Enzymatic Weight Update Algorithm for DNA-Based Molecular Learning

Abstract

Recent research in DNA nanotechnology has demonstrated that biological substrates can be used for computing at a molecular level. However, in vitro demonstrations of DNA computations use preprogrammed, rule-based methods which lack the adaptability that may be essential in developing molecular systems that function in dynamic environments. Here, we introduce an in vitro molecular algorithm that ‘learns’ molecular models from training data, opening the possibility of ‘machine learning’ in wet molecular systems. Our algorithm enables enzymatic weight update by targeting internal loop structures in DNA and ensemble learning, based on the hypernetwork model. This novel approach allows massively parallel processing of DNA with enzymes for specific structural selection for learning in an iterative manner. We also introduce an intuitive method of DNA data construction to dramatically reduce the number of unique DNA sequences needed to cover the large search space of feature sets. By combining molecular computing and machine learning the proposed algorithm makes a step closer to developing molecular computing technologies for future access to more intelligent molecular systems.

DNA Input Classification by a Riboregulator-Based Cell-Free Perceptron

“…Alternatively, to achieve a fully biological implementation of the perceptron, an enzymatic approach can be applied, where specific RNA sequences are amplified or cleaved based on the current perceptron output. 78 By engineering such sequences to competitively bind to the transducer strands, and in doing so, blocking the toehold site, the effective weight pattern can be adapted without altering the concentration of weights added to the system. The current implementation of the perceptron presented here lacks this learning functionality, instead relying on the manual tuning of the classification threshold.…”

“…In a similar manner, continuous flow reactions could be implemented to update the weight pattern at set intervals following the real-time monitoring of the output concentration, providing semi-autonomous perceptron behavior. Alternatively, to achieve a fully biological implementation of the perceptron, an enzymatic approach can be applied, where specific RNA sequences are amplified or cleaved based on the current perceptron output . By engineering such sequences to competitively bind to the transducer strands, and in doing so, blocking the toehold site, the effective weight pattern can be adapted without altering the concentration of weights added to the system.…”

Learning without neurons in physical systems

“…At a more practical level, the learning process is bound to be dissipative in nature. For example, the molecular versions might involve polymerization, ligation or strand displacement (65,51,54) while mechanical systems involve dissipative re-arrangements in foams and other complex materials (5). The theoretical and practical constraints on learning by dissipation deserve further study.…”

Molecular Computing and Bioinformatics

“…Their simulation results proved the feasibility and accuracy of the adder/subtractor logic calculation model based on the domain label, which could extend its application for molecular logic circuits. Beak et al [16] developed an enzyme weight-updating algorithm on the basics of DNA molecular learning for future smart molecular computing systems. The new algorithm used a hypernetwork model, which integrated the internal circulation structure of DNA and ensemble learning to update the enzyme weight.…”