About

Seung Hoan Choi, Ph.D, is a Research Assistant Professor in the Department of Biostatistics at Boston University. His academic training and research expertise lie at the intersection of biostatistics, statistical genetics, and large-scale human genomics. Dr. Choi focuses on developing and applying rigorous statistical methods to analyze population-based sequencing data. His research encompasses both methodological development and applied genetic studies, particularly targeting cardiovascular disease, Alzheimer's disease, pulmonary function, and aging-related traits. He earned his Ph.D. and M.A. in Biostatistics from Boston University and holds a B.S. in Applied Math & Statistics and Mathematics from Stony Brook University, with a dual degree from Chungnam National University. Prior to his current academic appointment, Dr. Choi worked as a Computational Scientist II and Postdoctoral Associate at the Broad Institute of MIT and Harvard.

Research topics

• Biology
• Genetics
• Medicine
• Computational biology
• Computer Science
• Bioinformatics
• Sociology
• Evolutionary biology
• Political Science
• Demography
• Internal medicine
• Endocrinology
• Cardiology
• Data science

Selected publications

• Proton-activated chloride channel 1 is essential for innate host defense against bacterial sepsis

  Proceedings of the National Academy of Sciences · 2026-04-06

  articleOpen access

  Bacterial sepsis remains a devastating clinical problem. Here, we describe a protective role for the recently discovered acid-sensitive, proton-activated chloride channel, PACC1 (PAC/ASOR/TMEM206), during sepsis. Initially, we found PACC1 was enriched in healthy human and mouse mononuclear phagocytes, particularly macrophages, and differentially regulated by inflammatory stimuli, suggesting PACC1 involvement in innate immunity. To further investigate, we generated de novo Pacc1 knockout ( −/− ) mice, which presented without major immunologic abnormalities at baseline. Compared to wild-type (WT), Pacc1 −/− myeloid cells showed normal phagocytic uptake of acid-insensitive Escherichia coli BioParticles , but impaired development of the acidifying phagolysosome using acid-sensitive E. coli BioParticles. Transcriptomic profiling of Pacc1 −/− macrophages revealed dysregulated phagolysosomal and cytokine networks (e.g., interferons). Because phagolysosomal bacterial clearance is essential to resolve infection, we challenged Pacc1 −/− mice with intraperitoneal gram-negative E. coli sepsis. Pacc1 −/− mice displayed increased bacterial burden, immune cell infiltration, inflammation, and lethality. In contrast, phagocytosis-independent E. coli lipopolysaccharide (LPS)-induced endotoxemia yielded comparable WT and Pacc1 −/− survival, as well as similar inflammatory responses. Finally, we engineered Pacc1 -floxed ( fl/fl ) mice crossed with a myeloid lineage Cre-deleter strain to interrogate myeloid cell–intrinsic PACC1 in vivo. Consistent with a predominate role for PACC1 during phagocytosis and bacterial clearance in these cells, LysM-Cre/Pacc1 fl/fl mice exhibited impaired E. coli sepsis survival but indifferent endotoxemia phenotypes. In conclusion, PACC1 links sterilizing phagolysosomal activity with immune networks in sepsis pathobiology.

  PublisherOA PDFDOI

• Proton-Activated Chloride Channel 1 (PACC1) is essential for innate host defense against bacterial sepsis

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-01-02 · 1 citations

  articleOpen access

  Abstract Bacterial sepsis remains a devastating clinical problem. Here, we describe a protective role for the recently discovered acid-sensitive, proton-activated chloride channel, PACC1 (PAC/ASOR/TMEM206), during sepsis. Initially, we found PACC1 was enriched in healthy human and mouse mononuclear phagocytes, particularly macrophages, and differentially regulated by inflammatory stimuli, suggesting PACC1 involvement in innate immunity. To further investigate, we generated de novo Pacc1 knockout ( -/- ) mice, which presented without major immunologic abnormalities at baseline. Compared to wildtype (WT), Pacc1 -/- myeloid cells showed normal phagocytic uptake of acid-insensitive Escherichia coli BioParticles , but impaired development of the acidifying phagolysosome using acid-sensitive E. coli BioParticles. Transcriptomic profiling of Pacc1 -/- macrophages revealed dysregulated phagolysosomal and cytokine networks (e.g., interferons). Because phagolysosomal bacterial clearance is essential to resolve infection, we challenged Pacc1 -/- mice with intraperitoneal gram-negative E. coli sepsis. Pacc1 -/- mice displayed increased bacterial burden, immune cell infiltration, inflammation, and lethality. In contrast, phagocytosis-independent E. coli lipopolysaccharide (LPS)-induced endotoxemia yielded comparable WT and Pacc1 -/- survival, as well as similar inflammatory responses. Finally, we engineered Pacc1 -floxed ( fl/fl ) mice crossed with a myeloid lineage Cre-deleter strain to interrogate myeloid cell-intrinsic PACC1 in vivo . Consistent with a predominate role for PACC1 during phagocytosis and bacterial clearance in these cells, LysM-Cre/Pacc1 fl/fl mice exhibited impaired E. coli sepsis survival but indifferent endotoxemia phenotypes. In conclusion, PACC1 links sterilizing phagolysosomal activity with immune networks in sepsis pathobiology. Significance Statement Bacterial sepsis remains a major global health burden. Here, we report an essential role for the recently discovered acid-sensitive chloride channel, PACC1 (PAC/ASOR/TMEM206), in protective host defense during bacterial infection and sepsis. PACC1 is highly expressed in human and mouse phagocytic myeloid cells, particularly macrophages, where it regulates phagocytic bacterial clearance and inflammatory responses. Using de novo generated mice, we show that global or myeloid cell-targeted deletion of PACC1 impairs development of phagolysosomal acidification, confers susceptibility to bacterial infection and excessive inflammation, and undermines host defense. These findings warrant further investigation of PACC1 in sepsis pathobiology.

  PublisherOA PDFDOI

• GWAS Meta-analysis Identifies Novel Associated Loci and Points to Causal Tissues in Central Serous Chorioretinopathy

  medRxiv · 2026-05-22

  articleOpen access

  Objective: To define CSC genetic architecture and identify implicated ocular tissues, cell types, genes, and circulating proteins. Data Sources: Genome-wide data were assembled from FinnGen, All of Us, Mass General Brigham Biobank, Million Veteran Program, and a Dutch chronic CSC cohort. Serum protein quantitative trait loci, human single-cell ocular atlases, and UK Biobank macular optical coherence tomography (OCT) imaging were used for downstream analyses. Study Selection: Five European-ancestry cohorts with genome-wide data and cohort-specific CSC case-control definitions were included, comprising 2,584 cases and 1,044,455 controls. Variants present in at least 2 cohorts were meta-analyzed. Data Extraction and Synthesis: Cohort-level GWASs were adjusted for age, age squared, sex, genotyping array or batch, and 10 genetic principal components, then combined using fixed-effects inverse-variance meta-analysis. Post-GWAS analyses included gene prioritization, colocalization, Mendelian randomization, single-cell disease-relevance scoring, and testing of a CSC genetic risk score in UK Biobank OCT images. Main Outcome(s) and Measure(s): Genome-wide significant CSC loci, effector genes and proteins, tissue and cell-type enrichment, and CSC-relevant OCT abnormalities. Results: Across 11,068,938 variants, 10 loci reached genome-wide significance (P < 5e-8), including 3 novel loci near TGFB1, LINC00551, and LOC105375630 and 7 replicated loci near CFH, CD46, NOTCH4, PREX1, PTPRB, GATA5, and TNFRSF10A. Integrative analyses prioritized 10 candidate effector genes. Colocalization and Mendelian randomization implicated circulating TNFRSF10A, TGFB1, and CASP10 levels. Single-cell analyses localized genetic risk to sclera (P = 2.0e-4) and vascular endothelial cells (P = 4.0e-4), with fibroblast enrichment. In UK Biobank, OCT abnormalities were more frequent in the top vs bottom 1% of CSC genetic risk (18 of 109 [16.5%] vs 8 of 134 [6.0%]; odds ratio, 4.05; 95% CI, 1.65-10.87; P = .002). Conclusions and Relevance: In this GWAS meta-analysis, CSC susceptibility localized predominantly to scleral and vascular biology rather than primary retinal pigment epithelial dysfunction. These findings support CSC as a sclerovascular disorder and nominate complement regulation, endothelial signaling, and extracellular matrix pathways for future study.

  PublisherOA PDFDOI

• Unsupervised deep learning of electrocardiograms enables scalable human disease profiling

  npj Digital Medicine · 2025-01-12 · 17 citations

  articleOpen access

  The 12-lead electrocardiogram (ECG) is inexpensive and widely available. Whether conditions across the human disease landscape can be detected using the ECG is unclear. We developed a deep learning denoising autoencoder and systematically evaluated associations between ECG encodings and ~1,600 Phecode-based diseases in three datasets separate from model development, and meta-analyzed the results. The latent space ECG model identified associations with 645 prevalent and 606 incident Phecodes. Associations were most enriched in the circulatory (n = 140, 82% of category-specific Phecodes), respiratory (n = 53, 62%) and endocrine/metabolic (n = 73, 45%) categories, with additional associations across the phenome. The strongest ECG association was with hypertension (p < 2.2×10-308). The ECG latent space model demonstrated more associations than models using standard ECG intervals, and offered favorable discrimination of prevalent disease compared to models comprising age, sex, and race. We further demonstrate how latent space models can be used to generate disease-specific ECG waveforms and facilitate individual disease profiling.

  PublisherOA PDFDOI

• Association of genetic risk of Alzheimer’s disease and cognitive function in two European populations

  Scientific Reports · 2025-02-21 · 8 citations

  articleOpen access

  Although there is some evidence of an association between Alzheimer's disease polygenic risk score (AD PRS) and cognitive function, limited validations have been performed in large populations. We investigated the relationship between AD PRS and cognitive function in the UK Biobank in over 276,000 participants and further validated the association in the Alzheimer's Disease Neuroimaging Initiative (ADNI) sample. We developed the AD PRS (excluded the APOE variants) in the middle age UK Biobank participants (age ranged 39-72, mean age 57 years) of European ancestries by LDpred2. To validate the association of AD PRS and cognitive function internally in the UK Biobank, we linearly regressed standardized cognitive function on continuous standardized AD PRS with age at cognitive test, sex, genotyping array, first 10 principal components of genotyping, smoking, education in years, body mass index, and apolipoprotein E gene ε4 (APOE4) risk allele dosages. To validate the associations externally, we ran the linear mixed effects model in the ADNI sample free of dementia (age ranged 55-91, mean age 73), including similar covariates as fixed effects and participants' IDs as the random effect. Stratification by age, APOE4 carrier status, and cognitive status (cognitively normal or mild cognitive impairment) was also investigated. Our study validated the associations of AD PRS and cognitive function in both midlife and late-life observational cohorts. Although not all of the cognitive measures were significantly associated with AD PRS, non-verbal fluid reasoning (matrix pattern completion, β = - 0.022, P = 0.003), processing speed (such as symbol digit substitution, β = - 0.017, P = 1.08E-05), short-term memory and attention (such as pairs matching, β = - 0.014, P = 1.66E-10), and reaction time (β = - 0.010, P = 1.19E-06) were inversely associated with increasing AD PRS in the UK Biobank. Higher likelihood of cognitive impairment was also associated with higher AD PRS in the ADNI cognitive normal individuals (AD assessment scale β = 0.079, P = 0.02). In summary, we confirmed that poorer cognitive function was associated with a higher polygenic AD risk, and suggested the potential utility of the AD PRS in identifying those who may be at risk for further cognitive decline.

  PublisherOA PDFDOI

• Polygenic susceptibility to dilated cardiomyopathy underlies peripartum, alcohol-induced, and cancer therapy-related cardiomyopathies

  medRxiv · 2025-02-23 · 3 citations

  preprintOpen access

  ABSTRACT Background Rare (monogenic) variants linked to non-ischemic dilated cardiomyopathy (DCM) are enriched among individuals with peripartum (PPCM), alcohol-induced (ALCM), and cancer therapy-related (CCM) cardiomyopathies, but are present in less than 15% of cases. Whether a common variant (polygenic) predisposition to DCM also pervades these secondary cardiomyopathies remains unclear. Methods We evaluated the association of a DCM polygenic score with PPCM, ALCM, and CCM in the Mass General Brigham (MGB) Biobank (n = 42,137), with replication in the UK Biobank (n = 295,160) and FinnGen (n = 417,950). We then assessed the proportion of cases with a monogenic variant and/or a high polygenic score (defined as &gt; 80 th percentile of the score distribution). Finally, we queried medical charts to ascertain whether cardiomyopathy onset in those at high polygenic risk might have been heralded by relevant clinical risk factors. Results We identified 415 individuals with a secondary cardiomyopathy (30 with PPCM, 275 with ALCM, and 110 for CCM) across the three cohorts. The DCM polygenic score associated with PPCM (OR = 1.88 per 1 standard deviation (SD) increase in polygenic score, p= 0.001), ALCM (OR per SD = 1.38, p = 1.46E-07), and CCM (OR per SD = 1.58, p = 2.97E-06). Monogenic DCM variants were strongly associated with PPCM, ALCM, and CCM, but were present in less than 10% of cases. Roughly 40% of all secondary cardiomyopathy cases had a high polygenic score, which conferred ∼3-fold odds of cardiomyopathy (p &lt;0.001). Most secondary cardiomyopathy cases lacked known antecedent clinical risk factors. Conclusion Cases of PPCM, ALCM, and CCM are enriched for monogenic DCM variants and a high DCM polygenic score, further supporting a shared genetic susceptibility influenced by distinct environmental precipitants. Considering both monogenic and polygenic risk for DCM may improve identification of individuals predisposed to secondary cardiomyopathies, particularly among those lacking established clinical risk factors.

  PublisherOA PDFDOI

• Abstract 4373583: Exome Wide Rare Variant Analysis Identifies Novel Genes Associated with Coronary Artery Disease

  Circulation · 2025-11-03

  article1st authorCorresponding

  Background: Cardiovascular disease is a leading cause of morbidity and mortality worldwide, and coronary artery disease is its most common subtype. While common variants have been extensively studied, much of the genetic architecture of coronary artery disease remains unexplained. Rare coding variants offer an opportunity to discover novel disease-associated genes with direct biological relevance. Research question: Can we identify genes carrying rare loss of function variants associated with coronary artery disease, and what are their broader cardiometabolic effects? Methods: We analyzed sequencing data from three large biobanks including UK Biobank with 59,236 cases and 422,329 controls, All of Us with 48,395 cases and 267,650 controls, and Mass General Brigham Biobank with 10,400 cases and 39,888 controls, totaling over 847,000 individuals. We performed gene-based burden testing aggregating rare loss of function variants to identify genes associated with coronary artery disease. Our mixed effects models were adjusted for age, sex, population structure, and relatedness. Genes reaching statistical significance (P&lt;3.20x10 -6 ) were further evaluated using phenome-wide association of 1587 PheCodes and 7 cardiometabolic traits. Results: Six genes were significantly associated with coronary artery disease: LDLR (OR 4.74 P=1.06 ×10 -19 ), TTN (OR 1.49, P=8.59x10 -22 ), PKD1 (OR 1.67, P=1.25×10 -6 ), ADGRL1 (OR 4.2, P=7.24x10 -7 ), RMC1 (OR 2.84, P=1.58×10 -6 ), and RBM12 (OR 3.46, P=2.63×10 -6 ). ADGRL1, RMC1, and RBM12 showed significant associations despite no prior implication in coronary or cardiometabolic traits, in contrast to LDLR , TTN and PKD1 , which have established roles. Phenome wide analysis revealed links between the novel genes and conditions such as hypertension, obesity, diabetes, and arrhythmias. Risk factor analyses showed that ADGRL1 and RMC1 were associated with higher body mass index and triglycerides and lower high density lipoprotein cholesterol. RBM12 was additionally associated with elevated low density lipoprotein cholesterol. Conclusions and Relevance: We identified five novel genes associated with coronary artery disease through exome wide rare variant analysis. Follow up phenome wide and risk factor analyses suggest functional relevance, particularly for ADGRL1 , RMC1 , and RBM12 , offering new biological insights into coronary artery disease.

  PublisherDOI

• Meta-analysis of genome-wide associations and polygenic risk prediction for atrial fibrillation in more than 180,000 cases

  Nature Genetics · 2025-03-01 · 38 citations

  reviewOpen access
  PublisherOA PDFDOI

• Addressing Heterogeneity in Genetic Studies of Cognitive Function: Insights from the ADSP‐PHC

  Alzheimer s & Dementia · 2025-12-01

  articleOpen access

  Abstract Background The Alzheimer's Disease Sequencing Project Phenotype Harmonization Consortium (ADSP‐PHC) released harmonized longitudinal cognition data across 12 studies enabling genomic analyses of AD related endophenotypes (ng00067.v11 via NIAGADS). We conducted genome‐wide analyses of common and rare variants using whole genome sequencing (WGS) to identify genetic factors influencing cognitive function. Method We selected genetically unique ADSP participants with available cognitive function measures and WGS. After excluding individuals with unknown status or dementia at baseline, 9,698 participants with either mild cognitive impairment or normal cognition remained for a cross‐sectional study of baseline memory score. The analysis began by testing the assocations with common variant (minor allele frequency [MAF] &gt; 0.5%) using linear mixed‐effects model (Model_0) and associated aggregated rare variants (MAF ≤ 0.5%) in both coding and non‐coding regions using STAARpipeline approach. All tests were adjusted for sex, age, sequencing center, study, education, and PCs, with or without APOE status. To address heterogeneity across studies, three models were applied. Model_1 used residuals from memory scores regressed on covariates as the outcome in a linear mixed‐effects model, enabling study‐specific fixed effects. Model_2 allowed for study‐specific residual variance components. Model_3 combined the approaches of Models 1 and 2. Result Model_0 yielded a genomic inflation factor (λ) of 1.12. Adjusting for heterogeneity across studies reduced the λ values in all subsequent models: Model_1 (λ=1.11), Model_2 (λ=1.02), Model_3 (λ=1.01) (Figure 1). Across all models, APOE status was associated with baseline memory score. Additionally, a region on chromosome 5 (rs36110370, MAF=0.182), near the GOLPH3 gene, showed consistent association with lower memory score: Model_0 (β=‐0.054, p = 8.6x10 ‐8 ), Model_1( β=‐0.052, p = 1.8x10 ‐7 ), Model_2 (β=‐0.05, p = 1.4x10 ‐8 ), and Model_3 (β=‐0.047, p = 2.6x10 ‐8 ). Rare noncoding variant analysis revealed an association between TOMM40 and baseline memory score (Figure 2). Conclusion The PHC harmonized memory score retains residual structure which is effectively addressed by modeling study‐specific variance. There was a strong and consistent association of rs36110370 across models, however, the association did not always meet a genome‐wide significance threshold. Analyses of ADSP PHC language and executive scores are ongoing.

  PublisherOA PDFDOI

• Abstract 4367175: Effects of Device-Measured Physical Activity on Disease Progression and Cardiovascular Outcomes in Phenotype-Negative Carriers of Cardiomyopathy-Associated Rare Variants

  Circulation · 2025-11-03

  article

  Background: Exercise may lead to disease progression and higher risk of sudden death in certain cardiomyopathies conferred by rare genetic variants. Whether general activity recommendations (i.e.,150-300 minutes/week of moderate-to-vigorous physical activity (MVPA)] are appropriate for individuals carrying a cardiomyopathy-associated variant without overt disease (G+P-) are unclear. Research Question: Do the effects of MVPA on cardiovascular (CV) outcomes, cardiac structure and function, and risk of developing overt cardiomyopathy and arrhythmias vary according to G+P- status? Methods: In UK Biobank participants undergoing 1-week of accelerometry, we assessed: i) associations between MVPA (as a spline term) and incident CV outcomes (i.e.,atrial fibrillation [AF], heart failure [HF], myocardial infarction [MI], and stroke) using Cox models adjusted for demographic and lifestyle factors, deriving optimal levels of MVPA from splines for each outcome; ii) effects of MVPA on derived indices of cardiac structure and function in a subset with cardiac magnetic resonance (CMR) imaging; iii) effects on progression to overt cardiomyopathy and malignant arrhythmia using analogous models estimating risk of incident non-ischemic cardiomyopathy (NICM) and ventricular arrhythmias (VA). The presence of varying effects on account of G+P- status was assessed by comparing model results with non-carriers. Results: Among 84,733 individuals (age 62±8 years, 56.4% women, 3,986 G+P-) undergoing accelerometry, MVPA levels were nearly identical for G+P- individuals and non-carriers (both medians:230, quartile-1:115, quartile-3:403). In multivariable models, higher MVPA was broadly associated with lower risk of incident CV disease in the G+P- group (hazard ratio[HR] at optimal MVPA level vs zero, 95% CI, AF:0.80, 0.69–0.92; HF:0.57, 0.47–0.69; MI:0.44, 0.33–0.58; stroke:0.34,0.24–0.50]) (Figure 1 ). MVPA was also associated with similar degrees of cardiac remodeling (e.g., LV dilation) in G+P- vs non-carriers ( Figure 2 ). No associations were observed between MVPA and risk of NICM or VA in G+P- individuals. Conclusions and Relevance: MVPA generally within the range of guideline-based recommendations associates with lower risk of adverse CV outcomes and similar degrees of cardiac remodeling for G+P- individuals compared to non-carriers. Although individual-level risk stratification remains critical, G+P- individuals should be encouraged to adhere to guideline-based activity recommendations.

  PublisherDOI

Frequent coauthors

• Patrick T. Ellinor

  351 shared

• Steven A. Lubitz

  Broad Institute

  285 shared

• Lu‐Chen Weng

  Brigham and Women's Hospital

  222 shared

• Emelia J. Benjamin

  Boston Medical Center

  184 shared

• Kathryn L. Lunetta

  Boston University

  172 shared

• Carolina Roselli

  University of Groningen

  168 shared

• Mark Chaffin

  Broad Institute

  149 shared

• J. G. Smith

  Sahlgrenska University Hospital

  137 shared

Labs

• SHC LabPI