Biology subjects

Hwang, J.-H.

Publications and source records attributed to Hwang, J.-H..

6 recordsLinked to original sources

bioRxiv2026

Asymmetric inter-kingdom gene transfer and a horizontal transposon invasion in a powdery mildew fungus

Horizontal DNA transfer can introduce new genetic material into eukaryotic genomes, but how often recent transfers occur between multicellular lineages and what happens to transferred DNA after integration remain poorly understood. Plants and fungi provide a powerful system to address these questions because of their long-standing and intimate ecological associations. Here, we performed an annotation-independent, bidirectional whole-genome screen across 418 plant and 1,081 fungal genomes and identified 22 high-confidence recent horizontal transfer events. Twenty-one involved fungal DNA acquired by plants and were concentrated in bryophytes and Pooideae grasses, revealing pronounced directional and phylogenetic biases. The evolutionary fate of transferred DNA differed across recipient lineages. In Pooideae, fungal-derived genes underwent duplication, structural remodeling and expression divergence, and showed copy-number variation within species. In Marchantia polymorpha, transferred genes remained mostly single-copy but displayed presence-absence variation and ongoing loss among subspecies and natural accessions. In the opposite direction, we identified a plant-derived DNA transposon integrated into the genome of the tomato powdery mildew fungus Pseudoidium neolycopersici. Following transfer, this element remained active and generated a derived family of miniature inverted-repeat transposable elements that expanded to 172 copies. Recent detectable DNA transfer between plants and fungi is therefore uncommon and strongly asymmetric, but transferred sequences can remain highly dynamic after integration. Their subsequent loss, duplication, structural and regulatory diversification, or transposon-driven amplification links rare cross-kingdom acquisition events to ongoing genome variation and diversification.

evolutionary biology↗

bioRxiv2025

An updated resource of 180K soybean SNP genotyping array based on the T2T reference genome

Single nucleotide polymorphism (SNP) genotyping has revolutionized crop improvement by enabling high-resolution genomic analyses and accelerating breeding programs. In soybean (Glycine max (L.) Merr.), a globally important legume crop, existing genotyping data for 180,961 SNP markers from the Korean soybean core collection were generated using the outdated Williams 82 reference genome version 1 (Wm82.v1), which contains numerous assembly gaps and misassemblies that limit genomic resolution. While high-quality reference genomes including Wm82.v4 and Wm82.v6 (telomere-to-telomere assembly) are now available, the valuable existing SNP array data have not been integrated with these improved genomic resources. Here we show successful remapping of the 180K SNP array data to both Wm82.v4 and Wm82.v6 reference genomes through sequence-based alignment of flanking regions. We extracted flanking sequences from SNP marker positions in Wm82.v1 and mapped them to the newer reference versions based on sequence similarity, excluding markers with mapping failures, allele mismatches, low-identity alignments, or multiple mappings, which resulted in successful mapping of 175,202 and 175,763 markers to Wm82.v4 and Wm82.v6, respectively. We also remapped genotype data from 927 soybean accessions (497 USDA-GRIN accessions and 430 Korean core collection accessions) to both reference versions. This updated SNP dataset provides the soybean research community with a comprehensive genomic resource that leverages both existing genotyping investments and state-of-the-art reference genome assemblies for enhanced crop improvement and genomic studies.

genomics↗

bioRxiv2025

A Study of Calibration as a Measurement of Trustworthiness of Large Language Models in Biomedical Research

ObjectivesTo assess the calibration of 9 large language models (LLMs) within biomedical natural language processing (BioNLP) tasks, furthering understanding of trustworthiness and reliability in real-world settings. Materials and MethodsFor each LLM, we collected responses and corresponding confidence scores for all 13 datasets (grouped into 6 tasks) of the Biomedical Language Understanding & Reasoning Benchmark (BLURB). Confidence scores were assigned using 3 strategies: Verbal, Self-consistency, Hybrid. For evaluation, we introduced Flex-ECE (Flexible Expected Calibration Error): a novel adaptation of ECE that accounts for partial correctness in model responses, allowing for a more realistic assessment of calibration in language-based settings. Two post-hoc calibration techniques--isotonic regression and histogram binning--were evaluated. ResultsAcross tasks, mean calibration ranged from 23.9% (Population-Intervention-Comparison-Outcome extraction) to 46.6% (Relation Extraction). Across LLMs, Medicine-Llama3-8B had the best mean overall calibration (29.8%); Flan-T5-XXL had the highest ranking on 5/13 datasets. Across strategies, self-consistency (mean: 27.3%) had better calibration than Verbal (mean: 42.0%) and Hybrid (mean: 44.2%). Post-hoc methods substantially improved calibration, with best mean calibrated Flex-ECEs ranging from 0.1% to 4.1%. DiscussionThe poor out-of-the-box calibration of LLMs poses a risk to trustworthy deployment of such models in real-world BioNLP applications. Calibration can be improved post-hoc and is a recommended practice. Non-binary metrics for LLM evaluation such as Flex-ECE provide a more realistic assessment of trustworthiness of LLMs, and indeed any model that can be partially right/wrong. ConclusionThis study shows that out-of-the-box calibration of LLMs is very poor, but traditional post-hoc calibration techniques are useful to calibrate LLMs.

bioinformatics↗

bioRxiv2024

Chromosome-level genome assembly of milk thistle (Silybum marianum (L.) Gaertn.)

Silybum marianum (L.) Gaertn., commonly known as milk thistle, is a medicinal plant belonging to the Asteraceae family. This plant has been recognized for its medicinal properties for over 2,000 years. However, the genome of this plant remains largely undiscovered, having no reference genome at a chromosomal level. Here, we assembled the chromosome-level genome of S. marianum, allowing for the annotation of 53,552 genes and the identification of transposable elements comprising 58% of the genome. The genome assembly from this study showed 99.1% completeness as determined by BUSCO assessment, while the previous assembly (ASM154182v1) showed 36.7%. Functional annotation of the predicted genes showed 50,329 genes (94% of total genes) with known protein functions in public databases. Comparative genome analysis among Asteraceae plants revealed a striking conservation of collinearity between S. marianum and C. cardunculus . The genomic information generated from this study will be a valuable resource for milk thistle breeding and for use by the larger research community.

genomics↗

bioRxiv2020

New Genotypes and Diversity of Orientia tsutsugamushi Isolated from Scrub Typhus Patients in South Korea as Determined by Multilocus Sequence Typing

Orientia tsutsugamushi, an obligate intracellular organism, is the causative agent of scrub typhus, which is endemic in the Asia-Pacific region. No comparative studies on the genotypic properties of O. tsutsugamushi have been performed using multilocus sequence typing (MLST) in South Korea. Here, we characterized 51 clinical isolates from Jeonju, in southwestern Korea, and we compared them to isolates from Thailand, Laos, and Japan. We also identified 10 new alleles and six novel sequence types. Overall, our results suggest that the relative genetic stability and the clonal populations of O. tsutsugamushi strains in South Korea have remained mostly conserved. Author summaryScrub typhus is a life-threatening disease, caused by infection with O. tsutsugamushi, a Gram-negative intracellular bacterium. Approximately one million people are infected globally every year, especially in the Asia-Pacific region. Strains of O. tsutsugamushi are typically distinguished serologically on the basis of sequences of the highly polymorphic 56-kDa outer membrane protein. Multilocus sequence typing (MLST) is a generic typing method that provides a unified bacterial isolate characterization approach that can be used for evolutionary and population studies of bacteria. In this study, we describe the development and application of a MLST scheme that was applied to 51 O. tsutsugamushi isolates. We found 10 new alleles and six new STs, which yielded a total of seven O. tsutsugamushi STs in South Korea. Among seven different STs (ST 48, 93-98), ST 48 account for the largest proportion (49.0%) of O. tsutsugamushi STs in South Korea. With the exception of the appearance of six novel STs, the clonal populations have remained conserved but further study of population structure and evolutionary trends is warranted.

microbiology↗

bioRxiv2019

Salience network atrophy links neuron type-specific degeneration to loss of empathy in frontotemporal dementia

Each neurodegenerative syndrome reflects a stereotyped pattern of cellular, regional, and large-scale brain network degeneration. In behavioral variant frontotemporal dementia (bvFTD), a disorder of social-emotional function, von Economo neurons (VENs) and fork cells are among the initial neuronal targets. These large layer 5 projection neurons are concentrated in the anterior cingulate and frontoinsular (FI) cortices, regions that anchor the salience network, a large-scale system linked to social-emotional function. Here, we studied patients with bvFTD, amyotrophic lateral sclerosis (ALS), or both, given that these syndromes share common pathobiological and genetic factors. Our goal was to determine how neuron type-specific TAR DNA-binding protein of 43 kDa (TDP-43) pathobiology relates to atrophy in specific brain structures and to loss of emotional empathy, a cardinal feature of bvFTD. We combined questionnaire-based empathy assessments, in vivo structural MR imaging, and quantitative histopathological data from 16 patients across the bvFTD/ALS spectrum. We show that TDP-43 pathobiology within right FI VENs and fork cells is associated with salience network atrophy spanning insular, medial frontal, and thalamic regions. Gray matter degeneration within these structures mediated loss of emotional empathy, suggesting a chain of influence linking the cellular, regional/network, and behavioral levels in producing signature bvFTD clinical features.

neuroscience↗