Generative Flow Networks (GFlowNets) offer a powerful framework for sampling graphs in proportion to their rewards. However, existing approaches suffer from systematic biases due to inaccuracies in state transition probability computations. These biases, rooted in the inherent symmetries of graphs, impact both atom-based and fragment-based generation schemes. To address this challenge, we introduce Symmetry-Aware GFlowNets (SA-GFN), a method that incorporates symmetry corrections into the learning process through reward scaling. By integrating bias correction directly into the reward structure, SA-GFN eliminates the need for explicit state transition computations. Empirical results show that SA-GFN enables unbiased sampling while enhancing diversity and consistently generating high-reward graphs that closely match the target distribution.

SUMMARY: Polygenic risk scores (PRSs) are essential tools for predicting individual phenotypic risk but often lack accuracy in non-European ancestry groups. Transfer Learning for Polygenic Risk Scores (TL-PRS) addresses this challenge by leveraging European PRSs to improve prediction in underrepresented ancestries but requires privacy-sensitive individual-level data and has low computational efficiency. Therefore, we introduce Pseudovalidated Transfer Learning for PRS (PTL-PRS), an extension of TL-PRS that incorporates pseudovalidation to eliminate the need for individual-level data and includes further software optimization. For pseudovalidation, PTL-PRS generates pseudo-summary statistics for training and validation and evaluates model performance with the pseudo-R2 metric. To improve computational efficiency, PTL-PRS software was optimized with C++, blockwise early stopping, and direct genotype retrieval. Overall, PTL-PRS enhances usability while maintaining TL-PRS's predictive performance. AVAILABILITY AND IMPLEMENTATION: The PTL.PRS R package is publicly available on GitHub at https://github.com/bokeumcho/PTL.PRS. The summary statistics used in this paper are available in the public domain: UK Biobank (https://pheweb.org/UKB-TOPMed), PGS Catalog (https://www.pgscatalog.org), COVID-19 Host Genetics Initiative (https://www.covid19hg.org) and GenOMICC (https://genomicc.org/data).

SUMMARY: The polygenic risk score (PRS) is an important method for assessing genetic susceptibility to diseases; however, its clinical utility is limited by a lack of interpretability tools. To address this problem, we introduce eXplainable PRS (XPRS), an interpretation and visualization tool that decomposes PRSs into genes/regions and single nucleotide polymorphism (SNP) contribution scores via Shapley additive explanations (SHAPs), which provide insights into specific genes and SNPs that significantly contribute to the PRS of an individual. This software features a multilevel visualization approach, including Manhattan plots, LocusZoom-like plots, and tables at the population and individual levels, to highlight important genes and SNPs. By implementing with a user-friendly web interface, XPRS allows for straightforward data input and interpretation. By bridging the gap between complex genetic data and actionable clinical insights, XPRS can improve communication between clinicians and patients. AVAILABILITY AND IMPLEMENTATION: The XPRS software is publicly available on GitHub at https://github.com/nayeonkim93/XPRS and can see the demo through our cloud-based web service at https://xprs.leelabsg.org/.

BACKGROUND: Genetic variants can contribute differently to trait heritability by their functional categories, and recent studies have shown that incorporating functional annotation can improve the predictive performance of polygenic risk scores (PRSs). In addition, when only a small proportion of variants are causal variants, PRS methods that employ a Bayesian framework with shrinkage can account for such sparsity. It is possible that the annotation group level effect is also sparse. However, the number of PRS methods that incorporate both annotation information and shrinkage on effect sizes is limited. We propose a PRS method, PRSbils, which utilizes the functional annotation information with a bilevel continuous shrinkage prior to accommodate the varying genetic architectures both on the variant-specific level and on the functional annotation level. RESULTS: We conducted simulation studies and investigated the predictive performance in settings with different genetic architectures. Results indicated that when there was a relatively large variability of group-wise heritability contribution, the gain in prediction performance from the proposed method was on average 8.0% higher AUC compared to the benchmark method PRS-CS. The proposed method also yielded higher predictive performance compared to PRS-CS in settings with different overlapping patterns of annotation groups and obtained on average 6.4% higher AUC. We applied PRSbils to binary and quantitative traits in three real world data sources (the UK Biobank, the Michigan Genomics Initiative (MGI), and the Korean Genome and Epidemiology Study (KoGES)), and two sources of annotations: ANNOVAR, and pathway information from the Kyoto Encyclopedia of Genes and Genomes (KEGG), and demonstrated that the proposed method holds the potential for improving predictive performance by incorporating functional annotations. CONCLUSIONS: By utilizing a bilevel shrinkage framework, PRSbils enables the incorporation of both overlapping and non-overlapping annotations into PRS construction to improve the performance of genetic risk prediction. The software is available at https://github.com/styvon/PRSbils .

OBJECTIVES: To develop recommendations regarding the use of weights to reduce selection bias for commonly performed analyses using electronic health record (EHR)-linked biobank data. MATERIALS AND METHODS: We mapped diagnosis (ICD code) data to standardized phecodes from 3 EHR-linked biobanks with varying recruitment strategies: All of Us (AOU; n = 244 071), Michigan Genomics Initiative (MGI; n = 81 243), and UK Biobank (UKB; n = 401 167). Using 2019 National Health Interview Survey data, we constructed selection weights for AOU and MGI to represent the US adult population more. We used weights previously developed for UKB to represent the UKB-eligible population. We conducted 4 common analyses comparing unweighted and weighted results. RESULTS: For AOU and MGI, estimated phecode prevalences decreased after weighting (weighted-unweighted median phecode prevalence ratio [MPR]: 0.82 and 0.61), while UKB estimates increased (MPR: 1.06). Weighting minimally impacted latent phenome dimensionality estimation. Comparing weighted versus unweighted phenome-wide association study for colorectal cancer, the strongest associations remained unaltered, with considerable overlap in significant hits. Weighting affected the estimated log-odds ratio for sex and colorectal cancer to align more closely with national registry-based estimates. DISCUSSION: Weighting had a limited impact on dimensionality estimation and large-scale hypothesis testing but impacted prevalence and association estimation. When interested in estimating effect size, specific signals from untargeted association analyses should be followed up by weighted analysis. CONCLUSION: EHR-linked biobanks should report recruitment and selection mechanisms and provide selection weights with defined target populations. Researchers should consider their intended estimands, specify source and target populations, and weight EHR-linked biobank analyses accordingly.

MOTIVATION: In the genome-wide association analysis of population-based biobanks, most diseases have low prevalence, which results in low detection power. One approach to tackle the problem is using family disease history, yet existing methods are unable to address type I error inflation induced by increased correlation of phenotypes among closely related samples, as well as unbalanced phenotypic distribution. RESULTS: We propose a new method for genetic association test with family disease history, mixed-model-based Test with Adjusted Phenotype and Empirical saddlepoint approximation, which controls for increased phenotype correlation by adopting a two-variance-component mixed model, accounts for case-control imbalance by using empirical saddlepoint approximation, and is flexible to incorporate any existing adjusted phenotypes, such as phenotypes from the LT-FH method. We show through simulation studies and analysis of UK Biobank data of white British samples and the Korean Genome and Epidemiology Study of Korean samples that the proposed method is robust and yields better calibration compared to existing methods while gaining power for detection of variant-phenotype associations. AVAILABILITY AND IMPLEMENTATION: The summary statistics and code generated in this study are available at https://github.com/styvon/TAPE. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Population stratification (PS) is a major confounder in genome-wide association studies (GWAS) and can lead to false-positive associations. To adjust for PS, principal component analysis (PCA)-based ancestry prediction has been widely used. Simple projection (SP) based on principal component loadings and the recently developed data augmentation, decomposition and Procrustes (ADP) transformation, such as LASER and TRACE, are popular methods for predicting PC scores. However, the predicted PC scores from SP can be biased toward NULL. On the other hand, ADP has a high computation cost because it requires running PCA separately for each study sample on the augmented dataset. RESULTS: We develop and propose two alternative approaches: bias-adjusted projection (AP) and online ADP (OADP). Using random matrix theory, AP asymptotically estimates and adjusts for the bias of SP. OADP uses a computationally efficient online singular value decomposition algorithm, which can greatly reduce the computation cost of ADP. We carried out extensive simulation studies to show that these alternative approaches are unbiased and the computation speed can be 16-16 000 times faster than ADP. We applied our approaches to the UK Biobank data of 488 366 study samples with 2492 samples from the 1000 Genomes data as the reference. AP and OADP required 0.82 and 21 CPU hours, respectively, while the projected computation time of ADP was 1628 CPU hours. Furthermore, when inferring sub-European ancestry, SP clearly showed bias, unlike the proposed approaches. AVAILABILITY AND IMPLEMENTATION: The OADP and AP methods, as well as SP and ADP, have been implemented in the open-source Python software FRAPOSA, available at github.com/daviddaiweizhang/fraposa. CONTACT: [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The objective of this article is to introduce valid and robust methods for the analysis of rare variants for family-based exome chips, whole-exome sequencing or whole-genome sequencing data. Family-based designs provide unique opportunities to detect genetic variants that complement studies of unrelated individuals. Currently, limited methods and software tools have been developed to assist family-based association studies with rare variants, especially for analyzing binary traits. In this article, we address this gap by extending existing burden and kernel-based gene set association tests for population data to related samples, with a particular emphasis on binary phenotypes. The proposed approach blends the strengths of kernel machine methods and generalized estimating equations. Importantly, the efficient generalized kernel score test can be applied as a mega-analysis framework to combine studies with different designs. We illustrate the application of the proposed method using data from an exome sequencing study of autism. Methods discussed in this article are implemented in an R package 'gskat', which is available on CRAN and GitHub.