Generative and predictive neural networks for the design of functional RNA molecules.

RNA is a remarkably versatile molecule that has been engineered for applications in therapeutics, diagnostics, and in vivo information-processing systems. However, the complex relationship between the sequence and structural properties of an RNA molecule and its ability to perform specific functions often necessitates extensive experimental screening of candidate sequences. Here we present a generalized neural network architecture that utilizes the sequence and structure of RNA molecules (SANDSTORM) to inform functional predictions. We demonstrate that this approach achieves state-of-the-art performance across several distinct RNA prediction tasks, while learning interpretable abstractions of RNA secondary structure. We paired these predictive models with generative adversarial RNA design networks (GARDN), allowing the generative modelling of novel mRNA 5' untranslated regions and toehold switch riboregulators exhibiting a predetermined fitness. This approach enabled the design of novel toehold switches with a 43-fold increase in experimentally characterized dynamic range compared to those designed using classic thermodynamic algorithms. SANDSTORM and GARDN thus represent powerful new predictive and generative tools for the development of diagnostic and therapeutic RNA molecules with improved function.

Electrochemical Strategy for Low-Cost Viral Detection.

Sexually transmitted infections, including the human immunodeficiency virus (HIV) and the human papillomavirus (HPV), disproportionally impact those in low-resource settings. Early diagnosis is essential for managing HIV. Similarly, HPV causes nearly all cases of cervical cancer, the majority (90%) of which occur in low-resource settings. Importantly, infection with HPV is six times more likely to progress to cervical cancer in women who are HIV-positive. An inexpensive, adaptable point-of-care test for viral infections would make screening for these viruses more accessible to a broader set of the population. Here, we report a novel, cost-effective electrochemical platform using gold leaf electrodes to detect clinically relevant viral loads. We have combined this platform with loop-mediated isothermal amplification and a CRISPR-based recognition assay to detect HPV. Lower limits of detection were demonstrated down to 104 total copies of input nucleic acids, which is a clinically relevant viral load for HPV DNA. Further, proof-of-concept experiments with cervical swab samples, extracted using standard extraction protocols, demonstrated that the strategy is extendable to complex human samples. This adaptable technology could be applied to detect any viral infection rapidly and cost-effectively.

Rapid electrostatic DNA enrichment for sensitive detection of Trichomonas vaginalis in clinical urinary samples.

Estimated to be the most common non-viral sexually transmitted infection globally, Trichomonas vaginalis (TV) can lead to pelvic inflammatory disease, pregnancy complications, and increased risk of acquiring and transmitting HIV. Once diagnosed, TV infection can be treated with oral antibiotics; however, infected individuals are often asymptomatic and do not seek treatment. The WHO and others have identified a need for point-of-care tests to expand access to TV testing and screening; ideal test characteristics include high sensitivity and specificity and the ability to use urine as a sample type, rather than invasively collected swab samples. Here, we report on a proof-of-concept prototype for rapid, electrostatic enrichment of DNA from urine samples and demonstrate the use of large volumes of urine to increase sensitivity of downstream nucleic acid amplification testing. We developed an internally controlled thermophilic helicase-dependent amplification (tHDA) assay with lateral flow immunoassay readout and demonstrate that this tHDA assay can be performed directly on our DNA capture filter. We validated our method using clinical urine samples with qPCR-quantified TV loads. Using 62 clinical urine samples and a simple sample processing device, our tHDA assay displayed 96.6% sensitivity and 100% specificity. Our analytical limit of detection was found to be approximately 7 genomic equivalents of TV DNA per mL of sample when 1 mL of sample was tested, comparable to existing isothermal tests for TV. Using large-volume simulated samples (40 mL of buffered urine with spiked-in TV DNA), we also demonstrated that sensitivity could be improved 28-fold to 0.25 genomic equivalents of TV DNA per mL, with a sample processing time of only 2 minutes.