Characterization of artificial riboswitches for Coxsackievirus B3 detection using machine learning
With the increase of recycled water to offset water demand, the potential possibility to spread contagious RNA viruses, such as Coxsackievirus B3, increases. However, detection of viral particles remains challenging because of low viral concentrations in wastewater and high mutation rates of the RNA virus. Robust monitoring is needed with low cost and low infrastructure technologies to increase…
Artificial riboswitches have been developed to detect Coxsackievirus B3 in wastewater, aiming to monitor the spread of contagious RNA viruses. These riboswitches bind to target viruses and activate reporter genes, amplifying detection signals. Machine learning models can optimize candidate nucleic acid sequences for detection, and a novel ML model has been presented for classifying riboswitch performance.
The model utilizes RNA sequence data along with secondary structural features derived from free energy calculations and single strandedness parameters. It employs a sparsely gated Mixture of Experts (MoE) architecture, which efficiently routes different features to specialized subnetworks, resulting in strong generalization across cross-validation and held-out testing.
When compared to traditional classifiers like Decision Tree, Random Forest, Gaussian Naive Bayes, and Dense Neural Network, the MoE model showcased near-zero classification errors. Additionally, post-hoc feature importance analysis and k-mer analyzes indicated that sequence constructs, along with folding mechanics, were crucial for achieving near-perfect predictive performance in silico.
An ablation study further confirmed that while the Dense Neural Network also performed well, the MoE architecture was preferred due to its specialized subnetworks allowing for computational efficiency.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
