Biology subjects

Liu, Y.-Y.

Publications and source records attributed to Liu, Y.-Y..

At least 19 recordsLinked to original sources

bioRxiv2026

AI-guided discovery of antimicrobial peptides for urinary tract infections leveraging a new catalogue of the human urinary microbiome

Urinary tract infections (UTIs) are common infections that pose a critical burden on healthcare and society. Despite growing recognition that the human urinary tract harbors its own microbiome, its composition, functional potential, and alterations in UTI remain limited. Here, we leveraged the publicly available whole-metagenome shotgun sequencing data from 450 urinary microbiome samples collected in four independent cohorts together with genome assembly and metagenomic binning to construct an extensive human urinary microbiome catalog consisting of [~]1.3 million non-redundant microbial genes and 705 non-redundant metagenome-assembled genomes (nrMAGs). We found that microbiomes from patients with UTI carry significantly more genes linked to antibiotic resistance and virulence vs controls. There was an enrichment of multiple Escherichia strains in patients with UTI from two independent case-control cohorts. UTIs are becoming multidrug-resistant, and we used machine learning models to identify potential antimicrobial peptides (AMPs) in 705 nrMAGs. Furthermore, we experimentally demonstrated that two of these AMPs exhibited strong inhibitory activity against uropathogenic Escherichia coli strains. Our study provides a valuable resource for studying the human urinary microbiome and suggests urinary microbiome-derived AMPs represent a source of new therapeutics for UTIs.

microbiology↗

bioRxiv2026

Oral Gut Microbial Axis in Inflammatory Bowel Disease and Primary Sclerosing Cholangitis

The oral cavity is increasingly recognized as a reservoir of microbes that can translocate to and influence the gut microbiome. This oral-gut microbial axis may contribute to the pathogenesis of chronic gastrointestinal and hepatobiliary disorders, including inflammatory bowel disease (IBD) and its comorbidity, primary sclerosing cholangitis (PSC). To investigate the oral-gut microbial axis in IBD and PSC, we enrolled 191 participants spanning Crohns disease (CD), ulcerative colitis (UC), CD with PSC (CD-PSC), UC with PSC (UC-PSC), and healthy controls. We generated and analyzed the whole-metagenome shotgun sequencing data from saliva, tongue swabs, and fecal samples. Across oral and gut niches, we identified multiple microbial species differentially enriched in participants with UC compared with healthy controls, including Fusobacterium nucleatum and Gemella sanguinis in saliva, Catonella massiliensis and Tannerella serpentiformis on the tongue, and G. sanguinis and Sellimonas intestinalis in feces. Paired oral-gut analyses revealed 15 potential oral-origin species enriched in fecal samples; notably, G. sanguinis and Veillonella rogosae were consistently enriched in participants with UC. These findings were further supported by an external validation dataset comprising 1,716 gut metagenomes from five independent IBD cohorts, highlighting the reproducibility of oral microbial signatures. In exploratory analyses, smoking history was associated with shifts in the oral microbiome toward a UC-like state in healthy controls, suggesting a possible environmental modifier of the oral-gut microbial axis. Collectively, our study identifies distinct oral microbial signatures and their potential translocation to the gut in IBD and PSC, underscoring the role of the oral- gut microbial axis in disease progression.

microbiology↗

bioRxiv2026

Machine learning-based Personalized Dietary Recommendations to Achieve Desired Gut Microbial Compositions

Dietary intervention is an effective way to alter the gut microbiome to promote human health. Yet, due to our limited knowledge of diet-microbe interactions and the highly personalized gut microbial compositions, an efficient method to prescribe personalized dietary recommendations to achieve desired gut microbial compositions is still lacking. Here, we propose a machine learning framework to resolve this challenge. Our key idea is to implicitly learn the diet-microbe interactions by training a machine learning model using paired gut microbiome and dietary intake data from a population-level cohort. The well-trained machine learning model enables us to predict the microbial composition of any given species collection and dietary intake. Next, we prescribe personalized dietary recommendations by solving an optimization problem to achieve the desired microbial compositions. We systematically validated this Machine learning-based Personalized Dietary Recommendation (MPDR) framework using synthetic data generated from an established microbial consumer-resource model. We then validated MPDR using real data collected from a diet-microbiome association study. The presented MPDR framework demonstrates the potential of machine learning for personalized nutrition.

bioinformatics↗

bioRxiv2026

The human gut virome is a non-redundant and clinically informative component of the microbiome

The gut virome represents a vast reservoir of genetic diversity with profound implications for human health, yet it remains the "dark matter" of the microbiome due to the staggering complexity of reproducible viral profiling. It remains fundamentally contested whether biologically informative virome signals can be robustly recovered from routine whole-metagenome sequencing (WMS), and to what extent these signals offer ecological insights independent of the bacteriome. Here we present VIP2B, a framework that leverages Type IIB restriction tags to extract multifaceted viral features (encompassing taxonomy, coverage, function, and phenotype) directly from bulk WMS data. Through extensive benchmarking across incomplete references, unseen genomes, and high bacterial or host background, we demonstrate that VIP2B achieved high precision and robust taxonomic concordance. By applying VIP2B to paired bulk and virus-like particle (VLP)-enriched datasets, we reveal a species-level overlap far greater than previously recognized, proving that standard bulk metagenomes contain a wealth of recoverable viral information. Analysis of 20 clinical cohorts demonstrates that coverage-, function-, and phenotype-resolved viral features consistently identify disease-associated signatures that escape taxonomic analysis alone, significantly improving diagnostic models over bacteriome-only approaches. Finally, we define two distinct gut virome community states at the population scale (n=6,090), characterized by divergent diversity profiles and health associations. Our findings establish the gut virome as a non-redundant, clinically actionable component of the human holobiont and provide the methodology necessary to transition microbiome research toward a truly multi-kingdom framework.

bioinformatics↗

bioRxiv2025

Neocortical temporal patterning by a two-layered regulatory network

In the developing neocortex, a diverse array of neurons with defined types and abundances are systematically generated by a limited population of radial glial progenitors (RGPs) as they undergo successive fate changes. The molecular regulation behind this intricate temporal patterning remains elusive. We undertook in-depth single-cell multi-omics analyses, discovering a two-layered regulatory framework at the core of this process. Central to this are global temporal regulators positioned above a temporal network, consisting of series of transcriptional factor (TF) hub groups under sequential state transitions. This temporal network operates not by restricting TF expressions to discrete temporal windows, but through coordinated transcriptional and chromatin-accessibility dynamics that modulate transient TF regulatory activity. Moreover, global temporal regulators specify the duration of each cascading stage and, consequently, the number of progenies generated at each stage. Loss of global temporal regulators protracts RGP lineage progression, whereas their increased activity accelerates it. These findings suggest a two-layer temporal regulatory system controlling RGP lineage progression and neural progeny output duality in mammalian neocortical development.

neuroscience↗

bioRxiv2025

Revealing Interactions between Microbes, Metabolites, and Dietary Compounds using Genome-scale Analysis

BackgroundThe role of gut microbiome in predicting diet response and developing personalized dietary recommendations has been increasingly recognized. Yet, we still lack comprehensive, genome-based insights into which gut microbes metabolize specific dietary compounds. ResultsHere, we leveraged the metabolic networks constructed from well-annotated microbial genomes to characterize the potential interactions between microbes and metabolites, specifically emphasizing the interactions between microbes and dietary compounds. We revealed a substantial, approximately four-fold variation in both the number of metabolites and dietary compounds in the microbial genome-scale metabolic networks across different genera, whereas species within the same genus showed a high metabolic similarity (mean coefficient of variation in microbial network degree CV = 0.023 for metabolites and 0.015 for dietary compounds). We found that the number of species that can utilize a metabolite drastically varies, ranging from 1 to 818 species, with some metabolites being used by a wide range of species (211 out of 1390 metabolites used by more than 95% of species) and others only by a few species (435 metabolites used by less than 5% of species). Leveraging a longitudinal microbiome study, we observed that microbial taxa with similar metabolic capacity tend to have positively correlated abundances, and the gut microbiomes capacity to process dietary compounds is functionally stable. Finally, we propose a network-based method to identify the dietary compounds that are specific to no more than 10 microbial species, offering a new strategy for combining a dietary compound and its linked microbial species to design synbiotics. ConclusionsOur results quantitatively reveal large-scale variation and redundancy in gut microbial metabolism and identify dietary compounds linked to only a few microbial species. These findings improve understanding of microbe-metabolite interactions and provide a foundation for the rational design of microbiome-based interventions for healthy benefits.

bioinformatics↗

bioRxiv2025

Suppression of Glucosylceramide Synthase Reverses Drug Resistance in Cancer Cells Harbor Homozygous p53 Mutants

Glucosylceramide synthase (GCS) catalyzes ceramide glycosylation in response to cell stress that produces glucosylceramide and other glycosphingolipids. GCS overexpression is a cause of drug resistance and enriches cancer stem cells (CSCs) during cancer chemotherapy. Previous studies showed that GCS modulates expression of p53 mutants and oncogenic gain-of-function (GOF) in heterozygous knock-in cell models (TP53 R273H-/+). However, it is unclear whether GCS can modulate the effects of homozygous p53 mutations, which are common in many cancer cases. We report herewith that inhibition of GCS, via UGCG-knockout and using new inhibitor (Genz-161), effectively re-sensitizes drug resistance and diminishes CSCs in colon cancer cells carrying the homozygous p53 R273H mutation. In aggressive WiDr cells carrying TP53 R273H mutation, knockout of UGCG gene using CRISPR/Cas9 editing or inhibition of GCS with Genz-161 sensitized cancer cells to oxaliplatin, irinotecan and paclitaxel. With decreased ceramide glycosylation in lipidomic profiling, both UGCG-knockout and Genz-161 treatments substantially decreased wound healing, and diminished CSCs and tumor growth under chemotherapy. Interestingly, inhibition of RNA m6A methylation by neplanocin A reactivated p53 function and reversed drug resistance. Mechanistic investigation revealed that GCS inhibition downregulated METTL3 expression and repressed RNA-m6A modification on mutant p53 R273H effects. Altogether, our findings demonstrate that ceramide glycosylation promotes METTL3 expression and RNA m6A methylation in response to drug-induced stress, thereby promoting mutant p53 expression and associated GOF. Conversely, inhibition of GCS can diminish CSCs and drug resistance via reduction of m6A modification and reactivation of p53 function. GCS inhibition is an achievable approach for mutant cancer treatment.

cancer biology↗

bioRxiv2025

Adapting CRISPR-associated transposons for rapid and high-throughput reverse genetics

CRISPR-associated transposons (CAST) use guide RNAs to direct their transposition and are being harnessed as tools for programmable genome engineering across diverse bacterial species. However, CAST systems have not been adapted for high-throughput genetic screening. Here, we present MultiCAST, a streamlined platform for rapid and scalable guide RNA-directed transposon insertion in bacteria. MultiCAST generates targeted insertions in a single step through conjugative delivery of conditionally replicative plasmids encoding the CAST enzymatic machinery and a selectable mini-transposon expressing the guide RNA. By leveraging the inserted guide sequence as a molecular barcode, MultiCAST enables pooled, high-throughput genetic screens using only amplicon sequencing. We identified factors that influence transposition efficiency and the accuracy of insertion frequency measurements derived from guide sequencing. Adjusting the ratio of donor and recipient strain during conjugation mitigates guide-transposon "crosstalk", in which a single recipient cell acquires multiple donor plasmids containing distinct guides. Furthermore, we developed a machine learning-based predictive model for selecting highly active guides based on target sequence features that strongly correlate with activity. The nucleoid-associated protein H-NS was also found to inhibit CAST activity, providing a mechanistic explanation for variable insertion frequencies among non-essential genes. To demonstrate the scalability of MultiCAST, we screened a pooled mutant population created from >5,200 guides targeting 88 genes in E. coli across twelve nutrient conditions, accurately identifying genes with condition-specific fitness effects. The simplicity, speed, and throughput of MultiCAST make genome-scale functional screens more accessible across a wide range of bacterial species. SignificanceEfficient gene disruption is essential for understanding bacterial gene function, but traditional genetic approaches are labor-intensive and generally not well-suited for high-throughput studies. We developed MultiCAST, a simple and scalable method that harnesses guide RNA-directed CRISPR-associated transposons for targeted bacterial gene disruption. MultiCAST enables single and pooled transposon mutagenesis in a single step and eliminates the need for complex sequencing library preparation protocols by using the guide sequence as a quantifiable surrogate for mutant abundance. This approach allows thousands of mutants to be generated and screened simultaneously across multiple conditions using only amplicon sequencing. By dramatically reducing the time, cost, and complexity of reverse genetics, MultiCAST opens new possibilities for genome-scale functional studies, accelerating the discovery of bacterial gene functions.

microbiology↗

bioRxiv2025

An efficient artificial esterase with a dynamic conformation via conformational engineering

The catalytic turnover number (kcat) of current enzyme mimics and de novo designed enzymes is still orders of magnitude lower than that of natural enzymes, because mimicking the fast-dynamic feature of enzyme active centers remains a great challenge. Herein, we created a gold nanoparticle-based artificial esterase, Goldenzyme, by conformational engineering to reconstruct dynamic active centers resembling that of -chymotrypsin. NMR characterization shows that Goldenzyme possesses a fast-dynamic feature, about 6-fold faster than typical -helix. Therefore, Goldenzyme efficiently hydrolyzes p-nitrophenyl acetate with a net kcat of 6.37 s-1 per active center, more than 5-fold of that of -chymotrypsin (1.19 s-1), and Goldenzyme can even hydrolyze some non-activated esters that -chymotrypsin cannot. Our conformational engineering approach demonstrates the possibility of creating artificial enzymes that surpass natural ones.

biochemistry↗

bioRxiv2025

Advancing the Discovery of Phage-Host Interactions and Disease Classification from Metagenomic Profiles Using Deep Learning

Microbial communities are essential for sustaining ecosystem functions in diverse environments, including the human gut. Phages interact dynamically with their prokaryotic hosts and play a crucial role in shaping the structure and function of microbial communities. Previous approaches for inferring phage-host interactions (PHIs) from metagenomic data are constrained by low sensitivity and the inability to accurately capture ecological relationships. To overcome these limitations, we developed PHILM (Phage-Host Interaction Learning from Metagenomic profiles), a deep learning framework that predicts PHIs directly from the taxonomic profiles of metagenomic data. We validated PHILM on both synthetic datasets generated by ecological models and real-world data, finding that it consistently outperformed the co-abundance-based approach for inferring PHIs. When applied to a large-scale metagenomic dataset comprising 7,016 stool samples from healthy individuals, PHILM identified 90% more genus-level PHIs than the traditional assembly-based approach. In a longitudinal dataset tracking PHI dynamics, PHILMs latent representations recapitulated microbial succession patterns originally described using taxonomic abundances. Furthermore, we demonstrated that PHILMs latent representations served as more discriminative features than taxonomic abundance-based features for disease classifications. In summary, PHILM represents a novel computational framework for predicting phage-host interactions from metagenomic data, offering valuable insights for both microbiome science and translational medicine.

bioinformatics↗

bioRxiv2024

Optimization-based dietary recommendations for healthy eating

Various diet scores have been developed to assess compliance with dietary guidelines. Yet, enhancing those diet scores is very challenging. Here, we tackle this issue by formalizing an optimization problem and solving it with simulated annealing. Our optimization-based dietary recommendation (ODR) approach, evaluated using Diet-Microbiome Association study data, provides efficient and reasonable recommendations for different diet scores. ODR has the potential to enhance nutritional counseling and promote dietary adherence for healthy eating.

bioinformatics↗

bioRxiv2024

Competition and Compromise between Exogenous Probiotics and Native Microbiota

Probiotic interventions are effective strategies to modulate the gut microbiome, but how exogenous probiotics compete with native gut microbiota remains elusive. Here, we use a mouse model and a well-documented probiotic Bifidobacterium animalis subsp. lactis V9 (BV9) to mechanistically investigate its competitive strategies. We perform metagenomic and whole genome sequencing of stool samples and isolated BV9, longitudinally collected from 24 mice orally administered with BV9 and different diets. Results show that a high-fiber diet most effectively supports the colonization of BV9, where BV9 selectively competes with Parabacteroides distasonis (P. distasonis), rather than extensively with other gut bacteria. By comparing the genomic structures of BV9 and P. distasonis isolated during the washout period, we infer their co-evolution mechanisms, highlighting their competition and compromise in utilizing inulin-derived glucose. Finally, our in vitro co-culture experiments validate such competitive dynamics. This study fills a critical gap in our understanding of niche competition in colonization.

microbiology↗

bioRxiv2024

Rational Design of Live Biotherapeutic Products for the Prevention of Clostridioides difficile Infection

Clostridioides difficile infection (CDI) is one of the leading causes of healthcare- and antibiotic-associated diarrhea. While fecal microbiota transplantation (FMT) has emerged as a promising therapy for recurrent CDI, its exact mechanisms of action and long-term safety are not fully understood. Defined consortia of clonal bacterial isolates, known as live biotherapeutic products (LBPs), have been proposed as an alternative therapeutic option. However, the rational design of LBPs remains challenging. Here, we employ a computational pipeline and three independent metagenomic datasets to systematically identify microbial strains that have the potential to inhibit CDI. We first constructed the CDI-related microbial genome catalog, comprising 3,741 non-redundant metagenome-assembled genomes (nrMAGs) at the strain level. We then identified multiple potential protective nrMAGs that can be candidates for the design of microbial consortia targeting CDI, including strains from Dorea formicigenerans, Oscillibacter welbionis, and Faecalibacterium prausnitzii. Importantly, some of these potential protective nrMAGs were found to play an important role in the success of FMT, and the majority of the top protective nrMAGs can be validated by various previously reported findings. Our results demonstrate a computational framework for the rational selection of microbial strains targeting CDI, paving the way for the computational design of microbial consortia against other enteric infections.

microbiology↗

bioRxiv2023

Perception of Motion Salience Shapes the Emergence of Collective Motions

As one of the most common and spectacular manifestations of coordinated behavior, collective motion is the spontaneous emergence of the ordered movement in a system consisting of many self-propelled agents, e.g., flocks of birds, schools of fish, herds of animals, and human crowds. Despite extensive studies on collective motions, a systems-level understanding of different motion patterns of collective behaviors is still lacking. This further hinders the adoption of bio-inspired mechanisms for applications of swarm robotics. Here, by leveraging three large bird-flocking datasets, we systematically investigate the emergence of different patterns of collective motions: mobbing, circling, and transit. We find that flocks with higher maneuverable motions (i.e., mobbing and circling) prefer to evolve a more nested structure of leader-follower (LF) relations and a clear hierarchy to mitigate the damage of individual freedom to group cohesion. In contrast, flocks with smooth motion (i.e., transit) do not display this tactful strategy to organize the group. To explain this empirical finding, we propose a measure based on the perception of motion salience (MS) to quantify the trade-off between individual freedom and group cohesion. Moreover, we perform the correlation analysis between LF and MS, finding that individuals with higher MS tend to lead the group with higher maneuverable motions. Those findings prompt us to develop a swarm model with adaptive MS-based (AMS) interactions and confirm that AMS interactions are responsible for the emergence of nested and hierarchical LF relations in the flocks with highly maneuverable motions. Finally, we implement AMS interactions in swarm robotics that consists of [~]102 miniature mobile robots. The swarm experiments of collective following and collective evacuation demonstrate that AMS interactions not only empower the swarm to promptly respond to the transient perturbation but also strengthen the self-organization of collective motions in terms of temporal cognition.

animal behavior and cognition↗

bioRxiv2023

Microbiome-based correction of nutrient profiles derived from self-reported dietary assessments

Since dietary intake is challenging to directly measure in large-scale cohort studies, we often rely on self-reported instruments (e.g., food frequency questionnaires, 24-hour recalls, and diet records) developed in nutritional epidemiology. Those self-reported instruments are prone to measurement errors, which can lead to inaccuracies in the calculation of nutrient profiles. Currently, few computational methods exist to address this problem. In the present study, we introduce a deep-learning approach --- Microbiome-based nutrient profile corrector (METRIC), which leverages gut microbial compositions to correct random errors in self-reported dietary assessments using 24-hour recalls or diet records. We demonstrate the excellent performance of METRIC in minimizing the simulated random errors, particularly for nutrients metabolized by gut bacteria in both synthetic and three real-world datasets. Further research is warranted to examine the utility of METRIC to correct actual measurement errors in self-reported dietary assessment instruments.

microbiology↗

bioRxiv2023

Data-driven prediction of colonization outcomes for complex microbial communities

Complex microbial interactions can lead to different colonization outcomes of exogenous species, be they pathogenic or beneficial in nature. Predicting the colonization of exogenous species in complex communities remains a fundamental challenge in microbial ecology, mainly due to our limited knowledge of the diverse physical, biochemical, and ecological processes governing microbial dynamics. Here, we proposed a data-driven approach independent of any dynamics model to predict colonization outcomes of exogenous species from the baseline compositions of microbial communities. We systematically validated this approach using synthetic data, finding that machine learning models (including Random Forest and neural ODE) can predict not only the binary colonization outcome but also the post-invasion steady-state abundance of the invading species. Then we conducted colonization experiments for two commensal gut bacteria species Enterococcus faecium and Akkermansia muciniphila in hundreds of human stool-derived in vitro microbial communities, confirming that the data-driven approach can successfully predict the colonization outcomes. Furthermore, we found that while most resident species were predicted to have a weak negative impact on the colonization of exogenous species, strongly interacting species could significantly alter the colonization outcomes, e.g., the presence of Enterococcus faecalis inhibits the invasion of E. faecium. The presented results suggest that the data-driven approach is a powerful tool to inform the ecology and management of complex microbial communities.

ecology↗

bioRxiv2023

Feasibility in MacArthur's Consumer-Resource Model

Finding the conditions that ensure the survival of species has occupied ecologists for decades. Theoretically, for mechanistic models such as MacArthurs consumer-resource model, most of the efforts have concentrated on proving the stability of an equilibrium assuming that it is feasible, but overlooking the conditions that ensure its feasibility. Here we address this gap by finding the range of conditions that lead to a feasible equilibrium of MacArthurs consumer-resource model and study how changes in the systems structural and parametric properties affect those ranges. We characterize the relationship between the loss of feasibility and the increase in complexity (measured by the systems richness and connectance) by a power law that can be extended to random competition matrices. Focusing on the pool of consumers, we find that while the feasibility of the entire system decreases with the size of the pool, the expected fraction of feasible consumers increases --safety in consumer numbers. Focusing on the pool of resources, we find that if resources grow linearly, the larger the pool of resources, the lower the feasibility of the system and the expected fraction of feasible consumers --danger in resource numbers. However, if resources grow logistically, this pattern is reversed with a sublinear increase in feasibility, as it has been previously reported in experimental work. This work provides testable predictions for consumer-resource systems and is a gateway to exploring feasibility in other mechanistic models.

ecology↗

bioRxiv2023

Identifying keystone species in microbial communities using deep learning

Previous studies suggested that microbial communities harbor keystone species whose removal can cause a dramatic shift in microbiome structure and functioning. Yet, an efficient method to systematically identify keystone species in microbial communities is still lacking. This is mainly due to our limited knowledge of microbial dynamics and the experimental and ethical difficulties of manipulating microbial communities. Here, we propose a Data-driven Keystone species Identification (DKI) framework based on deep learning to resolve this challenge. Our key idea is to implicitly learn the assembly rules of microbial communities from a particular habitat by training a deep learning model using microbiome samples collected from this habitat. The well-trained deep learning model enables us to quantify the community-specific keystoneness of each species in any microbiome sample from this habitat by conducting a thought experiment on species removal. We systematically validated this DKI framework using synthetic data generated from a classical population dynamics model in community ecology. We then applied DKI to analyze human gut, oral microbiome, soil, and coral microbiome data. We found that those taxa with high median keystoneness across different communities display strong community specificity, and many of them have been reported as keystone taxa in literature. The presented DKI framework demonstrates the power of machine learning in tackling a fundamental problem in community ecology, paving the way for the data-driven management of complex microbial communities.

microbiology↗