Portable Molecular Computers Capable of Executing Functions of a Multilayer Perceptron, used for Gene-Expression Profile Analysis and Point-of-Care Virus or Disease Diagnosis | Undergraduate Engineering Research Day
1 minute read
Gene-expression analyses can diagnose many diseases, but its implementation currently involves expensive instrumentation. In molecular computing, wires and transistors of a silicon-based computer are replaced with engineered bio-molecules, which relay and integrate signals via programmed self-assembly. This project aims to build a low-cost and accessible molecular computer capable of executing the functions of a deep-learning multilayer perceptron for point-of-care viral classification.
Current molecular computers composed of DNA strands are limited to solving simple problems. They are incapable of executing more computationally intensive deep-learning algorithms, such as a multilayer perceptron (MLP). As such, we are ao,omg tp design molecular computers consisting of a set of DNA strands that can process molecular signals autonomously in solution via sequence-specific binding, toehold mediated strand displacement, and controlled RNA transcription. The molecular computer will be designed to accept specific DNA sequences as inputs, compute their relative concentrations, and produce functional RNA molecules as outputs.
An exciting application of this technology is in point-of-care COVID-19 diagnostics, where levels of gene expression in patient samples can be directly analyzed using this molecular computer, which then generates different colored RNA molecules (e.g. fluorescent aptamers) that report on the presence of the virus.
Please find the poster here.