SEARCH · PubMed Health

Results for “Bayesian model selection”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

30 records · Page 2Linked to original sources

PubMed2026

Polygenic Risk Scores for Incident Dementia in the Multi-Ethnic Study of Atherosclerosis.

Over 75 Alzheimer's disease (AD) and dementia-associated variants have been identified through genome-wide association studies, but the utility of polygenic risk scores (PRS) for predicting AD and dementia in diverse and admixed populations remains unclear. We compared how PRS approaches differing in p-value thresholds, variant weights, and source ancestry perform in predicting dementia in 6338 African American, Chinese, Hispanic, and White individuals from the Multi-Ethnic Study of Atherosclerosis. We tested clumping and thresholding (C+T) methods with varying parameters against Bayesian approaches (PRS-CS, PRS-CSx). We compared the ability of each method to predict incident dementia in all participants and in groups stratified by self-reported race/ethnicity. We additionally analyzed performance across groups stratified by estimated proportion of non-Finnish European (NFE)-like ancestry. Including more variants does not improve performance. We found comparable associations between dementia and PRS when comparing a C+T method with only 15 SNPs and PRS derived from Bayesian models that include > 800,000 SNPs (HR5e-08 = 1.18, 95% CI: 1.08-1.28; HRCSx = 1.17, 95% CI: 1.07-1.27). The p < 5e-08 C+T method was more strongly associated with incident dementia in populations genetically dissimilar from the source data (HRlowNFE_5e-08 = 1.27, 95% CI: 1.08-1.50; HRlowNFE_CSx = 1.12, 95% CI: 0.94-1.33). More selective PRS models using genome-wide significant SNPs may be preferable for dementia prediction in diverse populations.

Aged↗

PubMed2026

Enhancing detection of polygenic adaptation: a comparative study of machine learning and statistical approaches using simulated evolve-and-resequence data.

BACKGROUND: Detecting signals of polygenic adaptation remains a significant challenge in population genomics, as traditional methods often struggle to identify the associated subtle, multi-locus allele-frequency shifts. Here, we introduced and tested several novel approaches combining machine learning techniques with traditional statistical tests to detect polygenic adaptation patterns in time-series of allele frequency changes from whole genome data. We implemented a Naive Bayesian Classifier (NBC) and One-Class Support Vector Machines (OCSVM), and compared their performance against the classical Fisher's Exact Test (FET). Furthermore, we combined machine learning and statistical models (OCSVM-FET and NBC-FET), resulting in 5 competing approaches. The framework is mainly designed and validated for evolve-and-resequence (EaR) experimental designs, where defined selection pressures and temporal sampling are feasible, but might be applicable for certain natural experiments as well. RESULTS: Using a simulated dataset based on empirical C. riparius Pool-Seq data, we evaluated methods across evolutionary scenarios varying in generation, selection strength, and number of loci under selection. Our results demonstrate that the combined OCSVM-FET approach consistently outperformed competing methods, achieving the lowest false positive rate, highest area under the curve, and high accuracy. The performance peak aligned with what we term the 'late dynamic phase' of adaptation - the period after initial selection has occurred but before fixation - highlighting the method's sensitivity to ongoing selective processes. CONCLUSIONS: Furthermore, we emphasize the critical role of parameter tuning, balancing biological assumptions with methodological rigor. While broader applicability remains an important direction for future work, the present benchmarking is intentionally scoped to EaR experimental contexts.

Machine Learning↗

PubMed2025

Comparing artificial and convolutional neural networks with traditional models for Genomic prediction in wheat.

With the rapid development of sequencing technology, the application of genomic prediction has become more and more common in breeding schemes of livestocks and crops. Selecting an appropriate statistical model is of central importance to achieve high prediction accuracy. Recently, machine learning models have been expected to upgrade genomic prediction into a new era. However, the perspective still suffers from lack of evidence that machine learning models can generally outperform the traditional ones on empirical data sets. In this study, we compared two machine learning models based on artificial neural network (ANN) and convolutional neural network (CNN) with four traditional models, including genomic best linear unbiased prediction (GBLUP), Bayesian ridge regression (BRR), BayesA and BayesB, using three published data sets for grain yield in wheat. For each model, we considered two variants: modeling and ignoring the genotype-by-environment ([Formula: see text]) interaction. In the comparison, we considered two strategies of cross-validation: predicting genotypes that have not been evaluated in any environment (CV1) and predicting genotypes that have been tested in other environments (CV2). Our results showed that traditional Bayesian models (BayesA, BayesB, and BRR) outperformed GBLUP, ANN and CNN when considering [Formula: see text] interaction. The accuracies of ANN and CNN were higher than traditional models only in CV1 and when [Formula: see text] interaction was ignored. It was also found that the performance of the two machine learning models was significantly affected by the interaction between the CV strategy and the way of treating the [Formula: see text] interaction, while that of the four traditional models was only influenced by whether the [Formula: see text] interaction was considered or not. Thus, machine learning models can be a powerful complementary to the traditional ones and their superiority may depend on the prediction scenario. Among the two machine learning models, we observed that the accuracy of ANN was higher than CNN in most cases, indicating that it is still challenging to adapt complex machine learning models such as CNN to genomic prediction.

ANN↗

PubMed2025

BISON: bi-clustering of spatial omics data with feature selection.

MOTIVATION: The advent of next-generation sequencing-based spatially resolved transcriptomics (SRT) techniques has reshaped genomic studies by enabling high-throughput gene expression profiling while preserving spatial and morphological context. Understanding gene functions and interactions in different spatial domains is crucial, as it can enhance our comprehension of biological mechanisms, such as cancer-immune interactions and cell differentiation in various regions. It is necessary to cluster tissue regions into distinct spatial domains and identify discriminating genes (DGs) that elucidate the clustering result, referred to as spatial domain-specific DGs. Existing methods for identifying these genes typically rely on a two-stage approach, which can lead to the phenomenon known as double-dipping. RESULTS: To address the challenge, we propose a unified Bayesian latent block model that simultaneously detects a list of DGs contributing to spatial domain identification while clustering these DGs and spatial locations. The efficacy of our proposed method is validated through a series of simulation experiments, and its capability to identify DGs is demonstrated through applications to benchmark SRT datasets. AVAILABILITY AND IMPLEMENTATION: The R/C++ implementation of BISON is available at https://github.com/new-zbc/BISON.

Software↗

PubMed2025

Sparse polygenic risk score inference with the spike-and-slab LASSO.

MOTIVATION: Large-scale biobanks, with rich phenotypic and genomic data across hundreds of thousands of samples, provide ample opportunities to elucidate the genetics of complex traits and diseases. Consequently, there is growing demand for robust and scalable methods for disease risk prediction from genotype data. Inference in this setting is challenging due to the high-dimensionality of genomic data, especially when coupled with smaller sample sizes. Popular Polygenic Risk Score (PRS) inference methods address this challenge by adopting sparse Bayesian priors or penalized regression techniques, such as the Least Absolute Shrinkage and Selection Operator (LASSO). However, the former class of methods are not as scalable and do not produce exact sparsity, while the latter tends to over-shrink large coefficients. RESULTS: In this study, we present SSLPRS, a novel PRS method based on the Spike-and-Slab LASSO (SSL) prior, which offers a theoretical bridge between the two frameworks. We extend previous work to derive a coordinate-ascent inference algorithm that operates on GWAS summary statistics, which is orders-of-magnitude more efficient than corresponding individual-level-based implementations. To illustrate the statistical properties of the proposed model, we conducted experiments involving nine simulation configurations and nine quantitative phenotypes from the UK Biobank. Our results demonstrate that SSLPRS is competitive with state-of-the-art methods in terms of prediction accuracy and exhibits superior variable selection performance, especially in sparse genetic architectures. In simulations, this translates to upwards of 50% improvement in positive predictive value. In analysis of real phenotypes, we show that selected variants are highly enriched for meaningful genomic annotations and have better replication rates in larger meta-analyses. AVAILABILITY AND IMPLEMENTATION: SSLPRS is available in the open-source package https://github.com/li-lab-mcgill/penprs.

Multifactorial Inheritance↗

PubMed2026

Precision targeting of teacher burnout using network-informed ecological momentary interventions.

Teacher well-being affects classroom functioning and workforce stability, yet generic digital programs rarely use person-specific affect dynamics to select support. This cluster-randomised trial evaluated whether micro-interventions selected from high expected influence (EI) nodes in teachers' contemporaneous affect networks produced larger changes in burnout-related EI and everyday happiness than content-matched random allocation. The objectives were to estimate allocation effects on changes in estimated network summaries and happiness, evaluate network change as a statistical mediator, examine personality moderation, and benchmark simpler allocation rules. A two-arm cluster randomised platform trial was conducted in 84 public schools across four urban districts in H Province. After a 14 day baseline of ecological momentary assessment (EMA), person specific partial correlation networks were estimated for happiness, exhaustion, detachment, efficacy and rumination. An optimisation engine prioritised three brief micro-intervention types per teacher according to baseline EI, while the active control received the same library without network information. EMA continued for 8 weeks; Bayesian multilevel models, permutation-based mediation, and benchmarking analyses were applied. EI-based targeting produced larger reductions in the composite EI-change index than active control (mean difference 0.11, 95% credible interval 0.08 to 0.14) and higher week 7 EMA happiness (4.4 points on a 0 to 100 scale, 95% credible interval 2.7 to 6.0), with a positive arm by week slope difference of 0.62 points per week (95% credible interval 0.39 to 0.85). Model-based mediation estimates were consistent with approximately one half of the happiness difference being statistically associated with change in the composite EI-change index (average conditional mediation estimate 3.5 points, 95% credible interval 2.0 to 5.2). Benchmarking showed smaller gains under severity, threshold, or group-level centrality rules. Effects were stronger among teachers higher in conscientiousness. The findings indicate that integrating EMA, network modelling, and EI-driven optimisation yields measurable gains beyond content-matched exposure, providing a proof of concept for district-scale precision mental health that requires prospective implementation testing. Replication in additional regions, expanded node sets, and longer follow up are warranted to assess durability and generalisability.

Female↗

PubMed2025

Genomic prediction and genome-wide association study for liver abscesses in crossbred beef cattle.

Liver abscesses are a concern in feedlot cattle, and little is known about the role of genetics in their development. This study aimed to estimate genetic parameters and to identify single-nucleotide polymorphisms (SNPs) associated with liver abscesses. Crossbred cattle representing 18 breeds in the U.S. Meat Animal Research Center Germplasm Evaluation Program were phenotyped for liver abscesses at slaughter (n = 9,044). Seventeen percent of cattle had liver abscesses. These cattle had genotypes that were imputed to sequence variant genotypes. After filtering and quality control, 340,723 SNPs were used in the analysis. Liver abscess prevalence was modeled with a single-step genomic best linear unbiased prediction (ssGBLUP) threshold model using a Bayesian framework. The model included contemporary group (sex, treatment group, and slaughter date), additive genomic, and residual effects. Genomic heritability was 0.039 (95% highest posterior density = 0.005, 0.081), which was very small. To assess prediction quality, a 5-fold random cross-validation structure was used. Method Linear Regression was used to assess accuracy, bias, and dispersion by comparing estimated breeding values (EBV) from full and reduced analyses. Cross-validation metrics showed EBV based on genotypes had 0.05 reliability (SD < 0.01) with no bias relative to EBV based on genotypes and phenotypes. For the genome-wide association study, SNP effects were back calculated from the EBV solutions from ssGBLUP. No SNPs were associated with liver abscesses at a Benjamini-Hochberg adjusted 0.05 significance level. Although a large dataset was used, this result was because of the low genomic heritability and imprecise EBV used to calculate SNP effects. Based on these results, environmental factors contribute to most of the variation in liver abscesses. Genetic selection to reduce liver abscesses would be slow because of the low genomic heritability, measurement late in life, and inability to measure breeding animals. A faster approach would be finding additional environmental interventions that maintain animal performance.

Animals↗

PubMed2026

Multi-omics analysis identifies key genes and functional loci affecting teat number in American Large White and Landrace pigs and their application in optimizing genomic selection models.

BACKGROUND: Teat number is a crucial economic trait in pigs. It directly affects the ability of sows to lactate, which in turn influences the survival and health of piglets. The teat number of French Large White pigs is close to 16, while the teat number of American Large White and Landrace pigs is about 14. In order to improve the teat number of American Landrace and Large White pigs through molecular approaches and precise breeding techniques, we genotyped 2,131 American Landrace and 4,564 American Large White with teat number phenotype using a 50 K SNP chip. Then, the SNP-chip data was imputed to the level of whole-genome sequencing (iWGS). Based on iWGS data, we conducted GWAS to identify novel, significant SNPs associated with teat number and to incorporate them into genomic selection. RESULTS: In Landrace pigs, significant SNPs for TTN mapped to SSC2, SSC7, SSC8, and SSC14; the SSC8 and SSC14 effects are novel. LTN mapped to SSC7, RTN to SSC7 and SSC8. The lead SSC7 SNP explained 2.60% of TTN phenotypic variance. In Large White pigs, significant SNPs were detected on SSC7 and SSC10 for TTN; SSC7, SSC10, and SSC12 for LTN; and SSC7 and SSC10 for RTN. The most significant locus on SSC7 accounted for 2.99% of the phenotypic variance in TTN. Additionally, a multi-population meta-analysis detected significant novel SNPs for LTN on SSC1 and SSC8. By utilizing Bayesian fine mapping, the most precise QTL confidence interval on SSC7 for both TTN and RTN in Large White pigs was reduced to 40 kb. By integrating functional gene annotation with RNA-seq and ATAC-seq data from Erhualian and Bamaxiang pigs mammary placodes at embryonic day 26, we prioritized PTPN13, TRPV3, ZDHHC13, and BRD2 as novel candidate genes for teat number. We then incorporated the significant SNPs to GBLUP and benchmarked genomic-selection accuracy. In both breeds, fitting the top SNP as fixed maximized prediction for TTN and RTN, whereas treating all significant loci as an additional random effect optimized LTN. CONCLUSIONS: Our findings provide a theoretical basis for dissecting new key genes affecting teat number and for advancing molecular breeding of teat number in pigs.

Animals↗

PubMed2026

Genome-wide association studies for feed efficiency, production and feeding behavior traits in Canadian purebred Duroc pigs.

This study aimed to identify potential genetic variants and candidate genes associated with feed efficiency (FE), production, and feeding behavior traits in Canadian purebred Duroc pigs. Genome-wide association studies (GWAS) were conducted using 8,861 individuals and an imputed Affymetrix PigGen Canada 50K panel v2.0 using a linear mixed model (LMM) and a Bayesian B model. This analysis used an adjusted P-value threshold (ranging from 6.6 × 10-5 to 1.3 × 10-4) using a false-discovery rate to determine significance. The number of significant SNPs identified for each trait was as follows: average daily gain (ADG, 48), daily feed intake (DFI, 85), feed conversion ratio (FCR, 101), residual feed intake (RFI, 37), residual gain (RG, 64), residual intake and gain (RIG, 55), backfat thickness (BF, 100), loin depth (LD, 6), Kleiber's ratio (KR, 0), total time spent eating per day (TPD, 7), and number of visits to the feeder per day (NVD, 6). Several traits (BF, DFI, FCR, RFI, RG, and RIG) showed strong overlapping signals on chromosomes 7 and 10 with 24 shared significant SNPs, indicating potential shared genetic mechanisms. These traits also had 71 overlapping candidate genes, such as PACSIN1, PTCH1, ADIPOR1, and ITPR3, associated with glucose, lipid, and cholesterol metabolism. Well-known candidate genes in literature associated with growth and fatness such as MC4R and CDH20 were also identified to be associated with ADG, BF, FCR, and DFI in this study. Gene ontology enrichment analysis revealed that a set of the candidate genes were involved in the gonadotropin-releasing hormone (GnRH) and the platelet-derived growth factor (PDGF) signaling pathways. Overall, this study contributed to understanding the genetic architecture and provided a biological foundation for improving FE, production, and feeding behavior traits in Canadian Duroc pigs, facilitating the selection of more efficient pigs.

Sus scrofa↗

PubMed2022

Comparison on Major Gene Mutations Related to Rifampicin and Isoniazid Resistance between Beijing and Non-Beijing Strains of Mycobacterium tuberculosis: A Systematic Review and Bayesian Meta-Analysis.

Objective: The Beijing strain of Mycobacterium tuberculosis (MTB) is controversially presented as the predominant genotype and is more drug resistant to rifampicin and isoniazid compared to the non-Beijing strain. We aimed to compare the major gene mutations related to rifampicin and isoniazid drug resistance between Beijing and non-Beijing genotypes, and to extract the best evidence using the evidence-based methods for improving the service of TB control programs based on genetics of MTB. Method: Literature was searched in Google Scholar, PubMed and CNKI Database. Data analysis was conducted in R software. The conventional and Bayesian random-effects models were employed for meta-analysis, combining the examinations of publication bias and sensitivity. Results: Of the 8785 strains in the pooled studies, 5225 were identified as Beijing strains and 3560 as non-Beijing strains. The maximum and minimum strain sizes were 876 and 55, respectively. The mutations prevalence of rpoB, katG, inhA and oxyR-ahpC in Beijing strains was 52.40% (2738/5225), 57.88% (2781/4805), 12.75% (454/3562) and 6.26% (108/1724), respectively, and that in non-Beijing strains was 26.12% (930/3560), 28.65% (834/2911), 10.67% (157/1472) and 7.21% (33/458), separately. The pooled posterior value of OR for the mutations of rpoB was 2.72 ((95% confidence interval (CI): 1.90, 3.94) times higher in Beijing than in non-Beijing strains. That value for katG was 3.22 (95% CI: 2.12, 4.90) times. The estimate for inhA was 1.41 (95% CI: 0.97, 2.08) times higher in the non-Beijing than in Beijing strains. That for oxyR-ahpC was 1.46 (95% CI: 0.87, 2.48) times. The principal patterns of the variants for the mutations of the four genes were rpoB S531L, katG S315T, inhA-15C > T and oxyR-ahpC intergenic region. Conclusion: The mutations in rpoB and katG genes in Beijing are significantly more common than that in non-Beijing strains of MTB. We do not have sufficient evidence to support that the prevalence of mutations of inhA and oxyR-ahpC is higher in non-Beijing than in Beijing strains, which provides a reference basis for clinical medication selection.

Isoniazid↗

PubMed2025

The association between milk fat intake and atopic dermatitis: A study based on NHANES from 1999 to 2006 and Mendelian randomization.

Atopic dermatitis (AD) is a prevalent chronic inflammatory skin disease imposing significant global burden. While dietary factors are implicated in AD, the relationship between milk fat intake and AD risk remains unclear, particularly regarding optimal fat levels. This study aimed to investigate the association between milk fat intake and AD risk in adults. Relevant data (included a total of 9760 participants) from National Health and Nutrition Examination Survey between 1999 and 2006 were selected, and the relationship between milk fat intake and AD was assessed using weighted multifactorial logistic regression. Subsequently, a 2-sample Mendelian randomization (MR) study was conducted using the summary statistics of genome-wide association studies, and the causal relationship between the 2 was verified through inverse variance weighting, Bayesian weighted MR, and other supplementary MR methods. Weighted multifactorial logistic regression analysis adjusted for other covariates showed that, compared with the intake of full-fat milk, the intake of 1% fat milk (M3: odds ratio [OR]: 1.476, 95% confidence interval [CI]: 1.157-1.874, P = .005), nonfat milk (M3: OR: 1.578, 95% CI: 1.288-1.930, P < .001), as well as for milk abstainers (M3: OR: 1.303, 95% CI: 1.061-1.600, P = .025) increased the risk of AD. MR analysis further validated a significant inverse association between milk fat intake and AD risk, with both primary methods demonstrating statistical significance (P < .05) and no significant pleiotropy or heterogeneity detected in sensitivity analyses. Compared with the population consuming full-fat milk, the risk of AD may be higher in American adults consuming 1% fat milk, nonfat milk, and milk abstainers.

Humans↗

PubMed2025

Navigating Sampling Bias in Discrete Phylogeographic Analysis: Assessing the Performance of an Adjusted Bayes Factor.

Bayesian phylogeographic inference is widely used in molecular epidemiological studies to reconstruct the dispersal history of pathogens. Discrete phylogeographic analysis treats geographic locations as discrete traits and infers lineage transition events among them, and is typically followed by a Bayes factor (BF) test to assess the statistical support. In the standard BF (BFstd) test, the relative abundance of the involved trait states is not considered, which can be problematic in the case of unbalanced sampling. Existing methods to correct sampling bias in discrete phylogeographic analyses using continuous-time Markov chain (CTMC) model, often require additional epidemiological information to balance the sampling effort among locations. As such data is not necessarily available, alternative approaches that rely solely on available genomic data are needed. In this perspective, we assess the performance of a modification of the BFstd, the adjusted Bayes factor (BFadj), which incorporates information on the relative abundance of samples by location when inferring support for transition events and root location inference without requiring additional data. Using a simulation framework, we assess the statistical performance of BFstd and BFadj under varying levels of sampling bias, estimating their type I and type II error rates. Our results show that BFadj complements the BFstd by reducing type I errors at the cost increasing type II errors for inferred transition events, while improving type I and type II errors in root location inference. Our findings provide guidelines for implementing the complementary BFadj to detect and mitigate sampling bias in discrete phylogeographic inference using CTMC modeling.

Bayes Theorem↗