Finally, it is unclear how PCR, a method that actively creates new DNA strands, can serve as the foundation for in-storage file manipulations without significant challenges associated with altering the composition and strand balance of a database. To address this set of challenges, we propose a simple but fundamental shift in DNA storage system architectures, inspired by single stranded DNA (ssDNA) 'toeholds' used in synthetic biology and DNA computing [17][18][19][20] , and by the way organisms naturally access information from their genome through transcription. As described below and in Figure 1C, we engineered a reusable DNA-based information storage system that can be created at scale, reduces off-target file access, and supports multiple in-storage file operations.…”

Section: Strategy: Molecular Technologies To Unlock Dynamic Features mentioning

confidence: 99%

Dynamic DNA-based information storage

Lin

Tuck

Keung

2019

Preprint

View full text Add to dashboard Cite

Technological leaps are often driven by key innovations that transform the underlying architectures of systems. Current DNA storage systems largely rely on polymerase chain reaction, which broadly informs how information is encoded, databases are organized, and files are accessed. Here we show that a hybrid 'toehold' DNA structure can unlock a fundamentally different, dynamic DNA-based information storage system architecture with broad advantages.This innovation increases theoretical storage densities and capacities by eliminating non-specific DNA-DNA interactions common in PCR and increasing the encodable sequence space. It also provides a physical handle with which to implement a range of in-storage file operations. Finally, it reads files non-destructively by harnessing the natural role of transcription in accessing information from DNA. This simple but powerful toehold structure lays the foundation for an information storage architecture with versatile capabilities.

show abstract

“…Microfluidic technologies come in many shapes and sizes, each offering different advantages. massive storage density [12,17,38] or parallel computation [45,52,63].…”

Section: Automated Experimentationmentioning

confidence: 99%

“…Instead of executing a pre-determined list of operations and reporting the output, the heterogeneous system can use information from sensors to dynamically make decisions. This combination of fluidic manipulation and computation is critical for emerging applications which depend on biology "in the loop", e.g., molecular data storage and computation [10,17,38,45,52] or automated experimentation [33-35, 48, 51].…”

Section: Introductionmentioning

confidence: 99%

Puddle

Willsey

Stephenson

Takahashi

et al. 2019

Proceedings of the Twenty-Fourth International Conference on Architectural Support for Programming Languages and Operating Syst

Self Cite

View full text Add to dashboard Cite

Microfluidic devices promise to automate wetlab procedures by manipulating small chemical or biological samples. This technology comes in many varieties, all of which aim to save time, labor, and supplies by performing lab protocol steps typically done by a technician. However, existing microfluidic platforms remain some combination of inflexible, error-prone, prohibitively expensive, and difficult to program. We address these concerns with a full-stack digital microfluidic automation platform. Our main contribution is a runtime system that provides a high-level API for microfluidic manipulations. It manages fluidic resources dynamically, allowing programmers to freely mix regular computation with microfluidics, which results in more expressive programs than previous work. It also provides real-time error correction through a computer vision system, allowing robust execution on cheaper microfluidic hardware. We implement our stack on top of a low-cost droplet microfluidic device that we have developed. We evaluate our system with the fully-automated execution of polymerase chain reaction (PCR) and a DNA sequencing preparation protocol. These protocols demonstrate high-level programs that combine computational and fluidic operations such as input/output of reagents, heating of samples, and data analysis. We also evaluate the impact of automatic error correction on our system's reliability.

show abstract

A Content-Addressable DNA Database with Learned Sequence Encodings

Cited by 23 publications

References 26 publications

Deep exploration networks for rapid engineering of functional DNA sequences

Deep exploration networks for rapid engineering of functional DNA sequences

Dynamic DNA-based information storage

Puddle

Contact Info

Product

Resources

About