Biology subjects

Peng, H.

Publications and source records attributed to Peng, H..

135 records · Page 8Linked to original sources

bioRxiv2018

Exploring morphological motifs for a single neuron based on multiple 3D reconstructions

The morphology of individual neurons is useful to study structures and functions of nervous system. Researchers have invented many semi or fully automatic tracing methods to efficiently generate a reconstruction from a single neuron. Different tracing methods have different design principles, and could produce different reconstructions. However, the \"common substructures\" of various reconstructions, called morphological motifs, should be highly reliable. In this work, we propose a Vaa3D based framework to explore morphological motifs of 3D reconstructions from a single neuron. The framework contains four steps: (1) resampling and sorting each reconstruction according to a standard reconstruction, such as a gold standard reconstruction, consensus reconstruction or a certain reliable reconstruction; (2) applying local alignment algorithm for each pair of the standard reconstruction and a reconstruction, or each pair of two reconstructions; (3) constructing overlaps based on selected points in local alignment pairs; (4) obtaining morphological motifs by post-processing these overlaps. Under the proposed framework, three methods were implemented and tested on a dataset of 73 fruitfly neurons released by the BigNeuron project (http://bigneuron.org), which contains a gold standard reconstruction, a consensus reconstruction and about 40 automatic reconstructions for each neuron. We quantitatively evaluated these three methods to choose reliable morphological motifs.

bioinformatics↗

bioRxiv2018

DeepNeuron: An Open Deep Learning Toolbox for Neuron Tracing

Reconstructing three-dimensional (3D) morphology of neurons is essential to understanding brain structures and functions. Over the past decades, a number of neuron tracing tools including manual, semi-automatic, and fully automatic approaches have been developed to extract and analyze 3D neuronal structures. Nevertheless, most of them were developed based on coding certain rules to extract and connect structural components of a neuron, showing limited performance on complicated neuron morphology. Recently, deep learning outperforms many other machine learning methods in a wide range of image analysis and computer vision tasks. Here we developed a new open source toolbox, DeepNeuron, which uses deep learning networks to learn features and rules from data and trace neuron morphology in light microscopy images. DeepNeuron provides a family of modules to solve basic yet challenging problems in neuron tracing. These problems include but not limited to: (1) detecting neuron signal under different image conditions, (2) connecting neuronal signals into tree(s), (3) pruning and refining tree morphology, (4) quantifying the quality of morphology, and (5) classifying dendrites and axons in real time. We have tested DeepNeuron using light microscopy images including bright-field and confocal images of human and mouse brain, on which DeepNeuron demonstrates robustness and accuracy in neuron tracing.

bioinformatics↗

bioRxiv2018

Long Noncoding RNA-Maternally Expressed Gene 3 Contributes to Hypoxic Pulmonary Hypertension

The expression and function of long noncoding RNAs (lncRNAs) in the development of hypoxic pulmonary hypertension, especially in the proliferation of pulmonary artery smooth muscle cells (PASMCs) are largely unknown. Here, we characterized the expression of lncRNA-maternally expressed gene 3 (lncRNA-MEG3) was significantly increased and primarily located in the cytoplasm of PASMCs by hypoxia. LncRNA-MEG3 knockdown by lung-specific delivery of small interfering RNAs (siRNAs) significantly prevented the development of hypoxic pulmonary hypertension in vivo. Silencing of lncRNA-MEG3 by siRNAs and gapmers attenuated PASMC responses to hypoxia in vitro. Mechanically, we found that lncRNA-MEG3 acts as a molecular sponge of microRNA-328 (miR-328); upon hypoxia, lncRNA-MEG3 interacts and sequesters miR-328, leading to the upregulation of insulin-like growth factor 1 receptor (IGF1R). Additionally, higher expression of lncRNA-MEG3 and IGF1R, and lower expression of miR-328 were observed in PASMCs of iPAH patients. These data provide insight into the contribution of lncRNA-MEG3 in hypoxia pulmonary hypertension. Upregulation of lncRNA-MEG3 sequesters cytoplasmic miR-328, eventually leading to the expression of IGF1R, revealing a regulatory mechanism by lncRNAs in hypoxia-induced PASMC proliferation.

molecular biology↗

bioRxiv2017

Potent and Selective Antitumor Activity of a T-Cell Engaging Bispecific Antibody Targeting a Membrane-Proximal Epitope of ROR1

T-cell engaging bispecific antibodies present a promising strategy for cancer immunotherapy and numerous bispecific formats have been developed for retargeting cytolytic T cells toward tumor cells. To explore the therapeutic utility of T-cell engaging bispecific antibodies targeting the receptor tyrosine kinase ROR1, which is expressed by tumor cells of various hematologic and solid malignancies, we used a bispecific ROR1 x CD3 scFv-Fc format based on a heterodimeric and aglycosylated Fc domain designed for extended circulatory half-life and diminished systemic T-cell activation. A diverse panel of ROR1-targeting scFv derived from immune and naive rabbit antibody repertoires was compared in this bispecific format for target-dependent T-cell recruitment and activation. A ROR1-targeting scFv with a membrane-proximal epitope, R11, revealed potent and selective antitumor activity in vitro and in vivo and emerged as a prime candidate for further preclinical and clinical studies. To elucidate the precise location and engagement of this membrane-proximal epitope, which is conserved between human and mouse ROR1, the three-dimensional structure of scFv R11 in complex with the kringle domain of ROR1 was determined by X-ray crystallography at 1.6-[A] resolution.

immunology↗

bioRxiv2017

SmartScope2: Simultaneous Imaging and Reconstruction of Neuronal Morphology

Quantitative analysis of neuronal morphology is critical in cell type classification and for deciphering how structure gives rise to function in the brain. Most current approaches to imaging and tracing neuronal 3D morphology are data intensive. We introduce SmartScope2, the first open source, automated neuron reconstruction machine integrating online image analysis with automated multiphoton imaging. SmartScope2 takes advantage of a neurons sparse morphology to improve imaging speed and reduce image data stored, transferred and analyzed. We show that SmartScope2 is able to produce the complex 3D morphology of human and mouse cortical neurons with six-fold reduction in image data requirements and three times the imaging speed compared to conventional methods.

bioinformatics↗

bioRxiv2017

Automated 3D Neuron Tracing with Precise Branch Erasing and Confidence Controlled Back-Tracking

The automatic reconstruction of single neuron cells from microscopic images is essential to enabling large-scale data-driven investigations in neuron morphology research. However, few previous methods were able to generate satisfactory results automatically from 3D microscopic images without human intervention. In this study, we developed a new algorithm for automatic 3D neuron reconstruction. The main idea of the proposed algorithm is to iteratively track backwards from the potential neuronal termini to the soma centre. An online confidence score is computed to decide if a tracing iteration should be stopped and discarded from the final reconstruction. The performance improvements comparing to the previous methods are mainly introduced by a more accurate estimation of the traced area and the confidence controlled back-tracking algorithm. The proposed algorithm supports large-scale batch-processing by requiring only one hyper-parameter for background segmentation. We bench-tested the proposed algorithm on the images obtained from both the DIADEM challenge and the BigNeuron challenge. Our proposed algorithm achieved the state-of-the-art results.

neuroscience↗

bioRxiv2016

Fast Assembling of Neuron Fragments in Serial 3D Sections

AbstractReconstructing neurons from 3D image-stacks of serial sections of thick brain tissue is very time-consuming and often becomes a bottleneck in high-throughput brain mapping projects. We developed NeuronStitcher, a software suite for stitching non-overlapping neuron fragments reconstructed in serial 3D image sections. With its efficient algorithm and user-friendly interface, NeuronStitcher has been used successfully to reconstruct very large and complex human and mouse neurons.

neuroscience↗

bioRxiv2016

Premature termination codons signaled targeted repair of frameshift mutation by endogenous RNA-directed gene editing

The molecular mechanisms for repairing DNA damages and point mutations have been well understood but it remains unclear how a frameshift mutation is repaired. Here we report that frameshift reversion occurs in E. coli more frequently than expected and appears to be a targeted gene repair signaled by premature termination codons (PTCs), producing high-level variations in the repaired genes. Genome resequencing shows that the revertant genome is highly stable, and the single-molecule variations in the repaired genes are derived from RNA editing. A multi-omics analysis shows that the expression levels change greatly in most the DNA and RNA manipulating genes. DNA replication, transcription, RNA editing, RNA degradation, nucleotide excision repair, mismatch repair, and homologous recombination were upregulated in the frameshift or revertant, but the base excision repair was not. Moreover, genes and transposons in a duplicate region silenced in wild type E. coli were activated in the frameshift. Finally, we propose a nonsense-mediated gene revising (NMGR) model for frame repair, which also acts as a driving force for molecular evolution. In essence, nonsense mRNAs are recognized, edited, and transported to template the repair of the coding gene by RNA-directed DNA repair, nucleotide excision, mismatch repair, and homologous recombination. Thanks to NMGR, the mutation rate temporarily rises in a frameshift gene, bringing genetic diversity while repairing the frameshift mutation and accelerating the evolution process without a high mutation rate in the genome.

genetics↗

bioRxiv2016

Automatic Tracing of Ultra-Volume of Neuronal Images

Despite substantial advancement in the automatic tracing of neurons' morphology in recent years, it is challenging to apply the existing algorithms to very large image datasets containing billions or more voxels. We introduce UltraTracer, a solution designed to extend any base neuron-tracing algorithm to be able to trace virtually unlimited data volumes. We applied this approach to neuron-tracing algorithms with completely different design principles and tested on challenging human and mouse neuron datasets that have hundreds of billions of voxels. Results indicate that UltraTracer is scalable, accurate, and about 3 to 6 times more efficient compared to other state-of-the-art approaches.

bioinformatics↗