Abstract
Download Conference Presentation
Abstract: Most DNA computing systems are single-use. After a computation is completed, the DNA strands in the test tube are discarded, requiring new material for each run. This increases cost, waste, and experimental effort. In molecular computing systems, kinetic proofreading and other error-correcting mechanisms prevent error states, but at an energy cost or complexity scale-up. Unlike these systems, thermodynamically favoured computers reach correct outputs naturally by moving to equilibrium. However, implementing this principle in molecular computing has remained a challenge for decades.
Our thermodynamically favoured Scaffolded DNA Computer (SDC) executes computations using a simple thermal anneal. We successfully demonstrated its <em>programmability by computations like arithmetic and cellular automata in under a minute encoding data on a 12-bit or 24-bit memory, and on 75 bits over a few hours, all without needing explicit error correction. The short runtimes highlight the system's <em>speed</em>, while the increasing memory size shows its ability to <em>scale-up</em>. Moreover, unlike traditional DNA computing approaches that require new material for each run, the SDC is <em>intrinsically renewable</em>. Instead of being discarded, the same mix can be reset and reused multiple times with just two DNA strands, one to reset the program and another to introduce the new input. We validated the renewal process on multiple programs, showing steady performance over repeated cycles.
We contend that these principles also apply to DNA storage platforms. Many existing DNA Storage platforms rely on destructive sequencing [3]. However, by leveraging thermodynamically favoured states, DNA storage can become computable, scalable, rewritable and sustainable. Our approach integrates renewability into molecular computation, reducing material waste, minimizing energy consumption, and paving the way for long-term, low-energy alternatives to traditional computing and storage technologies.