Search bioRxiv⌕ Search

Biology subjects

Min, Y.

Publications and source records attributed to Min, Y..

28 records · Page 2Linked to original sources
bioRxiv2021

Molecular Graph Contrastive Learning with Parameterized Explainable Augmentations

Learning generalizable, transferable, and robust representations for molecule data has always been a challenge. The recent success of contrastive learning (CL) for self-supervised graph representation learning provides a novel perspective to learn molecule representations. The most prevailing graph CL framework is to maximize the agreement of representations in different augmented graph views. However, existing graph CL frameworks usually adopt stochastic augmentations or schemes according to pre-defined rules on the input graph to obtain different graph views in various scales (e.g. node, edge, and subgraph), which may destroy topological semantemes and domain prior in molecule data, leading to suboptimal performance. Therefore, designing parameterized, learnable, and explainable augmentation is quite necessary for molecular graph contrastive learning. A well-designed parameterized augmentation scheme can preserve chemically meaningful structural information and intrinsically essential attributes for molecule graphs, which helps to learn representations that are insensitive to perturbation on unimportant atoms and bonds. In this paper, we propose a novel Molecular Graph Contrastive Learning with Parameterized Explainable Augmentations, MolCLE for brevity, that self-adaptively incorporates chemically significative information from both topological and semantic aspects of molecular graphs. Specifically, we apply deep neural networks to parameterize the augmentation process for both the molecular graph topology and atom attributes, to highlight contributive molecular substructures and recognize underlying chemical semantemes. Comprehensive experiments on a variety of real-world datasets demonstrate that our proposed method consistently outperforms compared baselines, which verifies the effectiveness of the proposed framework. Detailedly, our self-supervised MolCLE model surpasses many supervised counterparts, and meanwhile only uses hundreds of thousands of parameters to achieve comparative results against the state-of-the-art baseline, which has tens of millions of parameters. We also provide detailed case studies to validate the explainability of augmented graph views. CCS CONCEPTS* Mathematics of computing [->] Graph algorithms; * Applied computing [->] Bioinformatics; * Computing methodologies [->] Neural networks; Unsupervised learning.

bioinformatics↗
bioRxiv2021

Genetic architecture underlying variation in floral meristem termination in Aquilegia

Floral organs are produced by floral meristems (FMs), which harbor stem cells in their centers. Since each flower only has a finite number of organs, the stem cell activity of a FM will always terminate at a specific time point, a process termed floral meristem termination (FMT). Variation in the timing of FMT can give rise to floral morphological diversity, but how this process is fine-tuned at a developmental and evolutionary level is poorly understood. Flowers from the genus Aquilegia share identical floral organ arrangement except for stamen whorl numbers (SWN), making Aquilegia a well-suited system for investigation of this process: differences in SWN between species represent differences in the timing of FMT. By crossing A. canadensis and A. brevistyla, quantitative trait locus (QTL) mapping has revealed a complex genetic architecture with seven QTL. We identified potential candidate genes under each QTL and characterized novel expression patterns of select candidate genes using in situ hybridization. To our knowledge, this is the first attempt to dissect the genetic basis of how natural variation in the timing of FMT is regulated and our results provide insight into how floral morphological diversity can be generated at the meristematic level.

plant biology↗
bioRxiv2021

Quantitative live-imaging of Aquilegia floral meristems reveals distinct patterns of floral organ initiation and cell-level dynamics of floral meristem termination

In-depth investigation of any developmental process in plants requires knowledge of both the underpinning molecular networks and how they directly determine patterns of cell division and expansion over time. Floral meristems (FM) produce floral organs, after which they undergo floral meristem termination (FMT), and precise control of organ initiation and FMT is crucial to reproductive success of any flowering plant. Using a live confocal imaging, we characterized developmental dynamics during floral organ primordia initiation and FMT in Aquilegia coerulea (Ranunculaceae). Our results have uncovered distinct patterns of primordium initiation between stamens and staminodes compared to carpels, and provided insight into the process of FMT, which is discernable based on cell division dynamics preceding carpel initiation. To our knowledge, this is the first quantitative live imaging of meristem development in a system with numerous whorls of floral organs as well as an apocarpous gynoecium. This study provides crucial information for our understanding of how the spatial-temporal regulation of floral meristem behavior is achieved in both an evolutionary and developmental context.

developmental biology↗
bioRxiv2021

Transcriptional landscape of human microglia reveals robust gene expression signatures that implicates age, sex and APOE-related immunometabolic pathway perturbations

Microglia have fundamental roles in health and disease, however effects of age, sex and genetic factors on human microglia have not been fully explored. We applied bulk and single cell approaches to comprehensively characterize human microglia transcriptomes and their associations with age, sex and APOE. We identified a novel microglial signature, characterized its expression in bulk tissue and single cell microglia transcriptomes. We discovered microglial co-expression network modules associated with age, sex and APOE-{varepsilon}4 that are enriched for lipid and carbohydrate metabolism genes. Integrated analyses of modules with single cell transcriptomes revealed significant overlap between age-associated module genes and both pro-inflammatory and disease-associated microglial clusters. These modules and clusters harbor known neurodegenerative disease genes including APOE, PLCG2 and BIN1. Meta-analyses with published bulk and single cell microglial datasets further supported our findings. Thus, these data represent a well-characterized human microglial transcriptome resource; and highlight age, sex and APOE-related microglial immunometabolism perturbations with potential relevance in neurodegeneration.

genetics↗
bioRxiv2021

Genetic architecture of floral traits in bee- and hummingbird-pollinated sister species of Aquilegia (columbine)

Interactions with animal pollinators have helped shape the stunning diversity of flower morphologies across the angiosperms. A common evolutionary consequence of these interactions is that some flowers have converged on suites of traits, or pollination syndromes, that attract and reward specific pollinator groups. Determining the genetic basis of these floral pollination syndromes can help us understand the processes that contributed to the diversification of the angiosperms. Here, we characterize the genetic architecture of a bee-to-hummingbird pollination shift in Aquilegia (columbine) using QTL mapping of 17 floral traits encompassing color, nectar composition, and organ morphology. In this system, we find that the genetic architectures underlying differences in floral color are quite complex, and we identify several likely candidate genes involved in anthocyanin and carotenoid floral pigmentation. Most morphological and nectar traits also have complex genetic underpinnings; however, one of the key floral morphological phenotypes, nectar spur curvature, is shaped by a single locus of large effect.

plant biology↗
bioRxiv2020

A Hybrid Cellular and Heterogeneous Catalyst Strategy for the Production of Olefins from Glucose

Living systems provide a promising approach to chemical synthesis, having been optimized by evolution to convert renewable carbon sources such as glucose to an enormous range of small molecules. However, a large number of synthetic structures can still be difficult to obtain solely from cells, such as unsubstituted hydrocarbons. In this work, we demonstrate the use of a hybrid cellular-heterogeneous catalytic strategy to produce olefins from glucose, using a selective hydrolase to generate an activated intermediate that is readily deoxygenated. Using a new family of iterative thiolase enzymes, we have genetically engineered a microbial strain that produces 4.3 {+/-} 0.4 g L-1 of fatty acid from glucose with 86% captured as 3-hydroxyoctanoic and 3-hydroxydecanoic acids. This 3-hydroxy substituent serves as a leaving group enabling heterogeneous tandem decarboxylation-dehydration routes to olefinic products on Lewis acidic catalysts without the additional redox input required for enzymatic or chemical deoxygenation of simple fatty acids.

synthetic biology↗
bioRxiv2019

Oxygen tension regulates lysosomal activation and receptor tyrosine kinase degradation

Oxygen sensing is crucial for adaptation to variable habitats and physiological conditions. Low oxygen tension, or hypoxia, is a common feature of solid tumors and hypoxic tumors are often more aggressive and resistant to therapy. Here we show that, in mammalian tissue culture cells, hypoxia suppressed lysosomal acidification/activation and receptor tyrosine kinase (RTK) degradation. Hypoxia down-regulated mTORc1, reducing its ability to activate transcription factor EB (TFEB), a master regulator of v-ATPase, the lysosomal proton pump. Hypoxia prevented epidermal growth factor receptor (EGFR) degradation in tumor tissues, whereas activation of lysosomes enhanced tumor cell response to anti-EGFR treatment. Our results link oxygen tension and lysosomal activity, provide a molecular explanation of the malignant phenotype associated with hypoxic tumors, and suggest activation of lysosomes may provide therapeutic benefit in RTK-targeted cancer therapy.\n\nOne Sentence SummaryHypoxia suppresses lysosomal activation and proteolysis.

cell biology↗
bioRxiv2019

Vav1 is essential for HIF-1α activation in vascular response to ischemic stress

Vascular response to hypoxia is a major determinant of organ function under stress, which is particularly critical for vital organs such as the heart. This study identifies Vav1 as a key vascular regulator of hypoxia. Vav1 is present in vascular endothelium and is essential for HIF-1 activation under hypoxia. It regulates HIF-1 stabilization through the p38/Siah2/PHD3 pathway. Consequently, Vav1 deficient mice are predisposed to sudden death under cardiac ischemia with increased coronary endothelial apoptosis. Moreover, Vav1 binds to VEGFR1 that carries Vav1 to lysosomes for degradation in normoxia. Hypoxia upregulates Vav1 through inhibition of protein degradation. These findings reveal that regulation of Vav1 by hypoxia is analogous to HIF-1 regulation. Both proteins are constitutively produced allowing for rapid responses when stress occurs, and constantly degraded in normoxia. Hypoxia stabilizes Vav1, which is required for HIF-1 accumulation. Together they mediate the vascular response to hypoxia and maintain tissue homeostasis.

cancer biology↗
bioRxiv2018

Learning Protein Structural Fingerprints under the Label-Free Supervision of Domain Knowledge

Finding homologous proteins is the indispensable first step in many protein biology studies. Thus, building highly efficient \"search engines\" for protein databases is a highly desired function in protein bioinformatics. As of August 2018, there are more than 140,000 protein structures in PDB, and this number is still increasing rapidly. Such a big number introduces a big challenge for scanning the whole structure database with high speeds and high sensitivities at the same time. Unfortunately, classic sequence alignment tools and pairwise structure alignment tools are either not sensitive enough to remote homologous proteins (with low sequence identities) or not fast enough for the task. Therefore, specifically designed computational methods are required for quickly scanning structure databases for homologous proteins.\n\nHere, we propose a novel ContactLib-DNN method to quickly scan structure databases for homologous proteins. The core idea is to build structure fingerprints for proteins, and to perform alignment-free comparisons with the fingerprints. Specifically, the fingerprints are low-dimensional vectors representing the contact groups within the proteins. Notably, the Cartesian distance between two fingerprint vectors well matches the RMSD between the two corresponding contact groups. This is done by using RMSD as the domain knowledge to supervise the deep neural network learning. When comparing to existing methods, ContactLib-DNN achieves the highest average AUROC of 0.959. Moreover, the best candidate found by ContactLib-DNN has a probability of 70.0% to be a true positive. This is a significant improvement over 56.2%, the best result produced by existing methods.\n\nGitHub: https://github.com/Chenyao2333/contactlib/\n\nIndex Termshomologous proteins, protein structures, remote protein homolog detection, alignment-free comparisons

bioinformatics↗