SPLICER: A Highly Efficient Base Editing Toolbox That Enables In Vivo Therapeutic Exon Skipping.

Exon skipping technologies enable exclusion of targeted exons from mature mRNA transcripts, which has broad applications in molecular biology, medicine, and biotechnology. Existing exon skipping techniques include antisense oligonucleotides, targetable nucleases, and base editors, which, while effective for specific applications at some target exons, remain hindered by shortcomings, including transient effects for oligonucleotides, genotoxicity for nucleases and inconsistent exon skipping for base editors. To overcome these limitations, we created SPLICER, a toolbox of next-generation base editors consisting of near-PAMless Cas9 nickase variants fused to adenosine or cytosine deaminases for the simultaneous editing of splice acceptor (SA) and splice donor (SD) sequences. Synchronized SA and SD editing with SPLICER improves exon skipping, reduces aberrant outcomes, including cryptic splicing and intron retention, and enables skipping of exons refractory to single splice-site editing. To demonstrate the therapeutic potential of SPLICER, we targeted APP exon 17, which encodes the amino acid residues that are cleaved to form the Aß plaques in Alzheimer's disease. SPLICER reduced the formation of Aß42 peptides in vitro and enabled efficient exon skipping in a mouse model of Alzheimer's disease. Overall, SPLICER is a widely applicable and efficient toolbox for exon skipping with broad therapeutic applications.

Semi-quantitative group testing for efficient and accurate qPCR screening of pathogens with a wide range of loads.

BACKGROUND: Pathogenic infections pose a significant threat to global health, affecting millions of people every year and presenting substantial challenges to healthcare systems worldwide. Efficient and timely testing plays a critical role in disease control and transmission prevention. Group testing is a well-established method for reducing the number of tests needed to screen large populations when the disease prevalence is low. However, it does not fully utilize the quantitative information provided by qPCR methods, nor is it able to accommodate a wide range of pathogen loads. RESULTS: To address these issues, we introduce a novel adaptive semi-quantitative group testing (SQGT) scheme to efficiently screen populations via two-stage qPCR testing. The SQGT method quantizes cycle threshold (Ct) values into multiple bins, leveraging the information from the first stage of screening to improve the detection sensitivity. Dynamic Ct threshold adjustments mitigate dilution effects and enhance test accuracy. Comparisons with traditional binary outcome GT methods show that SQGT reduces the number of tests by 24% on the only complete real-world qPCR group testing dataset from Israel, while maintaining a negligible false negative rate. CONCLUSION: In conclusion, our adaptive SQGT approach, utilizing qPCR data and dynamic threshold adjustments, offers a promising solution for efficient population screening. With a reduction in the number of tests and minimal false negatives, SQGT holds potential to enhance disease control and testing strategies on a global scale.

DR-BERT: A protein language model to annotate disordered regions.

Despite their lack of a rigid structure, intrinsically disordered regions (IDRs) in proteins play important roles in cellular functions, including mediating protein-protein interactions. Therefore, it is important to computationally annotate IDRs with high accuracy. In this study, we present Disordered Region prediction using Bidirectional Encoder Representations from Transformers (DR-BERT), a compact protein language model. Unlike most popular tools, DR-BERT is pretrained on unannotated proteins and trained to predict IDRs without relying on explicit evolutionary or biophysical data. Despite this, DR-BERT demonstrates significant improvement over existing methods on the Critical Assessment of protein Intrinsic Disorder (CAID) evaluation dataset and outperforms competitors on two out of four test cases in the CAID 2 dataset, while maintaining competitiveness in the others. This performance is due to the information learned during pretraining and DR-BERT's ability to use contextual information.

Emergence of catalytic function in prebiotic information-coding polymers.

Life as we know it relies on the interplay between catalytic activity and information processing carried out by biological polymers. Here we present a plausible pathway by which a pool of prebiotic information-coding oligomers could acquire an early catalytic function, namely sequence-specific cleavage activity. Starting with a system capable of non-enzymatic templated replication, we demonstrate that even non-catalyzed spontaneous cleavage would promote proliferation by generating short fragments that act as primers. Furthermore, we show that catalytic cleavage function can naturally emerge and proliferate in this system. Specifically, a cooperative catalytic network with four subpopulations of oligomers is selected by the evolution in competition with chains lacking catalytic activity. The cooperative system emerges through the functional differentiation of oligomers into catalysts and their substrates. The model is inspired by the structure of the hammerhead RNA enzyme as well as other DNA- and RNA-based enzymes with cleavage activity that readily emerge through natural or artificial selection. We identify the conditions necessary for the emergence of the cooperative catalytic network. In particular, we show that it requires the catalytic rate enhancement over the spontaneous cleavage rate to be at least 102-103, a factor consistent with the existing experiments. The evolutionary pressure leads to a further increase in catalytic efficiency. The presented mechanism provides an escape route from a relatively simple pairwise replication of oligomers toward a more complex behavior involving catalytic function. This provides a bridge between the information-first origin of life scenarios and the paradigm of autocatalytic sets and hypercycles, albeit based on cleavage rather than synthesis of reactants.

A transcriptomic atlas of acute stress response to low pH in multiple Issatchenkia orientalis strains.

IMPORTANCE: Issatchenkia orientalis is a promising industrial chassis to produce biofuels and bioproducts due to its high tolerance to multiple environmental stresses such as low pH, heat, and other chemicals otherwise toxic for the most widely used microbes. Yet, little is known about specific mechanisms of such tolerance in this organism, hindering our ability to engineer this species to produce valuable biochemicals. Here, we report a comprehensive study of the mechanisms of acidic tolerance in this species via transcriptome profiling across variable pH for 12 different strains with different phenotypes. We found multiple regulatory mechanisms involved in tolerance to low pH in different strains of I. orientalis, marking potential targets for future gene editing and perturbation experiments.

FUN-PROSE: A deep learning approach to predict condition-specific gene expression in fungi.

mRNA levels of all genes in a genome is a critical piece of information defining the overall state of the cell in a given environmental condition. Being able to reconstruct such condition-specific expression in fungal genomes is particularly important to metabolically engineer these organisms to produce desired chemicals in industrially scalable conditions. Most previous deep learning approaches focused on predicting the average expression levels of a gene based on its promoter sequence, ignoring its variation across different conditions. Here we present FUN-PROSE-a deep learning model trained to predict differential expression of individual genes across various conditions using their promoter sequences and expression levels of all transcription factors. We train and test our model on three fungal species and get the correlation between predicted and observed condition-specific gene expression as high as 0.85. We then interpret our model to extract promoter sequence motifs responsible for variable expression of individual genes. We also carried out input feature importance analysis to connect individual transcription factors to their gene targets. A sizeable fraction of both sequence motifs and TF-gene interactions learned by our model agree with previously known biological information, while the rest corresponds to either novel biological facts or indirect correlations.

Functional convergence in slow-growing microbial communities arises from thermodynamic constraints.

The dynamics of microbial communities is complex, determined by competition for metabolic substrates and cross-feeding of byproducts. Species in the community grow by harvesting energy from chemical reactions that transform substrates to products. In many anoxic environments, these reactions are close to thermodynamic equilibrium and growth is slow. To understand the community structure in these energy-limited environments, we developed a microbial community consumer-resource model incorporating energetic and thermodynamic constraints on an interconnected metabolic network. The central element of the model is product inhibition, meaning that microbial growth may be limited not only by depletion of metabolic substrates but also by accumulation of products. We demonstrate that these additional constraints on microbial growth cause a convergence in the structure and function of the community metabolic network-independent of species composition and biochemical details-providing a possible explanation for convergence of community function despite taxonomic variation observed in many natural and industrial environments. Furthermore, we discovered that the structure of community metabolic network is governed by the thermodynamic principle of maximum free energy dissipation. Our results predict the decrease of functional convergence in faster growing communities, which we validate by analyzing experimental data from anaerobic digesters. Overall, the work demonstrates how universal thermodynamic principles may constrain community metabolism and explain observed functional convergence in microbial communities.

Predicting metabolic response to dietary intervention using deep learning.

Due to highly personalized biological and lifestyle characteristics, different individuals may have different metabolic responses to specific foods and nutrients. In particular, the gut microbiota, a collection of trillions of microorganisms living in our gastrointestinal tract, is highly personalized and plays a key role in our metabolic responses to foods and nutrients. Accurately predicting metabolic responses to dietary interventions based on individuals' gut microbial compositions holds great promise for precision nutrition. Existing prediction methods are typically limited to traditional machine learning models. Deep learning methods dedicated to such tasks are still lacking. Here we develop a new method McMLP (Metabolic response predictor using coupled Multilayer Perceptrons) to fill in this gap. We provide clear evidence that McMLP outperforms existing methods on both synthetic data generated by the microbial consumer-resource model and real data obtained from six dietary intervention studies. Furthermore, we perform sensitivity analysis of McMLP to infer the tripartite food-microbe-metabolite interactions, which are then validated using the ground-truth (or literature evidence) for synthetic (or real) data, respectively. The presented tool has the potential to inform the design of microbiota-based personalized dietary strategies to achieve precision nutrition.

Gut Microbiome-Host Metabolome Homeostasis upon Exposure to PFOS and GenX in Male Mice.

Alterations of the normal gut microbiota can cause various human health concerns. Environmental chemicals are one of the drivers of such disturbances. The aim of our study was to examine the effects of exposure to perfluoroalkyl and polyfluoroalkyl substances (PFAS)-specifically, perfluorooctane sulfonate (PFOS) and 2,3,3,3-tetrafluoro-2-(heptafluoropropoxy) propanoic acid (GenX)-on the microbiome of the small intestine and colon, as well as on liver metabolism. Male CD-1 mice were exposed to PFOS and GenX in different concentrations and compared to controls. GenX and PFOS were found to have different effects on the bacterial community in both the small intestine and colon based on 16S rRNA profiles. High GenX doses predominantly led to increases in the abundance of Clostridium sensu stricto, Alistipes, and Ruminococcus, while PFOS generally altered Lactobacillus, Limosilactobacillus, Parabacteroides, Staphylococcus, and Ligilactobacillus. These treatments were associated with alterations in several important microbial metabolic pathways in both the small intestine and colon. Untargeted LC-MS/MS metabolomic analysis of the liver, small intestine, and colon yielded a set of compounds significantly altered by PFOS and GenX. In the liver, these metabolites were associated with the important host metabolic pathways implicated in the synthesis of lipids, steroidogenesis, and in the metabolism of amino acids, nitrogen, and bile acids. Collectively, our results suggest that PFOS and GenX exposure can cause major perturbations in the gastrointestinal tract, aggravating microbiome toxicity, hepatotoxicity, and metabolic disorders.

Transformer Neural Networks for Protein Family and Interaction Prediction Tasks.

The scientific community is rapidly generating protein sequence information, but only a fraction of these proteins can be experimentally characterized. While promising deep learning approaches for protein prediction tasks have emerged, they have computational limitations or are designed to solve a specific task. We present a Transformer neural network that pre-trains task-agnostic sequence representations. This model is fine-tuned to solve two different protein prediction tasks: protein family classification and protein interaction prediction. Our method is comparable to existing state-of-the-art approaches for protein family classification while being much more general than other architectures. Further, our method outperforms other approaches for protein interaction prediction for two out of three different scenarios that we generated. These results offer a promising framework for fine-tuning the pre-trained sequence representations for other protein prediction tasks.

Predicting metabolomic profiles from microbial composition through neural ordinary differential equations.

Characterizing the metabolic profile of a microbial community is crucial for understanding its biological function and its impact on the host or environment. Metabolomics experiments directly measuring these profiles are difficult and expensive, while sequencing methods quantifying the species composition of microbial communities are well-developed and relatively cost-effective. Computational methods that are capable of predicting metabolomic profiles from microbial compositions can save considerable efforts needed for metabolomic profiling experimentally. Yet, despite existing efforts, we still lack a computational method with high prediction power, general applicability, and great interpretability. Here we develop a method - mNODE (Metabolomic profile predictor using Neural Ordinary Differential Equations), based on a state-of-the-art family of deep neural network models. We show compelling evidence that mNODE outperforms existing methods in predicting the metabolomic profiles of human microbiomes and several environmental microbiomes. Moreover, in the case of human gut microbiomes, mNODE can naturally incorporate dietary information to further enhance the prediction of metabolomic profiles. Besides, susceptibility analysis of mNODE enables us to reveal microbe-metabolite interactions, which can be validated using both synthetic and real data. The presented results demonstrate that mNODE is a powerful tool to investigate the microbiome-diet-metabolome relationship, facilitating future research on precision nutrition.

Fine-scale diversity of microbial communities due to satellite niches in boom and bust environments.

Recent observations have revealed that closely related strains of the same microbial species can stably coexist in natural and laboratory settings subject to boom and bust dynamics and serial dilutions, respectively. However, the possible mechanisms enabling the coexistence of only a handful of strains, but not more, have thus far remained unknown. Here, using a consumer-resource model of microbial ecosystems, we propose that by differentiating along Monod parameters characterizing microbial growth rates in high and low nutrient conditions, strains can coexist in patterns similar to those observed. In our model, boom and bust environments create satellite niches due to resource concentrations varying in time. These satellite niches can be occupied by closely related strains, thereby enabling their coexistence. We demonstrate that this result is valid even in complex environments consisting of multiple resources and species. In these complex communities, each species partitions resources differently and creates separate sets of satellite niches for their own strains. While there is no theoretical limit to the number of coexisting strains, in our simulations, we always find between 1 and 3 strains coexisting, consistent with known experiments and observations.

Mitigation of SARS-CoV-2 transmission at a large public university.

In Fall 2020, universities saw extensive transmission of SARS-CoV-2 among their populations, threatening health of the university and surrounding communities, and viability of in-person instruction. Here we report a case study at the University of Illinois at Urbana-Champaign, where a multimodal "SHIELD: Target, Test, and Tell" program, with other non-pharmaceutical interventions, was employed to keep classrooms and laboratories open. The program included epidemiological modeling and surveillance, fast/frequent testing using a novel low-cost and scalable saliva-based RT-qPCR assay for SARS-CoV-2 that bypasses RNA extraction, called covidSHIELD, and digital tools for communication and compliance. In Fall 2020, we performed >1,000,000 covidSHIELD tests, positivity rates remained low, we had zero COVID-19-related hospitalizations or deaths amongst our university community, and mortality in the surrounding Champaign County was reduced more than 4-fold relative to expected. This case study shows that fast/frequent testing and other interventions mitigated transmission of SARS-CoV-2 at a large public university.

A cross-study analysis of drug response prediction in cancer cell lines.

To enable personalized cancer treatment, machine learning models have been developed to predict drug response as a function of tumor and drug features. However, most algorithm development efforts have relied on cross-validation within a single study to assess model accuracy. While an essential first step, cross-validation within a biological data set typically provides an overly optimistic estimate of the prediction performance on independent test sets. To provide a more rigorous assessment of model generalizability between different studies, we use machine learning to analyze five publicly available cell line-based data sets: National Cancer Institute 60, ancer Therapeutics Response Portal (CTRP), Genomics of Drug Sensitivity in Cancer, Cancer Cell Line Encyclopedia and Genentech Cell Line Screening Initiative (gCSI). Based on observed experimental variability across studies, we explore estimates of prediction upper bounds. We report performance results of a variety of machine learning models, with a multitasking deep neural network achieving the best cross-study generalizability. By multiple measures, models trained on CTRP yield the most accurate predictions on the remaining testing data, and gCSI is the most predictable among the cell line data sets included in this study. With these experiments and further simulations on partial data, two lessons emerge: (1) differences in viability assays can limit model generalizability across studies and (2) drug diversity, more than tumor diversity, is crucial for raising model generalizability in preclinical screening.

Complementary resource preferences spontaneously emerge in diauxic microbial communities.

Many microbes grow diauxically, utilizing the available resources one at a time rather than simultaneously. The properties of communities of microbes growing diauxically remain poorly understood, largely due to a lack of theory and models of such communities. Here, we develop and study a minimal model of diauxic microbial communities assembling in a serially diluted culture. We find that unlike co-utilizing communities, diauxic community assembly repeatably and spontaneously leads to communities with complementary resource preferences, namely communities where species prefer different resources as their top choice. Simulations and theory explain that the emergence of complementarity is driven by the disproportionate contribution of the top choice resource to the growth of a diauxic species. Additionally, we develop a geometric approach for analyzing serially diluted communities, with or without diauxie, which intuitively explains several additional emergent community properties, such as the apparent lack of species which grow fastest on a resource other than their most preferred resource. Overall, our work provides testable predictions for the assembly of natural as well as synthetic communities of diauxically shifting microbes.

Stochastic social behavior coupled to COVID-19 dynamics leads to waves, plateaus, and an endemic state.

It is well recognized that population heterogeneity plays an important role in the spread of epidemics. While individual variations in social activity are often assumed to be persistent, that is, constant in time, here we discuss the consequences of dynamic heterogeneity. By integrating the stochastic dynamics of social activity into traditional epidemiological models, we demonstrate the emergence of a new long timescale governing the epidemic, in broad agreement with empirical data. Our stochastic social activity model captures multiple features of real-life epidemics such as COVID-19, including prolonged plateaus and multiple waves, which are transiently suppressed due to the dynamic nature of social activity. The existence of a long timescale due to the interplay between epidemic and social dynamics provides a unifying picture of how a fast-paced epidemic typically will transition to an endemic state.

Time-dependent heterogeneity leads to transient suppression of the COVID-19 epidemic, not herd immunity.

Epidemics generally spread through a succession of waves that reflect factors on multiple timescales. On short timescales, superspreading events lead to burstiness and overdispersion, whereas long-term persistent heterogeneity in susceptibility is expected to lead to a reduction in both the infection peak and the herd immunity threshold (HIT). Here, we develop a general approach to encompass both timescales, including time variations in individual social activity, and demonstrate how to incorporate them phenomenologically into a wide class of epidemiological models through reparameterization. We derive a nonlinear dependence of the effective reproduction number [Formula: see text] on the susceptible population fraction S. We show that a state of transient collective immunity (TCI) emerges well below the HIT during early, high-paced stages of the epidemic. However, this is a fragile state that wanes over time due to changing levels of social activity, and so the infection peak is not an indication of long-lasting herd immunity: Subsequent waves may emerge due to behavioral changes in the population, driven by, for example, seasonal factors. Transient and long-term levels of heterogeneity are estimated using empirical data from the COVID-19 epidemic and from real-life face-to-face contact networks. These results suggest that the hardest hit areas, such as New York City, have achieved TCI following the first wave of the epidemic, but likely remain below the long-term HIT. Thus, in contrast to some previous claims, these regions can still experience subsequent waves.

Structured sequences emerge from random pool when replicated by templated ligation.

The central question in the origin of life is to understand how structure can emerge from randomness. The Eigen theory of replication states, for sequences that are copied one base at a time, that the replication fidelity has to surpass an error threshold to avoid that replicated specific sequences become random because of the incorporated replication errors [M. Eigen, Naturwissenschaften 58 (10), 465-523 (1971)]. Here, we showed that linking short oligomers from a random sequence pool in a templated ligation reaction reduced the sequence space of product strands. We started from 12-mer oligonucleotides with two bases in all possible combinations and triggered enzymatic ligation under temperature cycles. Surprisingly, we found the robust creation of long, highly structured sequences with low entropy. At the ligation site, complementary and alternating sequence patterns developed. However, between the ligation sites, we found either an A-rich or a T-rich sequence within a single oligonucleotide. Our modeling suggests that avoidance of hairpins was the likely cause for these two complementary sequence pools. What emerged was a network of complementary sequences that acted both as templates and substrates of the reaction. This self-selecting ligation reaction could be restarted by only a few majority sequences. The findings showed that replication by random templated ligation from a random sequence input will lead to a highly structured, long, and nonrandom sequence pool. This is a favorable starting point for a subsequent Darwinian evolution searching for higher catalytic functions in an RNA world scenario.

Ecology-guided prediction of cross-feeding interactions in the human gut microbiome.

Understanding a complex microbial ecosystem such as the human gut microbiome requires information about both microbial species and the metabolites they produce and secrete. These metabolites are exchanged via a large network of cross-feeding interactions, and are crucial for predicting the functional state of the microbiome. However, till date, we only have information for a part of this network, limited by experimental throughput. Here, we propose an ecology-based computational method, GutCP, using which we predict hundreds of new experimentally untested cross-feeding interactions in the human gut microbiome. GutCP utilizes a mechanistic model of the gut microbiome with the explicit exchange of metabolites and their effects on the growth of microbial species. To build GutCP, we combine metagenomic and metabolomic measurements from the gut microbiome with optimization techniques from machine learning. Close to 65% of the cross-feeding interactions predicted by GutCP are supported by evidence from genome annotations, which we provide for experimental testing. Our method has the potential to greatly improve existing models of the human gut microbiome, as well as our ability to predict the metabolic profile of the gut.

Emergency ventilator for COVID-19.

The COVID-19 pandemic disrupted the world in 2020 by spreading at unprecedented rates and causing tens of thousands of fatalities within a few months. The number of deaths dramatically increased in regions where the number of patients in need of hospital care exceeded the availability of care. Many COVID-19 patients experience Acute Respiratory Distress Syndrome (ARDS), a condition that can be treated with mechanical ventilation. In response to the need for mechanical ventilators, designed and tested an emergency ventilator (EV) that can control a patient's peak inspiratory pressure (PIP) and breathing rate, while keeping a positive end expiratory pressure (PEEP). This article describes the rapid design, prototyping, and testing of the EV. The development process was enabled by rapid design iterations using additive manufacturing (AM). In the initial design phase, iterations between design, AM, and testing enabled a working prototype within one week. The designs of the 16 different components of the ventilator were locked by additively manufacturing and testing a total of 283 parts having parametrically varied dimensions. In the second stage, AM was used to produce 75 functional prototypes to support engineering evaluation and animal testing. The devices were tested over more than two million cycles. We also developed an electronic monitoring system and with automatic alarm to provide for safe operation, along with training materials and user guides. The final designs are available online under a free license. The designs have been transferred to more than 70 organizations in 15 countries. This project demonstrates the potential for ultra-fast product design, engineering, and testing of medical devices needed for COVID-19 emergency response.