Biology subjects

Coresh, J.

Publications and source records attributed to Coresh, J..

11 recordsLinked to original sources

bioRxiv2026

Machine learning cross-platform proteomic imputation enables protein quality scoring and replication of epidemiological associations

High-throughput affinity-based proteomics has advanced biomedical research, yet fundamental, persistent discordance between mainstream platforms (SomaScan and Olink) routinely undermines the replication of findings. This platform-driven non-replication complicates downstream biological validation and biomarker prioritization. Here, we develop a machine learning-based framework for cross-platform protein value imputation to resolve this translational bottleneck. Using paired proteomic data measured by both SomaScan and Olink from 5,325 participants of the Multi-Ethnic Study of Atherosclerosis, we developed models to impute cross-platform measurements and applied them to two independent and demographically distinct cohorts (Cardiovascular Health Study [N=3,171] and UK Biobank [UKB; N=41,405]) for external validation. Our bi-directional model 1) established an imputation performance-based protein fidelity index, validated against gold-standard measurements from Atherosclerosis Risk in Communities study (N=101) and Nurses Health Study (N=54), 2) enabled imputation of platform-exclusive protein measurements, and 3) facilitated calibration of overlapping proteins. We demonstrate the utility of this framework through three applications: 1) fidelity-informed analyses enhanced the replication of biomarker discovery, 2) recovery of SomaScan signals that were previously inaccessible in UKBs original Olink measurements, and 3) improved replication performance for overlapping proteins. Our study offers a translational roadmap that allows researchers to achieve reliable epidemiological replication, target specific assays for future optimization, and prioritize biological signal over platform noise.

bioinformatics↗

bioRxiv2025

cpiVAE: Robust and Interpretable Cross-Platform Proteomics Imputation

Large-scale plasma proteomic studies often use different high-throughput affinity platforms, and measurements of the same protein across platforms are often discordant. Discordance hinders cross-study integration. Improving proteomics data integration would enable more powerful meta-analyses, improve statistical power for biomarker discovery, and provide a better understanding of proteome-phenotype relationships. Here we present a cross-platform proteomics imputation variational autoencoder (cpiVAE), a deep generative model for bidirectional imputation of protein abundances between two widely used platforms: Olink and SomaScan. Using a training cohort of paired measurements from the China Kadoorie Biobank (CKB), cpiVAE learns a joint latent representation that enables cross-platform imputation. The cpiVAE method improves benchmarks provided by established methods, k-nearest neighbors (KNN) Weighted Nearest Neighbors (WNN, from Seurat v4). The cpiVAE method achieves up to 30% higher correlation between imputed and true values than KNN and WNN. The cpiVAE method also generalizes well to an independent cohort from the Atherosclerosis Risk in Communities Study (ARIC). Without retraining, cpiVAE maintains high performance compared to benchmarks. Associations of imputed protein levels with clinical phenotypes closely mirror results using the actual measurements and increases power in a meta-analysis scenario. A post-hoc feature importance matrix enables interpretation of this AI model. Protein pair features extracted from cpiVAE have significant overlap with known associations in the Search Tool for the Retrieval of Interacting Genes (STRING) database. In summary, cpiVAE offers an accurate, generalizable, and interpretable solution for cross-platform proteomic imputation, enabling integrated analyses across studies with proteomics measured on different platforms. This user-friendly framework and pre-trained model weights are available under a BSD2 open source license at https://github.com/joelbaderlab/cpiVAE_v1.

bioinformatics↗

bioRxiv2023

Human Plasma Proteomic Profile of Clonal Hematopoiesis

Plasma proteomic profiles associated with subclinical somatic mutations in blood cells may offer novel insights into downstream clinical consequences. Here, we explore such patterns in clonal hematopoiesis of indeterminate potential (CHIP), which is linked to several cancer and non-cancer outcomes, including coronary artery disease (CAD). Among 61,833 ancestrally diverse participants (3,881 with CHIP) from NHLBI TOPMed and UK Biobank with blood-based DNA sequencing and proteomic measurements (1,148 proteins by SomaScan in TOPMed and 2,917 proteins by Olink in UK Biobank), we identified 32 and 345 unique proteins from TOPMed and UK Biobank, respectively, associated with the most prevalent driver genes (DNMT3A, TET2, and ASXL1). These associations showed substantial heterogeneity by driver genes, sex, and race, and were enriched for immune response and inflammation pathways. Mendelian randomization in humans, coupled with ELISA in hematopoietic Tet2-/- vs wild-type mice validation, disentangled causal proteomic perturbations from TET2 CHIP. Lastly, we identified plasma proteins shared between CHIP and CAD.

genetics↗

bioRxiv2021

Large Bi-Ethnic Study of Plasma Proteome Leads to Comprehensive Mapping of cis-pQTL and Models for Proteome-wide Association Studies

Improved understanding of genetic regulation of proteome can facilitate the identification of causal mechanisms for complex traits. We analyzed data on 4,657 plasma proteins from 7,213 European American (EA) and 1,871 African American (AA) individuals from the ARIC study, and further replicated findings on 467 AA individuals from the AASK study. Here we identified 2,004 proteins in EA and 1,618 in AA, with majority overlapping, which showed associations with common variants in cis-regions. Availability of AA samples led to smaller credible sets and significant number of population-specific cis-pQTLs. Elastic-net produced powerful models for protein prediction in both populations. An application of proteome-wide association studies (PWAS) to serum urate and gout, implicated several proteins, including IL1RN, revealing the promise of the drug anakinra to treat acute gout flares. Our study demonstrates the value of large and diverse ancestry study for genetic mechanisms of molecular phenotypes and their relationship with complex traits.

genetics↗

bioRxiv2020

Polygenic Risk Scores for Kidney Function to the Circulating Proteome, and Incident Kidney Diseases: the Atherosclerosis Risk in Communities Study

Genome-wide association studies (GWAS) have revealed numerous loci for kidney function (estimated glomerular filtration rate, eGFR). The relationship of polygenic predictors of eGFR, risk of incident adverse kidney outcomes, and the plasma proteome is not known. We developed a genome-wide polygenic risk score (PRS) using a weighted average of 1.2 million SNPs for eGFR using the LDpred algorithm, summary statistics generated by a European-ancestry (EA) meta-analysis of the CKDGen Consortium (N=558,423) and UK Biobank GWAS for eGFR (90% of the cohort; N=289,432), followed by best parameter selection using data from the remaining 10% of the UK Biobank (N=32,159). We then tested the association of the PRS among 8,886 EA participants in the Atherosclerosis Risk in Communities (ARIC) study (mean age: 54{+/-}6 years, 53% female) with incident chronic kidney disease (CKD), end stage kidney disease (ESKD), kidney failure (KF), and acute kidney injury (AKI). We also examined 4,877 plasma proteins measured at two time points (visit 3 (1993-95) and visit 5 (2011-13)) in relation to the PRS and compared associations between the proteome and eGFR itself. All models were adjusted for age, sex, center, and the first 10 principal components of ancestry. The developed PRS had an R2 for eGFR of 0.07 in ARIC. Over 30 years of follow up, the number of incident CKD, ESKD, KF, and AKI were 2,959, 137, 470, and 1,723, respectively. The PRS showed significant associations with all outcomes: hazard ratios (95% CI) per 1 SD lower PRS were 1.33 (1.28, 1.39), 1.20 (1.00, 1.42), 1.17 (1.06, 1.28), and 1.07 (1.02, 1.12) for incident CKD, ESKD, KF, and AKI respectively. The PRS was significantly associated (Bonferroni threshold P<1.02 x 10-5) with 108 proteins at both time points. The strongest associations were with cystatin-C (a marker of kidney function used in clinical practice), collagen alpha-1 (XV) chain, and desmocollin-2. All significant correlations with the PRS were negative, except those of testican-2 and angiostatin. Correlations of proteins with eGFR were much stronger than those with the PRS. Overall, we demonstrated that the PRS for eGFR is now sufficiently strong to capture risk for a spectrum of incident kidney diseases as well as broadly influence the plasma proteome.

genetics↗

bioRxiv2020

Multiethnic catalog of structural variants and their translational impact for disease phenotypes across 19,652 genomes

Genome sequencing at population scale provides unprecedented access to the genetic foundations of human phenotypic diversity, but genotype-phenotype association analyses limited to small variants have failed to comprehensively characterize the genetic architecture of human health and disease because they ignore structural variants (SVs) known to contribute to phenotypic variation and pathogenic conditions1-3. Here we demonstrate the significance of SVs when assessing genotype-phenotype associations and the importance of ethnic diversity in study design by analyzing SVs across 19,652 individuals and the translational impact on 4,156 aptamerbased proteomic measurements across 4,021 multi-ethnic samples. The majority of 304,533 SVs detected are rare, although we identified 2,336 protein-coding genes impacted by common SVs.\ We identified 64 significant SV-protein associations that comprise 36 cis- and 28 trans-acting relationships, and 21 distinct SV regions overlapped with genome-wide association study loci. These findings represent a more comprehensive mapping of regulatory and translational endophenotypes underlying health and disease.

genomics↗

bioRxiv2019

Kidney Function and Blood Pressure: A Bi-directional Mendelian Randomisation Study

ObjectiveTo evaluate the bi-directional causal relation between kidney function and blood pressure. DesignMendelian randomisation study. SettingWe performed two-sample Mendelian randomisation analyses. Genetic instruments of kidney function traits were selected from summary statistics of genome-wide association studies (GWAS) of glomerular filtration rate estimated from serum creatinine (eGFRcr) and blood urea nitrogen (BUN) and were required to be associated with both eGFRcr and BUN to ensure that the instruments were more likely to represent the underlying kidney function. Genetic instruments of blood pressure were selected from summary statistics of GWAS of systolic and diastolic blood pressure. We investigated Mendelian randomisation hypothesis using several alternative approaches, including methods that are most robust to the presence of horizontal pleiotropy. ParticipantsThe summary statistics of eGFRcr included 567,460 participants from 54 cohorts, and the summary statistics of BUN included 243,031 participants from 48 cohorts from the Chronic Kidney Disease Genetics (CKDGen) Consortium. The summary statistics of systolic and diastolic blood pressure included 757,601 participants from the UK Biobank and 78 cohorts from the International Consortium for Blood Pressure (ICBP). ResultsSignificant evidence supported the causal effects of higher kidney function on lower blood pressure with multiple methods. Based on the mode-based Mendelian randomisation analysis approach, known for its robustness to the presence of pleiotropic effect, the effect estimate for 1 SD higher in eGFRcr was -0.17 SD unit (95 % CI: -0.09 to -0.24) in systolic blood pressure (SBP) and -0.15 SD unit (95% CI: -0.07 to -0.22) in diastolic blood pressure (DBP). In contrast, the causal effects of blood pressure on kidney function were not statistically significant. ConclusionsMendelian randomisation analyses support causal effects of higher kidney function on lower blood pressure. These results suggest preventing kidney function decline can reduce the public health burden of hypertension.

genetics↗

bioRxiv2019

Mitochondrial DNA Copy Number (mtDNA-CN) Can Influence Mortality and Cardiovascular Disease via Methylation of Nuclear DNA CpGs

BackgroundMitochondrial DNA copy number (mtDNA-CN) has been associated with a variety of aging-related diseases, including all-cause mortality. However, the mechanism by which mtDNA-CN influences disease is not currently understood. One such mechanism may be through regulation of nuclear gene expression via the modification of nuclear DNA (nDNA) methylation. MethodsTo investigate this hypothesis, we assessed the relationship between mtDNA-CN and nDNA methylation in 2,507 African American (AA) and European American (EA) participants from the Atherosclerosis Risk in Communities (ARIC) study. To validate our findings we assayed an additional 2,528 participants from the Cardiovascular Health Study (CHS) (N=533) and Framingham Heart Study (FHS) (N=1,995). We further assessed the effect of experimental modification of mtDNA-CN through knockout of TFAM, a regulator of mtDNA replication, via CRISPR-Cas9. ResultsThirty-four independent CpGs were associated with mtDNA-CN at genome-wide significance (P<5x10-8). Meta-analysis across all cohorts identified six mtDNA-CN associated CpGs at genome-wide significance (P<5x10-8). Additionally, over half of these CpGs were associated with phenotypes known to be associated with mtDNA-CN, including coronary heart disease, cardiovascular disease, and mortality. Experimental modification of mtDNA-CN demonstrated that modulation of mtDNA-CN directly drives changes in nDNA methylation and gene expression of specific CpGs and nearby transcripts. Strikingly, the neuroactive ligand receptor interaction KEGG pathway was found to be highly overrepresented in the ARIC cohort (P= 5.24x10-12), as well as the TFAM knockout methylation (P=4.41x10-4) and expression (P=4.30x10-4) studies. ConclusionsThese results demonstrate that changes in mtDNA-CN influence nDNA methylation at specific loci and result in differential expression of specific genes that may impact human health and disease via altered cell signaling.

genomics↗

bioRxiv2019

Improving Proteinuria Screening with Mailed Smartphone Urinalysis Testing in Previously Unscreened Patients with Hypertension: a Randomized Controlled Trial

BackgroundProteinuria screening is recommended for patients with hypertension to screen for kidney disease and identify those at elevated risk for cardiovascular disease. However, screening rates among hypertensive patients are low. Home testing strategies may be useful in improving proteinuria screening adherence. MethodsWe conducted an individual-level, randomized trial at 55 primary care clinic sites in the Geisinger Health System to evaluate the effectiveness of a strategy using home smartphone urinalysis test (Dip.io) to complete proteinuria screening in previously unscreened non-diabetic patient portal users with hypertension. All patients received an educational letter and a standing urinalysis lab order, and then were randomized to control (usual care) or intervention. Intervention arm participants were invited to complete proteinuria screening with a mailed home smartphone urinalysis test. Co-primary outcomes were completion of proteinuria screening and number of albuminuria cases (albumin/creatinine ratio [ACR] [≥] 30 mg/g or protein/creatinine ratio [≥] 150 mg/g) at the end of 3 months. We also evaluated patient satisfaction with the home test, and compliance with recommendations for patients with newly detected albuminuria. ResultsA total of 999 patients were randomized to intervention or control. Out of 499 patients assigned to the intervention arm, 253 were reached by phone, and 69/97 (71.1%) consented patients completed the home test. Overall, the intervention increased proteinuria screening completion (28.9% vs. 18.0%; p<0.001) with no effect on the number of albuminuria cases (4 vs. 4) although only 6/57 (10.5%) patients with trace or 1+ urine dipstick protein had a follow-up quantitative test. Among the 55 patients who completed a survey after the home test, 89% preferred testing at home rather than the physicians office. ConclusionsA strategy using a home urinalysis smartphone test increased proteinuria screening rates in previously unscreened patients with hypertension and may be useful in increasing rates of proteinuria screening compliance. Future studies are needed to determine whether improving early detection of kidney disease can improve future kidney health. Trial RegistrationClinical Trial Registry: NCT03470701 https://clinicaltrials.gov/ct2/show/NCT03470701

clinical trials↗

bioRxiv2017

A Comprehensive Evaluation of the Genetic Architecture of Sudden Cardiac Arrest

BackgroundSudden cardiac arrest (SCA) accounts for 10% of adult mortality in Western populations. While several risk factors are observationally associated with SCA, the genetic architecture of SCA in the general population remains unknown. Furthermore, understanding which risk factors are causal may help target prevention strategies.\n\nMethodsWe carried out a large genome-wide association study (GWAS) for SCA (n=3,939 cases, 25,989 non-cases) to examine common variation genome-wide and in candidate arrhythmia genes. We also exploited Mendelian randomization methods using cross-trait multi-variant genetic risk score associations (GRSA) to assess causal relationships of 18 risk factors with SCA.\n\nResultsNo variants were associated with SCA at genome-wide significance, nor were common variants in candidate arrhythmia genes associated with SCA at nominal significance. Using cross-trait GRSA, we established genetic correlation between SCA and (1) coronary artery disease (CAD) and traditional CAD risk factors (blood pressure, lipids, and diabetes), (2) height and BMI, and (3) electrical instability traits (QT and atrial fibrillation), suggesting etiologic roles for these traits in SCA risk.\n\nConclusionsOur findings show that a comprehensive approach to the genetic architecture of SCA can shed light on the determinants of a complex life-threatening condition with multiple influencing factors in the general population. The results of this genetic analysis, both positive and negative findings, have implications for evaluating the genetic architecture of patients with a family history of SCA, and for efforts to prevent SCA in highrisk populations and the general community.

genetics↗

bioRxiv2017

Association between Mitochondrial DNA Copy Number and Sudden Cardiac Death: Findings from the Atherosclerosis Risk in Communities Study (ARIC)

AimsSudden cardiac death (SCD) is a major public health burden. Mitochondrial dysfunction has been implicated in a wide range of cardiovascular diseases including cardiomyopathy, heart failure, and arrhythmias, but it is unknown if it also contributes to SCD risk. We sought to examine the prospective association between mtDNA copy number (mtDNA-CN), a surrogate marker of mitochondrial function, and SCD risk.\n\nMethods and ResultsWe measured baseline mtDNA-CN in 11,093 participants from the Atherosclerosis Risk in Communities (ARIC) study. mtDNA-CN was calculated from probe intensities of mitochondrial single nucleotide polymorphisms (SNP) on the Affymetrix Genome-Wide Human SNP Array 6.0. SCD was defined as a sudden pulseless condition presumed due to a ventricular tachyarrhythmia in a previously stable individual without evidence of a non-cardiac cause of cardiac arrest. SCD cases were reviewed and adjudicated by an expert committee. During a median follow-up of 20.4 years, we observed 361 SCD cases. After adjusting for age, race, sex, and center, the hazard ratio (HR) for SCD comparing the 1st to the 5th quintiles of mtDNA-CN was 2.24 (95% CI 1.58 to 3.19; p-trend <0.001). When further adjusting for traditional CVD risk factors, prevalent CHD, heart rate, and QT interval duration, the association remained statistically significant. Spline regression models showed that the association was approximately linear over the range of mtDNA-CN values. No apparent interaction by race or by sex was detected.\n\nConclusionIn this community-based prospective study, mtDNA-CN in peripheral blood was inversely associated with the risk of SCD.

epidemiology↗