Yonsei University

About: Yonsei University is a education organization based out in Seoul, South Korea. It is known for research contribution in the topics: Population & Cancer. The organization has 50162 authors who have published 106172 publications receiving 2279044 citations. The organization is also known as: Yonsei.

Graphyne: Hexagonal network of carbon with versatile Dirac cones

Abstract: We study \ensuremath{\alpha}, \ensuremath{\beta}, and \ensuremath{\gamma} graphyne, a class of graphene allotropes with carbon triple bonds, using a first-principles density-functional method and tight-binding calculation. We find that graphyne has versatile Dirac cones and it is due to remarkable roles of the carbon triple bonds in electronic and atomic structures. The carbon triple bonds modulate effective hopping matrix elements and reverse their signs, resulting in Dirac cones with reversed chirality in \ensuremath{\alpha} graphyne, momentum shift of the Dirac point in \ensuremath{\beta} graphyne, and switch of the energy gap in \ensuremath{\gamma} graphyne. Furthermore, the triple bonds provide chemisorption sites of adatoms which can break sublattice symmetry while preserving planar $s{p}^{2}$-bonding networks. These features of graphyne open new possibilities for electronic applications of carbon-based two-dimensional materials and derived nanostructures.

NOD-Like Receptors in Infection, Immunity, and Diseases

Abstract: Nucleotide-binding and oligomerization domain (NOD)-like receptors (NLRs) are pattern-recognition receptors similar to toll-like receptors (TLRs). While TLRs are transmembrane receptors, NLRs are cytoplasmic receptors that play a crucial role in the innate immune response by recognizing pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs). Based on their N-terminal domain, NLRs are divided into four subfamilies: NLRA, NLRB, NLRC, and NLRP. NLRs can also be divided into four broad functional categories: inflammasome assembly, signaling transduction, transcription activation, and autophagy. In addition to recognizing PAMPs and DAMPs, NLRs act as a key regulator of apoptosis and early development. Therefore, there are significant associations between NLRs and various diseases related to infection and immunity. NLR studies have recently begun to unveil the roles of NLRs in diseases such as gout, cryopyrin-associated periodic fever syndromes, and Crohn's disease. As these new associations between NRLs and diseases may improve our understanding of disease pathogenesis and lead to new approaches for the prevention and treatment of such diseases, NLRs are becoming increasingly relevant to clinicians. In this review, we provide a concise overview of NLRs and their role in infection, immunity, and disease, particularly from clinical perspectives.

Genome-wide association study identifies FCGR2A as a susceptibility locus for Kawasaki disease.

Abstract: Kawasaki disease is a systemic vasculitis of unknown etiology, with clinical observations suggesting a substantial genetic contribution to disease susceptibility. We conducted a genome-wide association study and replication analysis in 2,173 individuals with Kawasaki disease and 9,383 controls from five independent sample collections. Two loci exceeded the formal threshold for genome-wide significance. The first locus is a functional polymorphism in the IgG receptor gene FCGR2A (encoding an H131R substitution) (rs1801274; P = 7.35 × 10(-11), odds ratio (OR) = 1.32), with the A allele (coding for histadine) conferring elevated disease risk. The second locus is at 19q13, (P = 2.51 × 10(-9), OR = 1.42 for the rs2233152 SNP near MIA and RAB4B; P = 1.68 × 10(-12), OR = 1.52 for rs28493229 in ITPKC), which confirms previous findings(1). The involvement of the FCGR2A locus may have implications for understanding immune activation in Kawasaki disease pathogenesis and the mechanism of response to intravenous immunoglobulin, the only proven therapy for this disease.

Systematic comparison of variant calling pipelines using gold standard personal exome variants

Abstract: The success of clinical genomics using next generation sequencing (NGS) requires the accurate and consistent identification of personal genome variants. Assorted variant calling methods have been developed, which show low concordance between their calls. Hence, a systematic comparison of the variant callers could give important guidance to NGS-based clinical genomics. Recently, a set of high-confident variant calls for one individual (NA12878) has been published by the Genome in a Bottle (GIAB) consortium, enabling performance benchmarking of different variant calling pipelines. Based on the gold standard reference variant calls from GIAB, we compared the performance of thirteen variant calling pipelines, testing combinations of three read aligners—BWA-MEM, Bowtie2 and Novoalign—and four variant callers—Genome Analysis Tool Kit HaplotypeCaller (GATK-HC), Samtools mpileup, Freebayes and Ion Proton Variant Caller (TVC), for twelve data sets for the NA12878 genome sequenced by different platforms including Illumina2000, Illumina2500 and Ion Proton, with various exome capture systems and exome coverage. We observed different biases toward specific types of SNP genotyping errors by the different variant callers. The results of our study provide useful guidelines for reliable variant identification from deep sequencing of personal genomes.

Attention-Based Dropout Layer for Weakly Supervised Object Localization

Abstract: Weakly Supervised Object Localization (WSOL) techniques learn the object location only using image-level labels, without location annotations. A common limitation for these techniques is that they cover only the most discriminative part of the object, not the entire object. To address this problem, we propose an Attention-based Dropout Layer (ADL), which utilizes the self-attention mechanism to process the feature maps of the model. The proposed method is composed of two key components: 1) hiding the most discriminative part from the model for capturing the integral extent of object, and 2) highlighting the informative region for improving the recognition power of the model. Based on extensive experiments, we demonstrate that the proposed method is effective to improve the accuracy of WSOL, achieving a new state-of-the-art localization accuracy in CUB-200-2011 dataset. We also show that the proposed method is much more efficient in terms of both parameter and computation overheads than existing techniques.