Biology subjects

Weng, Z.

Publications and source records attributed to Weng, Z..

At least 19 recordsLinked to original sources

bioRxiv2026

A data model for longitudinal multi-modal biomedical cohorts

Longitudinal multimodal cohort studies require data models that can integrate heterogeneous molecular, phenotypic, and contextual data while preserving relationships across participants, biospecimens, study visits, and computational transformations. Here, we present the data model developed for the Multi-Omics for Health and Disease (MOHD) Consortium, a modality-agnostic, provenance-driven framework designed to support an evolving longitudinal biomedical resource. The model uses a compact set of linked entities and a central dataset abstraction to represent experimental, phenotypic and contextual, and computational data within a common framework. Stable identifiers, explicit provenance relationships, and flexible grouping structures enable data to be tracked across biological sources, processing workflows, and releases without requiring modality-specific restructuring of the core model. To address the complexity of participant context, we developed a harmonization framework that transforms branched REDCap survey data into a longitudinal, machine-readable representation with normalized variables and explicit response status. Interoperability with external data models and ontologies is supported through reproducible translation workflows while maintaining a stable canonical representation. We demonstrate the model through its implementation in the first production-scale MOHD data release, encompassing 956 participants, nearly 7,000 datasets, and more than 40,000 files across multiple molecular modalities and study visits. This implementation demonstrates the ability of the model to accommodate heterogeneous and longitudinal data, evolving computational products, participant withdrawal, and distinct data-access environments. Together, the MOHD implementation provides a practical framework and generalizable design principles for building longitudinal multimodal biomedical resources that remain traceable, interoperable, and adaptable as data and analytical requirements evolve.

bioinformatics↗

bioRxiv2026

Multi-source domain generalization with few-shot calibration for cross-dataset EEG state classification under proxy labels

Cross-dataset generalization of EEG-based classification under weak, proxy-derived labels remains an open problem for altered-states research. We present a reproducible eight-dataset alignment pipeline that maps eight heterogeneous EEG corpora (712,832 windows; 697,906 with valid labels) to a common 14-channel EPOC+ montage with 63-dimensional spectral features, and we recover the real 1-9 arousal self-assessments for MAHNOB-HCI from session.xml metadata. As a benchmark, Random Forest classifiers are trained on seven source domains and evaluated on the held-out target under both zero-shot and 20%-participant few-shot calibration. The benchmark exposes two concrete methodological pitfalls rather than a performance result: (i) per-class recall shows every target collapsing to a single majority class, and (ii) a within-dataset upper-bound experiment (Table 3) shows that six of eight proxy label sets sit at or below three-class chance even when trained and tested on the same dataset, so the cross-dataset failure is a label-validity problem rather than a transfer-method problem. Across the eight targets (20 seeds, 8,000 evaluation windows per target), zero-shot accuracy averages 36.85% (95% CI 34.40-39.30) and calibrated 43.76% (41.77-45.75), but balanced accuracy stays at 33.01-35.62% (Cohen's kappa <= 0.068), i.e. at chance. The +6.91pp mean change is driven almost entirely by a single target, ds006437 (6.31% -> 60.60%): the median paired change across all 160 seed-pairs is 0.00pp, and after Holm-Bonferroni correction only ds006437 and ds004572 remain significant, the latter with a practically null effect (+0.39pp). Balanced accuracy stays between 33.01% and 35.62% and Cohen's kappa at 0.009 +/- 0.032, i.e. at or barely above three-class chance, while per-class recall shows six of eight targets collapsing to Deep (96.7-100% recall) and two to Light (68.5-99.1%). The collapse persists under SMOTE oversampling, under an EEGNet-v4 deep-learning baseline, and under CORAL and AdaBN feature alignment, which locates the bottleneck in proxy-label validity and class overlap in the feature space rather than in classifier capacity. We position this work as a preliminary methodological study: its contribution is a reproducible eight-dataset alignment pipeline, recovered MAHNOB-HCI arousal self-assessments, a quantitative estimate of split-leakage inflation, and a transparently reported negative result rather than a performance claim.

neuroscience↗

bioRxiv2026

A linguistics-based algorithm for RBP motif and context discovery

RNA-binding proteins (RBPs) regulate their RNA targets by binding to short sequence motifs, but the underlying mechanisms enabling sequence-specific recognition within the vast transcriptome remain unclear for the majority of human RBPs. Sequence contexts are believed to be a significant contributing factor to RBP binding specificity but are often overlooked. Further, existing motif discovery algorithms do not consider the structure and composition of the motifs flanking regions in their construction, which represents a consequential shortcoming. Herein, we present a novel linguistics-inspired RBP motif and context discovery algorithm that is consensus-based, deterministic, and flexible. Our algorithm draws multiple parallels between natural language and genomic language and relies on three important k-mer properties that impart lexical, syntactic, and semantic structures and rules to the process of motif and context discovery. Critically, our algorithm integrates information from sequence contexts when constructing RBP motifs. We demonstrate that our algorithm achieves strong discovery accuracy against a ground-truth set, and even outperforms existing methods in primary motif ranking.

bioinformatics↗

bioRxiv2026

Natural language-based representation and modeling of RBP binding

RNA-binding proteins (RBPs) are critical regulators of the human transcriptome, but the binding patterns of most RBPs are insufficiently characterized. While sequence context facilitates RBP binding specificity, its precise contribution remains unclear. Existing computational methods to decipher RBP binding patterns are limited by their architecture-dependence, challenging interpretability, and, importantly, lack of focus on context. We present a novel comprehensive approach to address the aforementioned knowledge gaps. We first introduce a natural language-based representation to model RNA sequences using lexical, syntactic, and semantic forms, then devise a sequence decomposition method based on these structures to deconstruct RNA sequences into regions, each containing a target k-mer and its flanking contexts. We leverage this linguistic conceptualization to predict RBP binding under a Multiple Instance Learning (MIL) framework, which we solve using a novel method of significant region extraction termed "iterative relabeling". We demonstrate that our bottom-up approach discovers key regions contributing to RBP binding in an architecture-dependent, accurate, and interpretable manner.

bioinformatics↗

bioRxiv2025

Inferring binding specificities of human transcription factors with the wisdom of crowds

DNA motif discovery and, particularly, computational modeling of transcription factor binding motifs, has been a mecca of algorithmic bioinformatics for several decades. Here, we report the results of the largest open community challenge in Inferring BInding Specificities (IBIS), where participants all over the world were invited to construct binding specificity models from multi-assay experimental data for poorly studied human transcription factors. The submissions were rigorously tested against a rich held-out dataset. Benchmarking demonstrated a consistent advantage of properly designed deep learning models over traditional positional weight matrices and other machine learning methods. Yet, the positional weight matrices displayed a surprisingly strong performance out of the box, being only slightly behind the best deep learning models. A post-challenge assessment of a selection of other deep learning methods further solidified this finding. IBIS highlights the power of benchmarking in finding adequate DNA motif representations, emphasizes the pros and cons of various machine learning methods applied to DNA motif modeling, and establishes a rich dataset, benchmarking protocols, and computational framework for a fair cross-platform evaluation of future models of transcription factor binding motifs in DNA sequences. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=175 SRC="FIGDIR/small/688692v1_ufig1.gif" ALT="Figure 1"> View larger version (64K): org.highwire.dtl.DTLVardef@1c6677corg.highwire.dtl.DTLVardef@b4124aorg.highwire.dtl.DTLVardef@1ce2b1org.highwire.dtl.DTLVardef@66e917_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗

bioRxiv2025

An integrated multi-tissue atlas of epigenomic landscapes and regulatory elements in the bovine genome

Deciphering the regulatory syntax of the genome is essential to understand the genetic and molecular architecture of complex traits, as most trait-associated variants lie in non-coding regions. Yet, functional annotation of the bovine genome remains limited, hindering our ability to unravel the mechanisms underpinning complex traits of economic and ecological importance in cattle. Here, we present a comprehensive epigenetic atlas comprising 1,138 genome-wide epigenetic profiles, including chromatin accessibility, six histone modifications, CCCTC-binding factor (CTCF) transcription factor binding, DNA methylation, chromatin conformation, and transcriptomes across 53 adult tissues, five fetal tissues, and seven primary cell types. This atlas-level data enables us to annotate around 45% of the genome as putative regulatory elements exhibiting tissue- or cell-specific regulatory activity. Leveraging sequence-to-function deep learning models, we discovered 301 sequence motifs and predicted the functional impact of genetic variants through in silico mutagenesis, thereby facilitating the decoding of the regulatory syntax of the cattle genome and fine-mapping of GWAS loci for 22 complex traits. Cross-species analysis further revealed evolutionarily conserved features of regulatory architecture and provided evolutionary insights into complex traits and diseases in humans. Together, this atlas offers a foundational resource for advancing cattle functional genomics, sustainable breeding, and studies of regulatory evolution.

genomics↗

bioRxiv2025

LOCATE: using Long-read to Characterize All Transposable Elements

Transposons constitute [~]45% of the human genome, driving gene evolution and contributing to disease, but their repetitive nature complicates the identification of new insertions. We present LOCATE (Long-read to Characterize All Transposable Elements), an algorithm using long-read sequencing to detect and assemble transposon insertions. LOCATE outperforms existing tools on simulated datasets and achieves the best performance in two previous benchmarks, as well as in a new benchmark we constructed using real biological datasets. Applying LOCATE to public datasets revealed that pre-existing Alu copies create two hotspots for Alu and LINE1 insertions: the A-rich linker and the poly(A) tail. We further observed a preference for self-insertions over non-self-insertions in Alu and LINE1, suggesting a "feedforward" transposition mechanism in which Alu and LINE1 RNA transcripts target the hotspots of their source copies to generate new insertions. LOCATE enhances our ability to study transposons and their role in genome dynamics.

bioinformatics↗

bioRxiv2025

A novel NLP-based method and algorithm to discover RNA-binding protein (RBP) motifs, contexts, binding preferences, and interactions

RNA-binding proteins (RBPs) are essential modulators in the regulation of mRNA processing. The binding patterns, interactions, and functions of most RBPs are not well-characterized. Previous studies have shown that motif context is an important contributor to RBP binding specificity, but its precise role remains unclear. Despite recent computational advances to predict RBP binding, existing methods are challenging to interpret and largely lack a categorical focus on RBP motif contexts and RBP-RBP interactions. There remains a need for interpretable predictive models to disambiguate the contextual determinants of RBP binding specificity in vivo. Here, we present a novel and comprehensive pipeline to address these knowledge gaps. We devise a Natural Language Processing-based method to deconstruct sequences into entities comprising a target k-mer and its flanking regions, then use this representation to formulate RBP binding prediction as a weakly supervised Multiple Instance Learning problem. To interpret our predictions, we introduce a deterministic motif discovery algorithm to leverage our data structure, recapitulating the established motifs of numerous RBPs as validation. Importantly, we characterize the binding motifs and binding contexts for 71 RBPs in HepG2 and 74 RBPs in K562, with many of them being novel. Finally, through feature integration, transitive inference, and a new cross-prediction approach, we propose novel cooperative and competitive RBP-RBP interaction partners and hypothesize their potential regulatory functions. In summary, we present a complete framework for investigating the contextual determinants of specific RBP binding, and we demonstrate the significance of our findings in delineating RBP binding patterns, interactions, and functions.

genomics↗

bioRxiv2024

An Expanded Registry of Candidate cis-Regulatory Elements for Studying Transcriptional Regulation

Mammalian genomes contain millions of regulatory elements that control the complex patterns of gene expression. Previously, The ENCODE consortium mapped biochemical signals across many cell types and tissues and integrated these data to develop a Registry of 0.9 million human and 300 thousand mouse candidate cis-Regulatory Elements (cCREs) annotated with potential functions1. We have expanded the Registry to include 2.35 million human and 927 thousand mouse cCREs, leveraging new ENCODE datasets and enhanced computational methods. This expanded Registry covers hundreds of unique cell and tissue types, providing a comprehensive understanding of gene regulation. Functional characterization data from assays like STARR-seq, MPRA, CRISPR perturbation, and transgenic mouse assays now cover over 90% of human cCREs, revealing complex regulatory functions. We identified thousands of novel silencer cCREs and demonstrated their dual enhancer/silencer roles in different cellular contexts. Integrating the Registry with other ENCODE annotations facilitates genetic variation interpretation and trait-associated gene identification, exemplified by discovering KLF1 as a novel causal gene for red blood cell traits. This expanded Registry is a valuable resource for studying the regulatory genome and its impact on health and disease.

genomics↗

bioRxiv2024

Tissue clocks derived from histological signatures of biological aging enable tissue-specific aging predictions from blood

Aging, the leading risk factor for numerous diseases, manifests through diverse structural and architectural changes in human tissues, providing an opportunity to quantify and interpret tissue-specific aging. To address this, we present a comprehensive assessment of tissue changes occurring during human aging, utilizing a vast array of whole slide histopathological images from the Genotype-Tissue Expression Project (GTEx), primarily reflecting non-diseased tissue samples. Using deep learning, we analyzed 25,712 images from 40 distinct tissue types across 983 individuals, quantifying nuanced morphological changes that tissues undergo with age. We developed tissue clocks--predictors of biological age based on tissue images--which achieved a mean prediction error of 4.9 years. These clocks were associated with established aging markers, including telomere attrition, subclinical pathologies, and comorbidities. In a systematic assessment of biological age rates across organs, we identified pervasive non-uniform rates of aging across the human lifespan, with some organs exhibiting earlier changes (20-40 years old) and others showing bimodal patterns of age-related changes. We also uncovered several associations between demographic, lifestyle, and medical history factors and tissue-specific acceleration or deceleration of biological age, highlighting potential modifiable risk factors that influenced the aging process at the tissue level. Finally, by combining paired histological images and gene expression data, we developed a strategy to predict tissue-specific age gaps from blood samples. This approach was validated in independent cohorts covering eight diseases, ranging from acute conditions like stroke to chronic diseases such as cystic fibrosis and Alzheimers disease. It successfully recovered significant associations with disease-relevant organs and revealed patterns of systemic and tissue-specific aging that may reflect broader physiological changes in health and disease. This work offers a new perspective on the aging process by positioning tissue structure as an integrator of cellular and molecular changes that reflect the physiological state of organs in health and disease. It underscores the value of histopathological imaging as a tool for understanding human aging and provides a foundation for the monitoring of tissue-specific aging processes in age-associated diseases.

physiology↗

bioRxiv2024

A consensus variant-to-function score to functionally prioritize variants for disease

Identifying and functionally characterizing causal disease variants in genome-wide association studies remains a pressing challenge. Here, we construct a consensus variant-to-function (cV2F) score that assigns a single value to each common single-nucleotide variant in the genome, and helps to predict and characterize causal disease variants. The cV2F score leverages features reflecting variant-level experimentally and computationally predicted function (e.g. allelic imbalance and sequence-based deep learning models) and element-level function (e.g. predicted enhancers), and learns optimal combinations of features by training a gradient boosting model on GWAS fine-mapping results. The cV2F-annotated variants attained an AUPRC of 0.822 at identifying held-out fine-mapped variants. Variants with high cV2F scores are highly enriched for heritability (14.2x, s.e. 0.5) across 66 diseases/traits, are uniquely informative for disease heritability, and are highly predictive of variants implicated by reporter assays; cV2F substantially outperforms previous variant-to-function scores using all of these metrics. GWAS fine-mapping of 110 diseases/traits informed by cV2F identified 14.3% more confidently fine-mapped (PIP > 0.95) variants than non-functionally informed fine-mapping. We further constructed tissue/cell line-specific cV2F scores that prioritize variants based on regulatory potential in specific tissues/cell lines, attaining high heritability enrichment for tissue-related diseases/traits (15.6x, s.e. 2.3) while providing independent information (average correlation of 0.27 with the primary cV2F score). We highlight examples of GWAS loci for which cV2F pinpoints causal variants with high confidence and elucidates their functional role.

genetics↗

bioRxiv2024

NMI induces chemokine release and recruits neutrophils through the activation of NF-kappaB pathway

Neutrophils are essential components of the innate immune system, playing a pivotal role in immune responses. These cells rapidly migrate to sites of inflammation or tissue injury, facilitating pathogen clearance and tissue repair. The chemotactic signaling network regulating neutrophil recruitment is complex and not fully understood, particularly regarding damage-associated molecular patterns (DAMPs). In our previous research, we identified NMI as a DAMP that activates dendritic cells and macrophages, amplifying inflammatory responses and contributing to both acute and chronic inflammation. In this study, we investigated the role of NMI in neutrophil recruitment. We purified and characterized a recombinant murine NMI protein, ensuring endotoxin removal while preserving biological activity. In vivo experiments demonstrated that NMI enhances neutrophil recruitment in both a murine air pouch model and an acute peritonitis model, mediated by macrophage-derived chemokines. In vitro assays revealed a concentration-dependent increase in neutrophil migration induced by NMI, facilitated by chemokine secretion and subsequent migration through the CXCR2 receptor. Importantly, we established that NMI activates chemokine expression via the NF-{kappa}B signaling pathway. These findings provide insights into the mechanisms of NMI-induced neutrophil migration, enhancing our understanding of neutrophil recruitment during inflammation.

immunology↗

bioRxiv2024

Random Sanitization in DNA information storage using CRISPR-Cas12a

DNA information storage provides an excellent solution for metadata storage due to its high density, programmability, and long-term stability. However, current research in DNA storage primarily focuses on the processes of storing and reading data, lacking comprehensive solutions for the secure metadata wiping. Herein, we present a method of random sanitization in DNA information storage using CRISPR-Cas12a (RSDISC) based on precise control of the thermodynamic energy of primer-template hybridization. We utilize the collateral cleavage (trans-activity) of single-stranded DNA (ssDNA) by CRISPR-Cas12a to achieve selective sanitization of files in metadata. This method enables ssDNA degradation with different GC content, lengths, and secondary structures to achieve a sanitization efficiency up to 99.9% for 28,258 oligonucleotides in DNA storage within one round. We demonstrate that the number of erasable files could reach 1011.7 based on a model of primer-template hybridization efficiency. Overall, RSDISC provides a random sanitization approach to set the foundation of information encryption, file classification, memory deallocation and accurate reading in DNA data storage.

synthetic biology↗

bioRxiv2023

An encyclopedia of enhancer-gene regulatory interactions in the human genome

Identifying transcriptional enhancers and their target genes is essential for understanding gene regulation and the impact of human genetic variation on disease1-6. Here we create and evaluate a resource of >13 million enhancer-gene regulatory interactions across 352 cell types and tissues, by integrating predictive models, measurements of chromatin state and 3D contacts, and large-scale genetic perturbations generated by the ENCODE Consortium7. We first create a systematic benchmarking pipeline to compare predictive models, assembling a dataset of 10,411 element-gene pairs measured in CRISPR perturbation experiments, >30,000 fine-mapped eQTLs, and 569 fine-mapped GWAS variants linked to a likely causal gene. Using this framework, we develop a new predictive model, ENCODE-rE2G, that achieves state-of-the-art performance across multiple prediction tasks, demonstrating a strategy involving iterative perturbations and supervised machine learning to build increasingly accurate predictive models of enhancer regulation. Using the ENCODE-rE2G model, we build an encyclopedia of enhancer-gene regulatory interactions in the human genome, which reveals global properties of enhancer networks, identifies differences in the functions of genes that have more or less complex regulatory landscapes, and improves analyses to link noncoding variants to target genes and cell types for common, complex diseases. By interpreting the model, we find evidence that, beyond enhancer activity and 3D enhancer-promoter contacts, additional features guide enhancer-promoter communication including promoter class and enhancer-enhancer synergy. Altogether, these genome-wide maps of enhancer-gene regulatory interactions, benchmarking software, predictive models, and insights about enhancer function provide a valuable resource for future studies of gene regulation and human genetics.

genomics↗

bioRxiv2023

Single-cell transcriptomic and genomic changes in the aging human brain

Aging brings dysregulation of various processes across organs and tissues, often stemming from stochastic damage to individual cells over time. Here, we used a combination of single-nucleus RNA-sequencing and single-cell whole-genome sequencing to identify transcriptomic and genomic changes in the prefrontal cortex of the human brain across life span, from infancy to centenarian. We identified infant-specific cell clusters enriched for the expression of neurodevelopmental genes, and a common down-regulation of cell-essential homeostatic genes that function in ribosomes, transport, and metabolism during aging across cell types. Conversely, expression of neuron-specific genes generally remains stable throughout life. We observed a decrease in specific DNA repair genes in aging, including genes implicated in generating brain somatic mutations as indicated by mutation signature analysis. Furthermore, we detected gene-length-specific somatic mutation rates that shape the transcriptomic landscape of the aged human brain. These findings elucidate critical aspects of human brain aging, shedding light on transcriptomic and genomics dynamics.

genomics↗

bioRxiv2023

Long-Term In Vivo Molecular Monitoring Using Aptamer-Graphene Microtransistors

Long-term, real-time molecular monitoring in complex biological environments is critical for our ability to understand, prevent, diagnose, and manage human diseases. Aptamer-based electrochemical biosensors possess the promise due to their generalizability and a high degree of selectivity. Nevertheless, the operation of existing aptamer-based biosensors in vivo is limited to a few hours. Here, we report a first-generation long-term in vivo molecular monitoring platform, named aptamer-graphene microtransistors (AGMs). The AGM incorporates a layer of pyrene- (polyethylene glycol)5-alcohol and DNase inhibitor-doped polyacrylamide hydrogel coating to reduce biofouling and aptamer degradation. As a demonstration of function and generalizability, the AGM achieves the detection of biomolecules such as dopamine and serotonin in undiluted whole blood at 37 {degrees}C for 11 days. Furthermore, the AGM successfully captures optically evoked dopamine release in vivo in mice for over one week and demonstrates the capability to monitor behaviorally-induced endogenous dopamine release even after eight days of implantation in freely moving mice. The results reported in this work establish the potential for chronic aptamer-based molecular monitoring platforms, and thus serve as a new benchmark for molecular monitoring using aptamer-based technology.

bioengineering↗

bioRxiv2023

Aub, Vasa and Armi localization to phase separated nuage is dispensable for piRNA biogenesis and transposon silencing in Drosophila

From nematodes to placental mammals, key components of the germline transposon silencing piRNAs pathway localize to phase separated perinuclear granules. In Drosophila, the PIWI protein Aub, DEAD box protein Vasa and helicase Armi localize to nuage granules and are required for ping-pong piRNA amplification and phased piRNA processing. Drosophila piRNA mutants lead to genome instability and Chk2 kinase DNA damage signaling. By systematically analyzing piRNA pathway organization, small RNA production, and long RNA expression in single piRNA mutants and corresponding chk2/mnk double mutants, we show that Chk2 activation disrupts nuage localization of Aub and Vasa, and that the HP1 homolog Rhino, which drives piRNA precursor transcription, is required for Aub, Vasa, and Armi localization to nuage. However, these studies also show that ping-pong amplification and phased piRNA biogenesis are independent of nuage localization of Vasa, Aub and Armi. Dispersed cytoplasmic proteins thus appear to mediate these essential piRNA pathway functions.

developmental biology↗

bioRxiv2023

Rapid disassembly and Piwi-independent reassembly of Drosophila piRNA cluster heterochromatin following acute heat shock

Environmental stress activates transposons and is proposed to generate genetic diversity that facilitates adaptive evolution. piRNAs guide germline transposon silencing, but the impact of stress on the piRNA pathway is not well understood. In Drosophila, the Rhino-Deadlock-Cuff complex (RDC) drives transcription of clusters composed of nested transposon fragments, generating precursors that are processed into mature piRNAs in the cytoplasm. We show that acute heat shock triggers rapid, reversable, loss of RDC localization and cluster transcript expression with coordinate changes in the cytoplasmic processing machinery. Maternal piRNAs bound to Piwi are proposed to guide Rhino localization to clusters during early embryogenesis. However, RDC re-localization after heat shock is accelerated in piwi mutants and delayed in thoc7 mutants, which disrupt piRNA precursor binding to THO complex, and we show that maternally deposited piRNAs are dispensable for RDC localization to the major 42AB cluster. Cluster specification is reconsidered in light of these findings.

developmental biology↗