Biology subjects

Deng, M.

Publications and source records attributed to Deng, M..

50 records · Page 3Linked to original sources

bioRxiv2019

Rapidly cycling stem cells regenerate the intestine independent of Lgr5high cells

The +4 cells in intestinal crypts are DNA damage-resistant and contribute to regeneration. However, their exact identity and the mechanism underlying +4 cell-mediated regeneration remain unclear. Using lineage tracing, we show that cells marked by an Msi1 reporter (Msi1+) are enriched at the +4 position in intestinal crypts and exhibit DNA damage resistance. Single-cell RNA sequencing reveals that the Msi1+ cells are heterogeneous with the majority being intestinal stem cells (ISCs). The DNA damage-resistant subpopulation of Msi1+ cells is characterized by low-to-negative Lgr5 expression and is more rapidly cycling than Lgr5high radio-sensitive crypt base columnar stem cells (CBCs); they enable fast repopulation of the intestinal epithelium independent of CBCs that are largely depleted after irradiation. Furthermore, relative to CBCs, Msi1+ cells preferentially produce Paneth cells during homeostasis and upon radiation repair. Together, we demonstrate that the DNA damage-resistant Msi1+ cells are rapidly cycling ISCs that maintain and regenerate the intestinal epithelium.

cell biology↗

bioRxiv2019

Deepprune: Learning efficient and interpretable convolutional networks through weight pruning for predicting DNA-protein binding

Convolutional neural network (CNN) based methods have outperformed conventional machine learning methods in predicting the binding preference of DNA-protein binding. Although studies in the past have shown that more convolutional kernels help to achieve better performance, visualization of the model can be obscured by the use of many kernels, resulting in overfitting and reduced interpretation because the number of motifs in true models is limited. Therefore, we aim to arrive at high performance, but with limited kernel numbers, in CNN-based models for motif inference.\n\nWe herein present Deepprune, a novel deep learning framework, which prunes the weights in the dense layer and fine-tunes iteratively. These two steps enable the training of CNN-based models with limited kernel numbers, allowing easy interpretation of the learned model. We demonstrate that Deepprune significantly improves motif inference performance for the simulated datasets. Furthermore, we show that Deepprune outperforms the baseline with limited kernel numbers when inferring DNA-binding sites from ChIP-seq data.

bioinformatics↗

bioRxiv2019

Stromal ISLR promotes intestinal regeneration and cancer by suppressing epithelial Hippo signaling

The YAP signaling activation in epithelial cells is essential for intestinal regeneration and tumorigenesis. However, the molecular mechanism linking stromal signals to YAP-mediated intestinal regeneration and tumorigenesis is poorly defined. Here we report a stroma-epithelia YAP signaling axis essential for stromal cells to modulate epithelial cell growth during intestinal regeneration and tumorigenesis. Specifically, upon inflammation and in cancer, an oncogenic transcription factor ETS1 in stromal cells induces expression of a secreted protein ISLR that can directly binds to a transmembrane protocadherin FAT1 on the surface of epithelial cells. This binding suppressed the Hippo signaling by disrupting MST1-FAT1 association, resulting in YAP signaling activation. Deletion of Islr in stromal cells in mice markedly impaired intestinal regeneration, and suppressed tumorigenesis in the colon. Moreover, the expression of stromal cell-specific ISLR and ETS1 significantly increased in inflamed mucosa of human IBD patients and in human colorectal adenocarcinoma, accounting for the epithelial YAP hyperactivation. Collectively, our findings uncovered a molecular mechanism governing signals for communication between stroma and epithelium during tissue regeneration and tumorigenesis.

cancer biology↗

bioRxiv2019

Expectation pooling: An effective and interpretable pooling method for predicting DNA-protein binding

MotivationConvolutional neural networks (CNNs) have outperformed conventional methods in modeling the sequence specificity of DNA-protein binding. While previous studies have built a connection between CNNs and probabilistic models, simple models of CNNs cannot achieve sufficient accuracy on this problem. Recently, some methods of neural networks have increased performance using complex neural networks whose results cannot be directly interpreted. However, it is difficult to combine probabilistic models and CNNs effectively to improve DNA-protein binding predictions.\n\nResultsIn this paper, we present a novel global pooling method: expectation pooling for predicting DNA-protein binding. Our pooling method stems naturally from the EM algorithm, and its benefits can be interpreted both statistically and via deep learning theory. Through experiments, we demonstrate that our pooling method improves the prediction performance DNA-protein binding. Our interpretable pooling method combines probabilistic ideas with global pooling by taking the expectations of inputs without increasing the number of parameters. We also analyze the hyperparameters in our method and propose optional structures to help fit different datasets. We explore how to effectively utilize these novel pooling methods and show that combining statistical methods with deep learning is highly beneficial, which is promising and meaningful for future studies in this field.\n\nContactdengmh@pku.edu.cn, gaog@mail.cbi.pku.edu.cn\n\nSupplementary informationAll code is public in https://github.com/gao-lab/ePooling

bioinformatics↗

bioRxiv2019

Genomic landscape of the global oak phylogeny

O_LIThe tree of life is highly reticulate, with the history of population divergence buried amongst phylogenies deriving from introgression and lineage sorting. In this study, we test the hypothesis that there are regions of the oak (Quercus, Fagaceae) genome that are broadly informative about phylogeny and investigate global patterns of oak diversity.\nC_LIO_LIWe utilize fossil data and restriction-site associated DNA sequencing (RAD-seq) for 632 individuals representing ca. 250 oak species to infer a time-calibrated phylogeny of the worlds oaks. We use reversible-jump MCMC to reconstruct shifts in lineage diversification rates, accounting for among-clade sampling biases. We then map the > 20,000 RAD-seq loci back to a recently published oak genome and investigate genomic distribution of introgression and phylogenetic support across the phylogeny.\nC_LIO_LIOak lineages have diversified among geographic regions, followed by ecological divergence within regions, in the Americas and Eurasia. Roughly 60% of oak diversity traces back to four clades that experienced increases in net diversification due to climatic transitions or ecological opportunity.\nC_LIO_LIThe support we find for the phylogeny contrasts with high genomic heterogeneity in phylogenetic signal and introgression. Oaks are phylogenomic mosaics, and their diversity may in fact depend on the gene flow that shapes the oak genome.\nC_LI

evolutionary biology↗

bioRxiv2018

A Framework to Incorporate D-trace Loss into Compositional Data Analysis

The development of high-throughput sequencing technologies for 16S rRNA gene profiling provides higher quality compositional data for microbe communities. Inferring the direct interaction network under a specific condition and understanding how the network structure changes between two different environmental or genetic conditions are two important topics in biological studies. However, the compositional nature and high dimensionality of the data are challenging in the context of network and differential network recovery. To address this problem in the present paper, we proposed a framework to incorporate the data transformations developed for compositional data analysis into D-trace loss for network and differential network estimation, respectively. The sparse matrix estimators are defined as the minimizer of the corresponding lasso penalized loss. This framework is characterized by its straightforward application based on the ADMM algorithm for numerical solution. Simulations show that the proposed method outperforms other state-of-the-art methods in network and differential network inference under different scenarios. Finally, as an illustration, our method is applied to a mouse skin microbiome data.\n\nAuthor summaryInferring the direct interactions among microbes and how these interactions change under different conditions are important to understand community-wide dynamics. The compositional nature and high dimensionality are two distinctive features of microbial data, which invalidate traditional correlation analysis and challenge interaction network estimation. In this study, we set up a framework that combines data transformation with D-trace loss to infer the direct interaction network and differential network from compositional data. Simulations and real data analysis show that our proposed methods lead to results with higher accuracy and stability.

microbiology↗

bioRxiv2018

scRMD: Imputation for single cell RNA-seq data via robust matrix decomposition

MotivationSingle cell RNA-sequencing (scRNA-seq) technology enables whole transcriptome profiling at single cell resolution and holds great promises in many biological and medical applications. Nevertheless, scRNA-seq often fails to capture expressed genes, leading to the prominent dropout problem. These dropouts cause many problems in down-stream analysis, such as significant noise increase, power loss in differential expression analysis and obscuring of gene-to-gene or cell-to-cell relationship. Imputation of these dropout values thus becomes an essential step in scRNA-seq data analysis.\n\nResultsIn this paper, we model the dropout imputation problem as robust matrix decomposition. This model has minimal assumptions and allows us to develop a computational efficient imputation method scRMD. Extensive data analysis shows that scRMD can accurately recover the dropout values and help to improve downstream analysis such as differential expression analysis and clustering analysis.\n\nContactruibinxi@math.pku.edu.cn

bioinformatics↗

bioRxiv2018

Drosophila scribble mutant tumors undergo a transition from a growth arrest state to a proliferative state over time

The Drosophila neoplastic tumor suppressor gene (nTSG) mutant tumors have successfully modeled many aspects of human tumor progression. However, the fly nTSG mutant tumors progress rapidly over days. This is in contrast with most human tumors which develop slowly, harbor heterogeneous cell populations for selection and undergo an evolution-like process. Whether the fast-growing fly nTSG mutant tumors have capacity for evolution remains unclear. Through quantitative analysis of the scrib mutant tumor growth, we found that the scrib mutant tumors evolve to display different growth rates and cell cycle profiles over time. Multiple growth-regulatory signaling pathways show quantitative differences in early versus late scrib mutant tumors. These data suggest that the scrib mutant tumors undergo a transition from a growth arrest state to a proliferative state. Through longitudinal single cell RNA (scRNA) data analysis we found that the scrib mutant tumors harbor heterogeneous cell populations likely of distinct proliferative states, which are available for potential selection. This study raises the possibility of studying tumor evolution in a genetically accessible and fast-growing invertebrate tumor model.

developmental biology↗

bioRxiv2017

Real-value and confidence prediction of protein backbone dihedral angles through a hybrid method of clustering and deep learning

BackgroundProtein dihedral angles provide a detailed description of protein local conformation. Predicted dihedral angles can be used to narrow down the conformational space of the whole polypeptide chain significantly, thus aiding protein tertiary structure prediction. However, direct angle prediction from sequence alone is challenging.\n\nMethodIn this study, we present a novel method to predict realvalued angles by combining clustering and deep learning. That is, we first generate certain clusters of angles (each assigned a label) and then apply a deep residual neural network to predict the label posterior probability. Finally, we output real-valued prediction by a mixture of the clusters with their predicted probabilities. At the same time, we also estimate the bound of the prediction errors at each residue from the predicted label probabilities.\n\nResultIn this article, we present a novel method (named RaptorX-Angle) to predict real-valued angles by combining clustering and deep learning. Tested on a subset of PDB25 and the targets in the latest two Critical Assessment of protein Structure Prediction (CASP), our method outperforms the existing state-of-art method SPIDER2 in terms of Pearson Correlation Coefficient (PCC) and Mean Absolute Error (MAE). Our result also shows approximately linear relationship between the real prediction errors and our estimated bounds. That is, the real prediction error can be well approximated by our estimated bounds.\n\nConclusionsOur study provides an alternative and more accurate prediction of dihedral angles, which may facilitate protein structure prediction and functional study.

bioinformatics↗

bioRxiv2017

ZmCOL3, a CCT-domain containing gene affects maize adaptation as a repressor and upstream of ZmCCT

Flowering time is a vital trait to control the adaptation of flowering plants to different environments. CCT-domain containing genes are considered to play an important role in plants flowering. Among 53 maize CCT family genes, 28 of them were located in the flowering time QTL regions and 16 genes were significant associated with flowering time based on candidate gene-based association mapping analysis. Furthermore, a CCT gene named as ZmCOL3 was validated to be a flowering repressor upstream of ZmCCT which is one of the key genes regulating maize flowering. The overexpressed ZmCOL3 could delay flowering time about 4 days whether in long day or short day conditions. The absent of one cytosine in 3UTR and the present of 551bp fragment in promoter regions are likely the causal polymorphisms which may contribute to the maize adaptation from tropical to temperate regions. ZmCOL3 could transactivate ZmCCT transcription or interfere circadian clock to inhibit flowering which was integrated in the modified model of maize photoperiod pathway.\n\nHighlightMaize CCT genes influence flowering time in different latitude environments and one of them named ZmCOL3 is a flowering time repressor which could transactivate ZmCCT transcription to delay flowering.

genetics↗

bioRxiv2017

Liver physiological T1rho dynamics associated with age and gender

PurposeUsing a single breathhold black blood sequence, the current study aims to understand the physiological ranges of liver T1rho relaxation for women and men.\n\nMaterials and MethodsThis volunteer study was conducted with institutional ethics committee approval, and included 62 females (age mean: 38.9 years; range: 18-75 years) and 34 males (age mean: 44.7 years, range: 24-80 years). MRI was conducted with a 3.0 T scanner, with six spin-lock times of 0, 10, 20, 25, 35, 50msec and a single breathhold of 12 seconds. Six slices were acquired for each examination.\n\nResultsFemale liver T1rho value ranged between 35.07 to 51.97ms, showed an age-dependent decrease with younger women had a higher measurement. Male Liver T1rho values ranged between 34.94 to 43.39 ms, and there was no evidential age-dependence. For females, there was a trend that liver T1rho value could be 4%-5% lower during menstrual phase than nonmenstrual phase. For both females and males, no evidential association was seen between body mass index and liver T1rho.\n\nConclusionLiver T1rho physiological value for males have relatively narrow distribution, while physiological value for females have wider distribution, and decreases with age.\n\nKey points1. Liver T1rho shows an age-dependency in women, with young women showing higher measurement. This age-dependency of liver T1rho measurement is not evidential in men. Post-menopausal women have similar liver T1rho value as men.\n\n2. Women at menstrual phase may have slight lower liver T1rho measurement.\n\n3. No association was noted between body mass index and liver T1rho\n\n4. When blood signal suppression sequence is used, in a population of 62 healthy women and 34 healthy men, the highest measured liver T1rho was 52 msec for young women, 44.7 msec for post-menopausal women, and 43.4 msec for men.

biophysics↗

bioRxiv2017

An updated infrageneric classification of the oaks: review of previous taxonomic schemes and synthesis of evolutionary patterns

In this paper, we review major classification schemes proposed for oaks by John Claudius Loudon, Anders Sandoe Orsted, William Trelease, Otto Karl Anton Schwarz, Aimee Antoinette Camus, Yuri Leonardovich Menitsky, and Kevin C. Nixon. Classifications of oaks (Fig. 1) have thus far been based entirely on morphological characters. They differed profoundly from each other because each taxonomist gave a different weight to distinguishing characters; often characters that are homoplastic in oaks. With the advent of molecular phylogenetics our view has considerably changed. One of the most profound changes has been the realisation that the traditional split between the East Asian subtropical to tropical subgenus Cyclobalanopsis and the subgenus Quercus that includes all other oaks is artificial. The traditional concept has been replaced by that of two major clades, each comprising three infrageneric groups: a Palearctic-Indomalayan clade including Group Ilex (Ilex oaks), Group Cerris (Cerris oaks) and Group Cyclobalanopsis (cycle-cup oaks), and a predominantly Nearctic clade including Group Protobalanus (intermediate or golden cup oaks), Group Lobatae (red oaks) and Group Quercus (white oaks, with most species in America and some 30 species in Eurasia). The main morphological feature characterising these phylogenetic lineages is pollen morphology, a character overlooked in traditional classifications. This realisation, along with the now available (molecular-)phylogenetic framework, opens new avenues for biogeographic, ecological and evolutionary studies and a re-appraisal of the fossil record. We provide an overview about recent advances in these fields and outline how the results of these studies contribute to the establishment of a unifying systematic scheme of oaks. Ultimately, we propose an updated classification of Quercus recognising two subgenera with eight sections. This classification considers morphological traits, molecular-phylogenetic relationships, and the evolutionary history of one of the most important temperate woody plant genera.\n\nO_FIG O_LINKSMALLFIG WIDTH=156 HEIGHT=200 SRC=\"FIGDIR/small/168146_fig1.gif\" ALT=\"Figure 1\">\nView larger version (49K):\norg.highwire.dtl.DTLVardef@d5a05org.highwire.dtl.DTLVardef@91295forg.highwire.dtl.DTLVardef@3338fcorg.highwire.dtl.DTLVardef@e24a58_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOFigure 1:C_FLOATNO Classification schemes for Quercus from Loudon to Nixon (see Appendix). Colour coding denotes the actual systematic affiliation of species included in each taxon: of the Old World or mid-latitude clade section Cyclobalanopsis (cycle-cup oaks, yellow), section Cerris (Cerris oaks; orange), and section Ilex (Ilex oaks; green); and of the New World or high-latitude clade section Quercus (white oaks s.str.; blue), sections Virentes (cyan) and Ponticae (dark blue), section Protobalanus (intermediate oaks; purple), and section Lobatae (red oaks; red). Colour gradients are proportional, i.e. reflect the proportion of species with different systematic affiliation included in each taxon. Note: Menitsky (1984) and Trelease (1924) only treated the Eurasian and American oaks, respectively, and provided classifications in (nearly) full agreement with current phylogenies.\n\nC_FIG

plant biology↗

bioRxiv2017

Thoracolumbar Intervertebral Disc Area Morphometry In Elderly Chinese Men And Women: Radiographic Quantifications At Baseline And Changes At Year-4 Follow-Up

The manuscript submitted does not contain information about medical device(s)/drug(s). No benefits in any form have been or will be received from a commercial party related directly or indirectly to the subject of this manuscript.\n\nStudy DesignA population-based radiographic study with longitudinal follow-up.\n\nObjectiveTo develop a quantitative index for lumbar disc space narrowing (DSN) evaluation in elderly subjects; to determine how DSN in the elderly is influenced by osteoporosis and gender.\n\nSummary of Background DataThere is paucity of research on quantitative classification of lumbar DSN based on disc areal morphometry.\n\nMethodsWith the database of Osteoporotic Fractures in Men (Hong Kong) and Osteoporotic Fractures in Women (Hong Kong) Studies and those who attended the year-4 follow-up (n = 1519 for men and n = 1546 for women), data of 491 women and 592 men were randomly selected. The anterior, middle, and posterior heights, anteroposterior diameter and area of intervertebral discs (T4T5 to L4L5) were measured on lateral radiographs. Disc Area Index for Lumbar Spine (DAIL, disc area divided by the mean of the sum of square of the adjacent upper and lower vertebrae mid-height anterior-posterior diameter) was developed and compared with semi-quantitative DSN expert grading.\n\nResultsDAIL correlated with semi-quantitative grading, with sensitivity and specificity varying from 87.3% to 96.8% for grade-1 DSN (<30% reduction in disc height), and 92.9 % to 100% for grade-3 DSN (>60% reduction in disc height). The thoracolumbar disc area loss among men and women during 4-years follow-up period varied between 1.32% and 3.56%, and it was greater for women (mean: 2.44%) than for men (mean: 1.90%, p=0.044). Majority of lumbar DSN progressions during 72 to 76 years old were progression from normal disc space to grade-1DSN. Osteoporosis was associated with greater disc area decrease, both for thoracic and lumbar discs.\n\nConclusionLumbar DSN can be quantified using DAIL. In elderly Chinese, intervertebral disc narrowing over a 4-year period was greater in women than men, and associated with the presence of osteoporosis.

epidemiology↗

bioRxiv2017

A Combined Use Of Intravoxel Incoherent Motion MRI Parameters Can Differentiate Early Stage Hepatitis-B Fibrotic Livers From Healthy Livers

This study investigated a combined use of IVIM parameters Dslow (D), PF (f) and Dfast (D*) for liver fibrosis evaluation. 16 healthy volunteers (F0) and 33 hepatitis-b patients (stage F1= 15, stage F2-4 = 18) were included. With a 1.5-T MR scanner and respiration-gating, IVIM diffusion weighted imaging was acquired using a single-shot echo-planar imaging sequence with ten b-values of 10, 20, 40, 60, 80, 100, 150, 200, 400, and 800 s/mm2. Signal measurement was performed on right liver parenchyma. With a 3-dimensional tool, Dslow, PF, and Dfast values were placed along the x-axis, y-axis, and z-axis, and a plane was defined to separate healthy volunteers from patients. 3-dimensional tool demonstrated healthy volunteers and all patients with liver fibrosis could be separated. Classification and Regression Tree showed a combination of PF (PF < 12.55%), Dslow (Dslow < 1.152 x10-3 mm2/s) and Dfast (Dfast <13.36 x10-3 mm2/s) could differentiate healthy subjects and all fibrotic livers (F1-F4) with an area under the curve of logistic regression (AUC) of 0.986. The AUC for differentiation of healthy livers vs. F2-4 livers was 1. PF offered the best diagnostic value, followed by Dslow; however, all three parameters of PF, Dslow, and Dfast contributed to liver fibrosis detection.

biophysics↗