Biology subjects

Nie, Z.

Publications and source records attributed to Nie, Z..

28 records · Page 2Linked to original sources

bioRxiv2024

Multi-Scale Protein Language Model for Unified Molecular Modeling

Protein language models have demonstrated significant potential in the field of protein engineering. However, current protein language models primarily operate at the residue scale, which limits their ability to provide information at the atom level. This limitation prevents us from fully exploiting the capabilities of protein language models for applications involving both proteins and small molecules. In this paper, we propose ESM-AA (ESM All-Atom), a novel approach that enables atom-scale and residue-scale unified molecular modeling. ESM-AA achieves this by pretraining on multi-scale code-switch protein sequences and utilizing a multi-scale position encoding to capture relationships among residues and atoms. Experimental results indicate that ESM-AA surpasses previous methods in proteinmolecule tasks, demonstrating the full utilization of protein language models. Further investigations reveal that through unified molecular modeling, ESM-AA not only gains molecular knowledge but also retains its understanding of proteins.1

bioinformatics↗

bioRxiv2024

Hunting for peptide binders of specific targets with data-centric generative language models

Bioactive peptides have become strong candidates for a variety of clinical therapies due to their diverse advantages, which promotes the development of deep generative models for peptide design. Considering that existing methods cannot effectively deal with the conformational flexibility of peptides and are difficult to capture accurate residue-to-residue interaction dependencies, we propose a unified weakly order-dependent autoregressive language modeling architecture (PepGenWOA) for bioactive peptide generative design with tolerating out-of-order input as inductive bias. The superiority of PepGenWOA is demonstrated by generating three classes of therapeutic peptides, including antimicrobial peptides, anticancer peptides, and peptide binders. For antimicrobial and anticancer peptide generation, PepGenWOA not only comprehensively outperforms state-of-the-art baseline models, but also exhibits a significant propensity to incorporate specific types of residues that are beneficial for antimicrobial or anticancer bioactivity. For characteristic-guided peptide binder generation, the pretrained PepGenWOA is fine-tuned with Mixture-of-Experts-style plugins with lifelong learning paradigm, thereby achieving the best trade-off between memory stability for world knowledge and learning plasticity for new downstream tasks. The subsequent structure-based virtual screening anchors the 7 most promising candidates from the mega-scale synthetic sequence space, reducing the number of candidates for in vitro experimental validation by 5 orders of magnitude. A target binding rate of 28.6% with binding specificity further confirms the effectiveness of the fine-tuning strategy. Overall, PepGenWOA is a unified architecture for peptide generative design that can be flexibly customized for different task requirements, thereby harnessing generative language models to reach the "dark space" of biological language.

bioinformatics↗

bioRxiv2023

Evolution-guided large language model is a predictor of virus mutation trends

The increasing frequency of emerging viral infections necessitates a rapid human response, highlighting the cost-effectiveness of computational methods. However, existing computational approaches are limited by their input forms or incomplete functionalities, preventing a unified prediction of diverse viral variation drivers and hindering in-depth applications. To address this issue, we propose a unified evolution-driven framework for predicting virus variation drivers, named E2VD, which is guided by virus evolutionary traits priors. With evolution-inspired design, E2VD comprehensively and significantly outperforms state-of-the-art methods across various virus variation drivers prediction tasks. Moreover, E2VD effectively captures the fundamental patterns of virus evolution. It not only distinguishes different types of mutations but also accurately identifies rare beneficial mutations that are critical for virus to survival, while maintains generalization capabilities on different viral lineages. Importantly, with predicted biological drivers, E2VD perceives virus evolutionary trends, in which potential high-risk mutation sites are accurately recommended. Overall, E2VD represents a unified, structure-free, and interpretable approach for analyzing and predicting viral evolutionary fitness, providing an ideal alternative to costly wet-lab measurements to accelerate responses to emerging viral infections.

bioinformatics↗

bioRxiv2023

Millimeter-scale magnetic implants paired with a fully integrated wearable device for wireless biophysical and biochemical sensing

Implantable sensors can directly interface with various organs for precise evaluation of health status. However, extracting signals from such sensors must rely on transcutaneous wires, integrated circuit chips, or cumbersome readout equipment, which increases the risks of infection, reduces the biocompatibility, or limits the portability. Here, we develop a set of millimeter-scale, chip-less and battery-less magnetic implants that can measure biophysical and biochemical signals wirelessly. In particular, the implants form two-way communications with a fully integrated wearable device, where the wearable device can induce a large-amplitude damped vibration of the magnetic implants and capture their subsequent motions in a wireless manner. Such damped vibrations reflect not only the biophysical conditions surrounding the implants movements, but also the concentration of a specific biochemical depending on the surface modification. Experiments in rat models demonstrate the capabilities in measuring cerebrospinal fluid (CSF) viscosity, intracranial pressure (ICP), and CSF glucose levels. This miniaturized system opens possibility for continuous, wireless monitoring of a wide range of biophysical and biochemical conditions within the living organism.

bioengineering↗

bioRxiv2023

Perceptual error based on Bayesian cue combination drives implicit motor adaptation

The sensorimotor system can recalibrate itself without our conscious awareness, a type of procedural learning whose computational mechanism remains undefined. Recent findings on implicit motor adaptation, such as over-learning from small perturbations and fast saturation for increasing perturbation size, challenge existing theories based on sensory errors. We argue that perceptual error, arising from the optimal combination of movement-related cues, is the primary driver of implicit adaptation. Central to our theory is the increasing sensory uncertainty of visual cues with increasing perturbations, which was validated through perceptual psychophysics (Experiment 1). Our theory predicts the learning dynamics of implicit adaptation across a spectrum of perturbation sizes on a trial-by-trial basis (Experiment 2). It explains proprioception changes and their relation to visual perturbation (Experiment 3). By modulating visual uncertainty in perturbation, we induced unique adaptation responses in line with our model predictions (Experiment 4). Overall, our perceptual error framework outperforms existing models based on sensory errors, suggesting that perceptual error in locating ones effector, supported by Bayesian cue integration, underpins the sensorimotor systems implicit adaptation.

neuroscience↗

bioRxiv2023

Emerging cooperativity between Oct4 and Sox2 governs the pluripotency network in mouse embryos

During the first lineage segregation, mammalian embryos generate the inner cell mass (ICM) and trophectoderm (TE). ICM gives rise to the epiblast (EPI) that forms all cell types of the body, an ability referred to as pluripotency. The molecular mechanisms that induce pluripotency in embryos remain incompletely elucidated. Using knockout (KO) mouse models in conjunction with low-input ATAC-seq and RNA-seq, we found that Oct4 and Sox2 gradually come into play in the early ICM, coinciding with the initiation of Sox2 expression. Oct4 and Sox2 activate the pluripotency-related genes through the putative OCT-SOX enhancers in the early ICM. Furthermore, we observed a substantial reorganization of chromatin landscape and transcriptome from the morula to the early ICM stages, which was partially driven by Oct4 and Sox2, highlighting their pivotal role in promoting the developmental trajectory towards the ICM. Our study provides new insights into the establishment of the pluripotency network in mouse preimplantation embryos.

developmental biology↗

bioRxiv2023

Target-identification and mechanism-of-action studies of indole terpenoid mimics reveal that proper spindle microtubule assembly is essential for Cdh1-mediated proteolysis of CENP-A

Centromere protein A (CENP-A), a histone H3 variant specific to centromeres, is crucial for kinetochore positioning and chromosome segregation. However, its regulatory mechanism in human cells remains incompletely understood. We conducted a structure-activity relationship (SAR) study of the cell cycle-arresting indole terpenoid mimic JP18 and found two more potent analogues, (+)-6-Br-JP18 and (+)-6-Cl-JP18. Tubulin was identified as a potential cellular target of these halogenated analogues by using the drug affinity responsive target stability (DARTS) based method. X-ray crystallography analysis revealed that both molecules bind to the colchicine-binding site of {beta}-tubulin. Furthermore, we discovered that treatment of human cells with microtubule-targeting agents (MTAs), including these two compounds, led to CENP-A accumulation by destabilizing Cdh1, a co-activator of the APC/C E3 ubiquitin ligase. Our study establishes a link between microtubule dynamics and CENP-A accumulation using small-molecule tools and highlights the role of Cdh1 in CENP-A proteolysis.

cell biology↗

bioRxiv2022

Running ahead of evolution - AI based simulation for predicting future high-risk SARS-CoV-2 variants

The never-ending emergence of SARS-CoV-2 variations of concern (VOCs) has challenged the whole world for pandemic control. In order to develop effective drugs and vaccines, one needs to efficiently simulate SARS-CoV-2 spike receptor binding domain (RBD) mutations and identify high-risk variants. We pretrain a large protein language model with approximately 408 million protein sequences and construct a high-throughput screening for the prediction of binding affinity and antibody escape. As the first work on SARS-CoV-2 RBD mutation simulation, we successfully identify mutations in the RBD regions of 5 VOCs and can screen millions of potential variants in seconds. Our workflow scales to 4096 NPUs with 96.5% scalability and 493.9x speedup in mixed precision computing, while achieving a peak performance of 366.8 PFLOPS (reaching 34.9% theoretical peak) on Pengcheng Cloudbrain-II. Our method paves the way for simulating coronavirus evolution in order to prepare for a future pandemic that will inevitably take place. Our models are released at https://github.com/ZhiweiNiepku/SARS-CoV-2_mutation_simulation to facilitate future related work. JustificationWe develop a novel multi-constraint variation prediction framework to simulate SARS-CoV-2 RBD mutations, reaching a peak performance of 366.8 PFLOPS with 96.5% scalability and achieving 493.9x speedup. Our method facilitates the prediction and prioritization of future high-risk variants for the early deployment of drugs and vaccines. Performance attributes O_TBL View this table: org.highwire.dtl.DTLVardef@1b1873aorg.highwire.dtl.DTLVardef@25314eorg.highwire.dtl.DTLVardef@1c80130org.highwire.dtl.DTLVardef@1c61b21org.highwire.dtl.DTLVardef@11f21c5_HPS_FORMAT_FIGEXP M_TBL C_TBL Overview of the problemCoronavirus Disease 2019 (COVID-19) has spread rapidly to more than 200 countries or regions since December 2019. Due to its high infectivity, there have been over 645 million confirmed cases, including approximately 6.6 million deaths, reported by the World Health Organization (WHO) as of December 20221. In addition to being a serious threat to human health, COVID-19 has had a catastrophic impact on the global economy.

bioinformatics↗

bioRxiv2022

Dual control of lysogeny and phage defense by a phosphorylation-based toxin/antitoxin system

Regulatory systems that maintain prophage quiescence integrate phage and host gene expression with environmental conditions1,2. In the opportunistic bacterial pathogen Pseudomonas aeruginosa, Pf filamentous bacteriophages play critical roles in biofilm formation and virulence3-5, but mechanisms governing Pf prophage activation in biofilms are largely unknown. Here, we report a new type of prophage regulatory module in a widely-distributed P. aeruginosa lineage that not only controls virion production of co-resident Pf prophages, but also mediates defense against diverse lytic phages. By comparing two lineages of the prototype P. aeruginosa strain PAO1 that harbor different Pf prophages, we identified a prophage-encoded kinase-kinase-phosphatase (KKP) system that controls Pf production in biofilms. KKP components exhibit dynamic stoichiometry, where high kinase levels in planktonic conditions maintain phosphorylation of the host H-NS protein MvaU, repressing prophage activation. During biofilm formation, phosphatase expression is heightened, leading to MvaU dephosphorylation and alleviating repression of prophage gene expression. KKP clusters are present in hundreds of diverse temperate prophages and other mobile elements across Gram-negative bacteria. Characterization of KKP modules from different species revealed that, in addition to regulating Pf phage lysogeny, KKP functions as a tripartite toxin-antitoxin system that mediates host defense from predatory lytic phages. KKP represents a new phosphorylation-based mechanism for prophage regulation and for phage defense. The dual function of this module raises the question of whether other newly described phage defense systems6-9 also regulate intrinsic prophage biology in diverse hosts.

microbiology↗

bioRxiv2020

Psi promotes Drosophila wing growth through transcriptional repression of key developmental networks

Psi, the sole FUSE Binding Protein (FUBP) family single stranded DNA/RNA binding protein in Drosophila, is essential for proper cell and tissue growth, however its mechanism of function remains unclear. Here we use Targeted DamID combined with RNA-sequencing to generate the first genome-wide binding and expression profiles for Psi. Surprisingly, we demonstrate Psi drives growth in the Drosophila wing through transcriptional repression of key developmental pathways (e.g. Wnt, Notch and TGF{beta}). Thus, Psi patterns tissue growth by directly repressing transcription of developmental growth suppressors. Analysis of direct Psi targets identified novel growth inhibitors, including Tolkin (Zinc metallopeptidase implicated in TGF{beta} signalling), Ephexin (Rho-GEF) and emp (CD36 scavenger receptor-related protein). Their depletion not only suppressed impaired growth associated with Psi knockdown, but alone was sufficient to drive wing overgrowth. Thus, Psi drives wing growth twofold, through direct activation of Myc and through transcriptional repression of growth inhibitors comprising core developmental pathways.

developmental biology↗