Genome-wide Association Studies Hits Research Articles

Abstract PO5-09-03: Multi-tissue transcriptome-wide association studies identified genes for intrinsic subtypes of breast cancer

Abstract Background: Though several genome-wide association studies (GWAS) of breast cancer (BC) have identified common variants which differ between intrinsic subtypes, the genes through which these variants act through to impact BC risk have not been fully established. Furthermore, transcriptome-wide association studies (TWAS) have thus far been primarily employed to identify genes associated with overall BC risk, while overlooking how the influence of common variation on gene expression may contribute to subtype-specific differences. Methods: In this study, we performed two complementary multi-tissue TWASs for each of the following BC intrinsic subtypes: Luminal A-like, Luminal B-like, Luminal B/HER2-negative-like, HER2-enriched-like, Triple Negative BC. These two approaches included 1) an expression-based approach that collated TWAS signals from expression quantitative trait loci (eQTLs) across multiple tissues using the Aggregated Cauchy Association Test (ACAT) and 2) a splicing-based approach that utilized two applications of ACAT to collate splicing QTLs (sQTLs) for a given gene in a tissue and then across tissues. To perform these two TWASs, we utilized e/sQTL models trained in 11 tissues from the Genotype-Tissue Expression Project including breast, ovary, uterus, vagina, EBV-transformed lymphocytes, whole blood, spleen, liver, subcutaneous adipose, visceral adipose, and cell-cultured fibroblasts. GWAS summary statistics were previously generated from 133,384 BC cases and 113,789 controls who were participants in the Breast Cancer Association Consortium (BCAC). We further performed our TWAS while conditioning e/sQTL effect sizes on nearby GWAS index SNPs. Additionally, we utilized gene-based fine-mapping of eQTLs and sQTL to identify candidate causal genes for each intrinsic subtype. Results: Overall, we identified 164 genes in 69 loci that were associated with Luminal A-like, 19 genes in 9 loci with Luminal B-like, 18 genes in 11 loci with Luminal B/HER2-negative-like, 10 genes in 7 loci with HER2-enriched-like, and 29 genes in 12 loci with TNBC. Among these genes, 17 genes had not been reported in previous TWAS of BC, and 140 genes, 1 gene, 2 genes, 2 genes, and 16 genes were uniquely associated with each of the intrinsic subtypes, respectively. Additionally, we identified one gene associated with Luminal A-like and one gene with Luminal B-like BC that were each in a locus located at least 1.4 Mb from published GWAS hits. Furthermore, we identified 106, 11, 10, 5, and 21 candidate causal genes for each of the intrinsic subtypes, respectively, that had a posterior inclusion probability of 0.9 in at least one e/sQTL. Conclusion: In summary, our multi-tissue TWAS corroborated previous GWAS loci for overall BC risk and intrinsic subtypes, while underscoring how common variation impacts BC etiology by modulating the expression and splicing of genes in multiple tissue types. Citation Format: James Li, Julian McClellan, Haoyu Zhang, Guimin Gao, Dezheng Huo. Multi-tissue transcriptome-wide association studies identified genes for intrinsic subtypes of breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-09-03.

Genetic regulatory effects in response to a high-cholesterol, high-fat diet in baboons

Steady-state expression quantitative trait loci (eQTLs) explain only a fraction of disease-associated loci identified through genome-wide association studies (GWASs), while eQTLs involved in gene-by-environment (GxE) interactions have rarely been characterized in humans due to experimental challenges. Using a baboon model, we found hundreds of eQTLs that emerge in adipose, liver, and muscle after prolonged exposure to high dietary fat and cholesterol. Diet-responsive eQTLs exhibit genomic localization and genic features that are distinct from steady-state eQTLs. Furthermore, the human orthologs associated with diet-responsive eQTLs are enriched for GWAS genes associated with human metabolic traits, suggesting that context-responsive eQTLs with more complex regulatory effects are likely to explain GWAS hits that do not seem to overlap with standard eQTLs. Our results highlight the complexity of genetic regulatory effects and the potential of eQTLs with disease-relevant GxE interactions in enhancing the understanding of GWAS signals for human complex disease using non-human primate models.

Machine Learning Models of Polygenic Risk for Enhanced Prediction of Alzheimer Disease Endophenotypes.

Alzheimer disease (AD) has a polygenic architecture, for which genome-wide association studies (GWAS) have helped elucidate sequence variants (SVs) influencing susceptibility. Polygenic risk score (PRS) approaches show promise for generating summary measures of inherited risk for clinical AD based on the effects of APOE and other GWAS hits. However, existing PRS approaches, based on traditional regression models, explain only modest variation in AD dementia risk and AD-related endophenotypes. We hypothesized that machine learning (ML) models of polygenic risk (ML-PRS) could outperform standard regression-based PRS methods and therefore have the potential for greater clinical utility. We analyzed combined data from the Mayo Clinic Study of Aging (n = 1,791) and the Alzheimer's Disease Neuroimaging Initiative (n = 864). An AD PRS was computed for each participant using the top common SVs obtained from a large AD dementia GWAS. In parallel, ML models were trained using those SV genotypes, with amyloid PET burden as the primary outcome. Secondary outcomes included amyloid PET positivity and clinical diagnosis (cognitively unimpaired vs impaired). We compared performance between ML-PRS and standard PRS across 100 training sessions with different data splits. In each session, data were split into 80% training and 20% testing, and then five-fold cross-validation was used within the training set to ensure the best model was produced for testing. We also applied permutation importance techniques to assess which genetic factors contributed most to outcome prediction. ML-PRS models outperformed the AD PRS (r2 = 0.28 vs r2 = 0.24 in test set) in explaining variation in amyloid PET burden. Among ML approaches, methods accounting for nonlinear genetic influences were superior to linear methods. ML-PRS models were also more accurate when predicting amyloid PET positivity (area under the curve [AUC] = 0.80 vs AUC = 0.63) and the presence of cognitive impairment (AUC = 0.75 vs AUC = 0.54) compared with the standard PRS. We found that ML-PRS approaches improved upon standard PRS for prediction of AD endophenotypes, partly related to improved accounting for nonlinear effects of genetic susceptibility alleles. Further adaptations of the ML-PRS framework could help to close the gap of remaining unexplained heritability for AD and therefore facilitate more accurate presymptomatic and early-stage risk stratification for clinical decision-making.

Replication of previous autism-GWAS hits suggests the association between NAA1, SORCS3, and GSDME and autism in the Han Chinese population

BackgroundAutism is a severe neurodevelopmental disorder characterized by social interaction deficits, impairments in communication, and restricted and repetitive stereotyped behavior and activities. Family and twin studies suggested an essential role of genetic factors in the etiology of autism spectrum disorder (ASD). Also, other studies found SORCS3 and GSDME (DFNA5) might be involved in brain development and susceptible to ASD. MethodsIn this study, 17 genome-wide significant SNPs reported in previous ASD genome-wide association studies (GWAS) and 7 SNPs in strong linkage disequilibrium with known ASD GWAS hits were selected to investigate the association between these SNPs and autism in the Han Chinese population. Then, 10 tagSNPs in SORCS3 and 11 tagSNPs in GSDME were selected to analyze the association between these SNPs and autism. The selected 24 SNPs and tagSNPs were genotyped using the Agena MassARRAY SNP genotyping assay in 757 Han Chinese autism trios. ResultsRs1484144 in NAA11 was significantly associated with autism; significance remained after the Bonferroni correction (P < 0.0022). Also, rs79879286, rs12154597, and rs12540919 near GSDME, as well as rs9787523 and rs3750261 in SORCS3, were nominally associated with autism. ConclusionOur study suggests that rs1484144 in NAA11 is a significant SNP for autism in the Han Chinese population, while SORCS3 and GSDME might be the susceptibility genes for autism in this population.

Charting the genetic architecture of Alzheimer’s disease across APOE*4 and sex

AbstractBackgroundAPOE*4 status and sex (and their interaction) are among the strongest risk factors for late‐onset Alzheimer’s disease (AD). We thus sought to perform large‐scale, stratified genome‐wide association study (GWAS) of AD to chart the role of common genetic variation in AD sexual dimorphism and heterogeneity of APOE*4‐related AD risk.MethodThe study overview and GWAS models are shown in Figure.1A. SNP‐array AD GWAS datasets primarily composing the ADGC, together with whole‐genome sequencing (WGS) from ADSP (NG00067.v7), provided case‐control diagnoses for phase‐1. The UK Biobank provided subjects with ICD codes and family history of AD status for phase‐2. Subjects were of European ancestry. Linear mixed model regressions were performed (LMM‐BOLT v.2.3.4) on AD outcome measures, adjusting for sex, APOE*4/APOE*2 dosage, genetic principal components, and array/batch/center. GWAS hits (P&lt;5e‐*8) were evaluated for interactions effects (meta‐regression for phase1+2 meta‐analyses) and assessed in functional follow‐up analyses. The effect of stratification on genetic risk scores (GRS) predictive performance on AD risk was also evaluated.ResultStratified AD GWAS identified 28 loci/variants, including 5 known lead variants in known AD loci, 6 novel independent variants in known AD loci, and 17 novel loci (Figure.1B). 13 out of 17 novel loci showed varying degrees of functional support, with 4 also showing evidence for QTL colocalization (Figure.2). Many implicated genes showed prior support or effects relevant to AD: MARCHF10 was recently implicated as a risk gene for all‐cause dementia; EMP2 is a regulator of cell membrane composition, cell‐matrix adhesion, and blood vessel endothelial cell migration; SLIT2 binds APP, and EYA4 and TBC1D9 have respectively been implicated in brain tau deposition and MAPT haplotype‐stratified GWAS. Nonetheless, most novel hits were low/uncommon frequency variants and QTL colocalization evidence was limited. Further, there appeared to be no predictive benefit for stratified GRS analyses (Figure.3).ConclusionWe identified novel loci/variants differentially associated with AD risk across APOE*4 and/or sex. Yet, we inferred relatively limited evidence for stratified effects (propensity for low frequency hits and no benefit for stratified GRS). Overall, our findings contribute to our understanding of the genetic etiology of AD, which in turn contributes to advance personalized genetic medicine.

Systematic differences in discovery of genetic effects on gene expression and complex traits.

Most signals in genome-wide association studies (GWAS) of complex traits implicate noncoding genetic variants with putative gene regulatory effects. However, currently identified regulatory variants, notably expression quantitative trait loci (eQTLs), explain only a small fraction of GWAS signals. Here, we show that GWAS and cis-eQTL hits are systematically different: eQTLs cluster strongly near transcription start sites, whereas GWAS hits do not. Genes near GWAS hits are enriched in key functional annotations, are under strong selective constraint and have complex regulatory landscapes across different tissue/cell types, whereas genes near eQTLs are depleted of most functional annotations, show relaxed constraint, and have simpler regulatory landscapes. We describe a model to understand these observations, including how natural selection on complex traits hinders discovery of functionally relevant eQTLs. Our results imply that GWAS and eQTL studies are systematically biased toward different types of variant, and support the use of complementary functional approaches alongside the next generation of eQTL studies.

Persistent homology analysis of type 2 diabetes genome-wide association studies in protein-protein interaction networks.

Genome-wide association studies (GWAS) involving increasing sample sizes have identified hundreds of genetic variants associated with complex diseases, such as type 2 diabetes (T2D); however, it is unclear how GWAS hits form unique topological structures in protein-protein interaction (PPI) networks. Using persistent homology, this study explores the evolution and persistence of the topological features of T2D GWAS hits in the PPI network with increasing p-value thresholds. We define an n-dimensional persistent disease module as a higher-order generalization of the largest connected component (LCC). The 0-dimensional persistent T2D disease module is the LCC of the T2D GWAS hits, which is significantly detected in the PPI network (196 nodes and 235 edges, P0.05). In the 1-dimensional homology group analysis, all 18 1-dimensional holes (loops) of the T2D GWAS hits persist over all p-value thresholds. The 1-dimensional persistent T2D disease module comprising these 18 persistent 1-dimensional holes is significantly larger than that expected by chance (59 nodes and 83 edges, P0.001), indicating a significant topological structure in the PPI network. Our computational topology framework potentially possesses broad applicability to other complex phenotypes in identifying topological features that play an important role in disease pathobiology.

Alzheimer’s disease-associated complement gene variants influence plasma complement protein levels

BackgroundAlzheimer’s disease (AD) has been associated with immune dysregulation in biomarker and genome-wide association studies (GWAS). GWAS hits include the genes encoding complement regulators clusterin (CLU) and complement receptor 1 (CR1), recognised as key players in AD pathology, and complement proteins have been proposed as biomarkers.Main bodyTo address whether changes in plasma complement protein levels in AD relate to AD-associated complement gene variants we first measured relevant plasma complement proteins (clusterin, C1q, C1s, CR1, factor H) in a large cohort comprising early onset AD (EOAD; n = 912), late onset AD (LOAD; n = 492) and control (n = 504) donors. Clusterin and C1q were significantly increased (p < 0.001) and sCR1 and factor H reduced (p < 0.01) in AD plasma versus controls. ROC analyses were performed to assess utility of the measured complement biomarkers, alone or in combination with amyloid beta, in predicting AD. C1q was the most predictive single complement biomarker (AUC 0.655 LOAD, 0.601 EOAD); combining C1q with other complement or neurodegeneration makers through stepAIC-informed models improved predictive values slightly. Effects of GWS SNPs (rs6656401, rs6691117 in CR1; rs11136000, rs9331888 in CLU; rs3919533 in C1S) on protein concentrations were assessed by comparing protein levels in carriers of the minor vs major allele. To identify new associations between SNPs and changes in plasma protein levels, we performed a GWAS combining genotyping data in the cohort with complement protein levels as endophenotype. SNPs in CR1 (rs6656401), C1S (rs3919533) and CFH (rs6664877) reached significance and influenced plasma levels of the corresponding protein, whereas SNPs in CLU did not influence clusterin levels.ConclusionComplement dysregulation is evident in AD and may contribute to pathology. AD-associated SNPs in CR1, C1S and CFH impact plasma levels of the encoded proteins, suggesting a mechanism for impact on disease risk.

Association of Polygenic Score With Tumor Molecular Subtypes in Papillary Thyroid Carcinoma.

Genome-wide association studies have identified germline variants associated with elevated PTC risk. It is also known that somatic driver mutations contribute to PTC development and as such PTCs can be further categorized into different molecular subtypes based on their somatic alterations. However, it remains unknown whether identified germline variants predictive of PTC risk are associated with specific molecular subtypes. The primary goal of the present study is to determine whether germline genetic risk, as assessed using a polygenic score (PGS) is associated with molecular subtypes of papillary thyroid carcinoma (PTC), defined based on tumor driver mutation status. This study was carried out using data from The Cancer Genome Atlas (TCGA) thyroid cancer study. A previously validated 10-single-nucleotide variation PGS for PTC derived from genome-wide association study hits was calculated to ascertain germline genetic risk. The primary molecular subtypes of interest were defined by tumor driver mutation status (BRAFV600E-mutated vs RAS-mutated vs "other"). We also explored associations between PGS and molecular subtypes defined by messenger RNA (mRNA) expression, microRNA expression, and DNA methylation patterns. Polytomous logistic regression analysis was used to assess the association between PGS and PTC molecular subtype with and without adjustment for clinical variables. Odds ratios (ORs) with their 95% CIs were estimated. A total of 359 patients were included in the study. PGS was significantly associated specific tumor molecular subtypes defined by tumor driver mutation status. Increasing germline risk was associated with having a higher odd of BRAFV600E-mutated PTC compared to PTCs without driver mutations in the "other" category. No significant difference was detected in terms of PGS tumor categorization in the RAS subtype compared to BRAFV600E. In exploratory analyses, PGS was also associated with mRNA-, microRNA-, and DNA methylation-defined molecular subtypes, as defined by the TCGA PTC study. PGS has molecular subtype-specific associations in PTC, which has implications for their use in risk prediction.

Multiset correlation and factor analysis enables exploration of multi-omics data

Multi-omics datasets are becoming more common, necessitating better integration methods to realize their revolutionary potential. Here, we introduce multi-set correlation and factor analysis (MCFA), an unsupervised integration method tailored to the unique challenges of high-dimensional genomics data that enables fast inference of shared and private factors. We used MCFA to integrate methylation markers, protein expression, RNA expression, and metabolite levels in 614 diverse samples from the Trans-Omics for Precision Medicine/Multi-Ethnic Study of Atherosclerosis multi-omics pilot. Samples cluster strongly by ancestry in the shared space, even in the absence of genetic information, while private spaces frequently capture dataset-specific technical variation. Finally, we integrated genetic data by conducting a genome-wide association study (GWAS) of our inferred factors, observing that several factors are enriched for GWAS hits and trans-expression quantitative trait loci. Two of these factors appear to be related to metabolic disease. Our study provides a foundation and framework for further integrative analysis of ever larger multi-modal genomic datasets.

A transcriptome association study of AD risk genes in the pathogenesis of Alzheimer’s disease

AbstractBackgroundAlzheimer’s disease (AD) affects millions of people with no effective treatments making it vital to identify potential therapeutic targets. Genome‐wide association studies (GWAS) have consistently identified AD susceptibility loci or ‘hits’. Few of these are differentially expressed in brain transcriptomic studies, meaning that their effects may be manifest via interacting or downstream transcripts. We have recently used machine learning on human brain transcriptomic data to identify and validate an association between lactoferrin and amyloid (1). Here we apply the same method to identify the transcripts associated with AD susceptibility ‘loci’.MethodAnalyses were performed on the disease‐linked genes from our machine learning analyses (LTF, ADAMTS2, PPDPF, PRTN3) and the combined 58 GWAS risk loci from two recent studies (2, 3). Boruta was used to find transcripts associated with each GWAS ‘hit’ and our genes of interest in two pathologically confirmed AD cohorts (Religious Orders Study and Memory and Aging project, Mount Sinai Brain Bank). The GWAS hits and associated transcripts were correlated with disease markers (cognition, plaque number, Braak score, CERAD score) to find direct and indirect effects.ResultsTREM2 was the GWAS ‘hit’ most correlated with disease markers in both cohorts but most GWAS ‘hits’ were either not correlated or negatively correlated with disease markers. Among the GWAS ‘hit’‐associated transcripts, SWAP70, EMP3, and WWTR1 were positively and VGF negatively correlated with disease markers in both cohorts. For our genes of interest, ADAMTS2 and PPDPF shared overlaps between their associated transcripts and that of GWAS ‘hit’ ABCA7. LTF was linked with TREM2 and associated with plaques in both cohorts.ConclusionThis study has teased apart nascent associations between known susceptibility genes and gene expression data from the AD brain, but it also hints that consideration of pathological sub‐phenotypes are required to make greater progress in understanding AD pathogenesis.1. Tsatsanis et al. The acute phase protein lactoferrin is a key feature of AD …. Mol Psychiatry. 2021;26(10):5516‐31.2. Yang et al. A human brain vascular atlas reveals diverse cell mediators of AD risk. BioRxiv [Preprint]. 2021.3. Wightman et al. A GWAS with 1,126,563 individuals identifies new risk loci for Alzheimer’s disease. Nat Genet. 2021;53(9):1276‐82.

A joint transcriptome-wide association study across multiple tissues identifies candidate breast cancer susceptibility genes

Genome-wide association studies (GWASs) have identified more than 200 genomic loci for breast cancer risk, but specific causal genes in most of these loci have not been identified. In fact, transcriptome-wide association studies (TWASs) of breast cancer performed using gene expression prediction models trained in breast tissue have yet to clearly identify most target genes. To identify candidate genes, we performed a GWAS analysis in a breast cancer dataset from UK Biobank (UKB) and combined the results with the GWAS results of the Breast Cancer Association Consortium (BCAC) by a meta-analysis. Using the summary statistics from the meta-analysis, we performed a joint TWAS analysis that combined TWAS signals from multiple tissues. We used expression prediction models trained in 11 tissues that are potentially relevant to breast cancer from the Genotype-Tissue Expression (GTEx) data. In the GWAS analysis, we identified eight loci distinct from those reported previously. In the TWAS analysis, we identified 309 genes at 108 genomic loci to be significantly associated with breast cancer at the Bonferroni threshold. Of these, 17 genes were located in eight regions that were at least 1 Mb away from published GWAS hits. The remaining TWAS-significant genes were located in 100 known genomic loci from previous GWASs of breast cancer. We found that 21 genes located in known GWAS loci remained statistically significant after conditioning on previous GWAS index variants. Our study provides insights into breast cancer genetics through mapping candidate target genes in a large proportion of known GWAS loci and discovering multiple new loci.

What does heritability of Alzheimer's disease represent?

Both late-onset Alzheimer's disease (AD) and ageing have a strong genetic component. In each case, many associated variants have been discovered, but how much missing heritability remains to be discovered is debated. Variability in the estimation of SNP-based heritability could explain the differences in reported heritability. We compute heritability in five large independent cohorts (N = 7,396, 1,566, 803, 12,528 and 3,963) to determine whether a consensus for the AD heritability estimate can be reached. These cohorts vary by sample size, age of cases and controls and phenotype definition. We compute heritability a) for all SNPs, b) excluding APOE region, c) excluding both APOE and genome-wide association study hit regions, and d) SNPs overlapping a microglia gene-set. SNP-based heritability of late onset Alzheimer's disease is between 38 and 66% when age and genetic disease architecture are correctly accounted for. The heritability estimates decrease by 12% [SD = 8%] on average when the APOE region is excluded and an additional 1% [SD = 3%] when genome-wide significant regions were removed. A microglia gene-set explains 69-84% of our estimates of SNP-based heritability using only 3% of total SNPs in all cohorts. The heritability of neurodegenerative disorders cannot be represented as a single number, because it is dependent on the ages of cases and controls. Genome-wide association studies pick up a large proportion of total AD heritability when age and genetic architecture are correctly accounted for. Around 13% of SNP-based heritability can be explained by known genetic loci and the remaining heritability likely resides around microglial related genes.

Design and quality control of large-scale two-sample Mendelian randomization studies.

Mendelian randomization (MR) studies are susceptible to metadata errors (e.g. incorrect specification of the effect allele column) and other analytical issues that can introduce substantial bias into analyses. We developed a quality control (QC) pipeline for the Fatty Acids in Cancer Mendelian Randomization Collaboration (FAMRC) that can be used to identify and correct for such errors. We collated summary association statistics from fatty acid and cancer genome-wide association studies (GWAS) and subjected the collated data to a comprehensive QC pipeline. We identified metadata errors through comparison of study-specific statistics to external reference data sets (the National Human Genome Research Institute-European Bioinformatics Institute GWAS catalogue and 1000 genome super populations) and other analytical issues through comparison of reported to expected genetic effect sizes. Comparisons were based on three sets of genetic variants: (i) GWAS hits for fatty acids, (ii) GWAS hits for cancer and (iii) a 1000 genomes reference set. We collated summary data from 6 fatty acid and 54 cancer GWAS. Metadata errors and analytical issues with the potential to introduce substantial bias were identified in seven studies (11.6%). After resolving metadata errors and analytical issues, we created a data set of 219 842 genetic associations with 90 cancer types, generated in analyses of 566 665 cancer cases and 1 622 374 controls. In this large MR collaboration, 11.6% of included studies were affected by a substantial metadata error or analytical issue. By increasing the integrity of collated summary data prior to their analysis, our protocol can be used to increase the reliability of downstream MR analyses. Our pipeline is available to other researchers via the CheckSumStats package (https://github.com/MRCIEU/CheckSumStats).

Deletion mapping of regulatory elements for GATA3 in T cells reveals a distal enhancer involved in allergic diseases

GATA3 is essential for Tcell differentiation and is surrounded by genome-wide association study (GWAS) hits for immune traits. Interpretation of these GWAS hits is challenging because gene expression quantitative trait locus (eQTL) studies lack power to detect variants with small effects on gene expression in specific cell types and the genome region containing GATA3 contains dozens of potential regulatory sequences. To map regulatory sequences for GATA3, we performed a high-throughput tiling deletion screen of a 2 Mb genome region in Jurkat Tcells. This revealed 23 candidate regulatory sequences, all but one of which is within the same topological-associating domain (TAD) as GATA3. We then performed a lower-throughput deletion screen to precisely map regulatory sequences in primary T helper 2 (Th2) cells. We tested 25 sequences with ∼100bp deletions and validated five of the strongest hits with independent deletion experiments. Additionally, we fine-mapped GWAS hits for allergic diseases in a distal regulatory element, 1 Mb downstream of GATA3, and identified 14 candidate causal variants. Small deletions spanning the candidate variant rs725861 decreased GATA3 levels in Th2 cells, and luciferase reporter assays showed regulatory differences between its two alleles, suggesting a causal mechanism for this variant in allergic diseases. Our study demonstrates the power of integrating GWAS signals with deletion mapping and identifies critical regulatory sequences for GATA3.

Embracing diversity: a genetic marker dataset with increased marker density facilitates association studies in maize

Maize is known for its phenotypic and genetic diversity. On average, two maize lines diverge from one another as much as humans do from chimpanzees (Buckler et al., 2006). This diversity is attributed to pollen flow between domesticated maize and its wild relative teosinte, as well as to trading between farmers (Hake & Ross-Ibarra, 2015). Maize diversity contributes to its adaptability to new climates and growth conditions so that, currently, maize is grown across a wider area than any other crop (Hake & Ross-Ibarra, 2015). Analysing maize populations based on DNA sequence polymorphisms (markers) is the premise for identifying selection targets and understanding geographic relations. Many studies (Brandenburg et al., 2017; Chia et al., 2012) have revealed changes in genetic diversity by re-sequencing maize lines with differing sequencing depth, and with samples from across America and Europe. However, comparing markers between datasets of different studies can be challenging: datasets can differ in allele frequencies, in the single nucleotide polymorphism (SNP)-calling pipelines, or in stochastic distributions of read depths, and therefore might identify different markers even in the same genomic region. Therefore, Marcin Grzybowski, James Schnable and team set out to unify previous datasets as well as incorporate newly re-sequenced samples in one dataset (Grzybowski et al., 2023). Schnable obtained his PhD from the University of California–Berkeley, working on plant comparative genomics. During doing lab rotations, he got hooked on computational biology because this approach leads to much faster results than wet-lab experiments. After a short postdoc at the Chinese Academy of Agricultural Sciences in Beijing, he was hired as a Professor at the University of Nebraska–Lincoln to work on computational biology. He soon got involved in a new plant phenotyping platform being started there, and brought together teams of biologists, computer scientists and statisticians to work with the platform. “It's been a lot of fun”, he says! For the diversity dataset, Grzybowski and colleagues used whole genome resequencing data from 1515 maize samples, comprising lines from the Wisconsin Diversity panel (Hansey et al., 2011), inbred lines from Poland, as well as wild relatives, tropical landraces, archaeological samples and modern open-pollinated varieties. Overall, these samples originated in or were developed over six continents (Figure 1a). The sequence data were aligned to the maize reference genome, and over 350 million potential DNA sequence polymorphisms were identified. The dataset is therefore much bigger than the approximately 83 million variants identified in the maize HapMap3 project, which included over 1200 maize accessions, but used a lower sequencing depth (Bukowski et al., 2018). Second-stage quality filtering of their new dataset reduced the number of variants to approximately 46 million higher confidence variants. Marker dataset for global maize diversity can be used to identify new genes associated with a trait. (a) Geographical distribution of the countries of origin for the 1515 maize individuals used in the study. (b) Association test using the marker set defined in this study identified MADS69 and ZCN8 based on female flowering data (days to silking). Adapted from Grzybowski et al. (2023). Population genetic analyses are not only based on DNA sequence diversity, but also require the analysis of phenotypic traits. However, comparing phenotypes across different environments adds more variance and thereby reduces the statistical power to link genotype and phenotype. Comparing genotypes within the same environment is desirable, but not all genotypes are adapted to the same environments and can complete their life cycle. To tackle this problem, researchers use association panels, which maximize genetic diversity by selecting genotypes adapted to a specific environment. Grzybowski and colleagues used their marker set to analyse the Wisconsin Diversity Panel (Hansey et al., 2011) and found that the lines retained over 70% DNA sequence variation compared to the wild relative Zea mays ssp. parviglumis, suggesting that there is still a lot of genetic variation in this diversity panel. Therefore, the marker dataset can serve as a resource for other researchers to calculate accurate values of DNA sequence diversity for their populations. To analyse the impact of the high marker density on the outcomes for genome-wide association studies (GWAS), Grzybowski and colleagues used a published set of female flowering data generated from temperate-adapted maize inbreds (Mural et al., 2022). A previous GWAS using around 400 000 RNA-sequencing-based markers identified the flowering time gene MADS69 (Mural et al., 2022). With the newly generated marker set, Grzybowski and colleagues identified both MADS69 and a new locus, ZCN8, a gene that contributes (Guo et al., 2018) to maize adaption to temperate climates (Figure 1b). Grzybowski and colleagues speculate that ZCN8 was previously not discovered because the dataset used RNA-sequencing-based genetic markers and therefore missed significant SNPs in the intergenic space. Grzybowski and colleagues are hopeful that the higher density of markers will also help to achieve more precise localization of the causal variants associated with specific GWAS hits. Because of the diversity range of the maize lines used, this dataset can also be used to detect selection patterns in the genome associated with traits of interest; for example, those related to domestication, adaptation to the environment or genetic improvement during breeding. In the case of MADS69, Grzybowski and colleagues found less DNA sequence diversity in the promoter in tropical and temperate maize lines than in the wild relative. For ZCN8, they confirmed previous studies that had shown less DNA sequence diversity in domesticated maize than in teosinte (Guo et al., 2018) and, additionally, they showed that the decline was biggest between temperate and tropical domesticated maize populations, whereas tropical maize diversity was similar to that of teosinte, suggesting that ZCN8 was selected during adaptation to temperate conditions rather than during domestication. Grzybowski, who is currently a senior researcher at the University of Warsaw, hopes to use the dataset to identify genes underlying maize adaptation to cold and understand their evolutionary history, a topic that relates back to his PhD studies on maize adaptations to cold spring conditions. Including global maize diversity in the dataset, instead of focusing on a single GWAS population adapted to a single environment, will make it a lot easier for researchers to track the evolutionary histories and impact of selection on functional variants once they have been identified in a GWAS study.

Complement receptor 1 is expressed on brain cells and in the human brain.

Genome wide association studies (GWAS) have highlighted the importance of the complement cascade in pathogenesis of Alzheimer's disease (AD). Complement receptor 1 (CR1; CD35) is among the top GWAS hits. The long variant of CR1 is associated with increased risk for AD; however, roles of CR1 in brain health and disease are poorly understood. A critical confounder is that brain expression of CR1 is controversial; failure to demonstrate brain expression has provoked the suggestion that peripherally expressed CR1 influences AD risk. We took a multi-pronged approach to establish whether CR1 is expressed in brain. Expression of CR1 at the protein and mRNA level was assessed in human microglial lines, induced pluripotent stem cell (iPSC)-derived microglia from two sources and brain tissue from AD and control donors. CR1 protein was detected in microglial lines and iPSC-derived microglia expressing different CR1 variants when immunostained with a validated panel of CR1-specific antibodies; cell extracts were positive for CR1 protein and mRNA. CR1 protein was detected in control and AD brains, co-localizing with astrocytes and microglia, and expression was significantly increased in AD compared to controls. CR1 mRNA expression was detected in all AD and control brain samples tested; expression was significantly increased in AD. The data unequivocally demonstrate that the CR1 transcript and protein are expressed in human microglia ex vivo and on microglia and astrocytes in situ in the human brain; the findings support the hypothesis that CR1 variants affect AD risk by directly impacting glial functions.

PALM: a powerful and adaptive latent model for prioritizing risk variants with functional annotations.

The findings from genome-wide association studies (GWASs) have greatly helped us to understand the genetic basis of human complex traits and diseases. Despite the tremendous progress, much effects are still needed to address several major challenges arising in GWAS. First, most GWAS hits are located in the non-coding region of human genome, and thus their biological functions largely remain unknown. Second, due to the polygenicity of human complex traits and diseases, many genetic risk variants with weak or moderate effects have not been identified yet. To address the above challenges, we propose a powerful and adaptive latent model (PALM) to integrate cell-type/tissue-specific functional annotations with GWAS summary statistics. Unlike existing methods, which are mainly based on linear models, PALM leverages a tree ensemble to adaptively characterize non-linear relationship between functional annotations and the association status of genetic variants. To make PALM scalable to millions of variants and hundreds of functional annotations, we develop a functional gradient-based expectation-maximization algorithm, to fit the tree-based non-linear model in a stable manner. Through comprehensive simulation studies, we show that PALM not only controls false discovery rate well, but also improves statistical power of identifying risk variants. We also apply PALM to integrate summary statistics of 30 GWASs with 127 cell type/tissue-specific functional annotations. The results indicate that PALM can identify more risk variants as well as rank the importance of functional annotations, yielding better interpretation of GWAS results. The source code is available at https://github.com/YangLabHKUST/PALM. Supplementary data are available at Bioinformatics online.

HTRX: an R package for learning non-contiguous haplotypes associated with a phenotype.

Haplotype Trend Regression with eXtra flexibility (HTRX) is an R package to learn sets of interacting features that explain variance in a phenotype. Genome-wide association studies (GWAS) have identified thousands of single nucleotide polymorphisms (SNPs) associated with complex traits and diseases, but finding the true causal signal from a high linkage disequilibrium block is challenging. We focus on the simpler task of quantifying the total variance explainable not just with main effects but also interactions and tagging, using haplotype-based associations. HTRX identifies haplotypes composed of non-contiguous SNPs associated with a phenotype and can naturally be performed on regions with a GWAS hit before or after fine-mapping. To reduce the space and computational complexity when investigating many features, we constrain the search by growing good feature sets using 'Cumulative HTRX', and limit the maximum complexity of a feature set. As the computational time scales linearly with the number of SNPs, HTRX has the potential to be applied to large chromosome regions. HTRX is implemented in R and is available under GPL-3 licence from CRAN (https://cran.r-project.org/web/packages/HTRX/readme/README.html). The development version is maintained on GitHub (https://github.com/YaolingYang/HTRX). yaoling.yang@bristol.ac.uk. Supplementary data are available at Bioinformatics Advances online.

Self-supervised graph representation learning integrates multiple molecular networks and decodes gene-disease relationships

Leveraging molecular networks to discover disease-relevant modules is a long-standing challenge. With the accumulation of interactomes, there is a pressing need for powerful computational approaches to handle the inevitable noise and context-specific nature of biological networks. Here, we introduce Graphene, a two-step self-supervised representation learning framework tailored to concisely integrate multiple molecular networks and adapted to gene functional analysis via downstream re-training. In practice, we first leverage GNN (graph neural network) pre-training techniques to obtain initial node embeddings followed by re-training Graphene using a graph attention architecture, achieving superior performance over competing methods for pathway gene recovery, disease gene reprioritization, and comorbidity prediction. Graphene successfully recapitulates tissue-specific gene expression across disease spectrum and demonstrates shared heritability of common mental disorders. Graphene can be updated with new interactomes or other omics features. Graphene holds promise to decipher gene function under network context and refine GWAS (genome-wide association study) hits and offers mechanistic insights via decoding diseases from genome to networks to phenotypes.