(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Exome sequencing identifies novel genetic variants associated with varicose veins [1]

['Dan-Dan Zhang', 'Department Of Neurology', 'Qingdao Municipal Hospital', 'Qingdao University', 'Qingdao', 'Xiao-Yu He', 'National Center For Neurological Disorders', 'Huashan Hospital', 'State Key Laboratory Of Medical Neurobiology', 'Moe Frontiers Center For Brain Science']

Date: 2024-08

A total of 13,823,269 autosomal genetic variants were obtained after quality control. We identified 36 VV-related independent common variants mapping to 34 genes by single-variant analysis and three rare variant genes (PIEZO1, ECE1, FBLN7) by collapsing analysis, and most associations between genes and VV were replicated in FinnGen. PIEZO1 was the closest gene associated with VV (P = 5.05 × 10 −31 ), and it was found to reach exome-wide significance in both single-variant and collapsing analyses. Two novel rare variant genes (ECE1 and METTL21A) associated with VV were identified, of which METTL21A was associated only with females. The pleiotropic effects of VV-related genes suggested that body size, inflammation, and pulmonary function are strongly associated with the development of VV.

In this study, whole-exome sequencing data from the UK Biobank were used to explore the effect of genetic variants on varicose veins (VV) and search for new VV-related genes. In contrast to traditional association studies, large-scale whole-exome sequencing analysis is more capable of identifying rare genetic variants (MAF < 1%) in diseases. The current study identified 34 VV-associated common variant genes by single-variant analysis and three rare variant genes (PIEZO1, ECE1, FBLN7) by collapsing analysis, and most associations were validated in FinnGen. In addition to replicating several genes reported in previous genome-wide association studies, we identified 17 novel genes that may be associated with VV. Through subsequent phenome-wide association analyses of identified genes, we found that these genes are also strongly associated with body size, inflammation, and pulmonary function. These findings contribute to understanding the underlying mechanisms of pathogenesis and developing novel therapeutic strategies for VV.

Funding: JT, Yu was supported by grants from the Science and Technology Innovation 2030 Major Projects (2022ZD0211600), National Natural Science Foundation of China (82071201, 82071997), Shanghai Municipal Science and Technology Major Project (2018SHZDZX01), Research Start-up Fund of Huashan Hospital (2022QD002), Excellence 2025 Talent Cultivation Program at Fudan University (3030277001), and Shanghai Talent Development Funding for The Project (2019074). W Cheng was supported by grants from the Shanghai Rising-Star Program (21QA1408700). JF Feng was supported by grants from the 111 Project (B18015). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability: The main data, including individual-level phenotype and sequencing data, used in this study were accessed from the UK Biobank under application number 19542 ( https://www.ukbiobank.ac.uk/ ). Statistics for varicose veins in FinnGen are available on the website ( http://r8.finngen.fi ) and are indicated with the phenotype code I9_VARICVE. The scRNA-seq data used in the present study were acquired from GEO database ( https://www.ncbi.nlm.nih.gov/geo/ ) with the accession number: GSE201333. The following software and packages were used for data analysis: FUMA v.1.3.8 ( https://fuma.ctglab.nl/ ), SnpEff ( https://pcingola.github.io/SnpEff/ ), GCTA v.1.93 ( https://yanglab.westlake.edu.cn/software/gcta/#COJO ), SAIGE-GENE+ ( https://github.com/saigegit/SAIGE ), BHR ( https://github.com/ajaynadig/bhr ), PLINK ( https://www.cog-genomics.org/plink/ ) and R v.4.2.0 ( https://www.r-project.org/ ), DAVID, ( https://david.ncifcrf.gov/ ), Gene-SCOUT ( https://astrazeneca-cgr-publications.github.io/gene-scout/ ). Scripts used to perform the analyses are available at https://github.com/ddzhang877/vv_wes .

In contrast to traditional studies, large-scale whole-exome sequencing (WES) analysis is more capable of identifying rare genetic variants (MAF < 1%) in diseases [ 16 ]. Rare variants are genetic markers of high disease risk and help to identify novel genetic targets for drug interventions [ 17 ]. Apart from revealing the full spectrum of protein-coding variants, WES can facilitate the identification of novel loss-of-function (LOF) variants and enhance the ability to detect the associations between LOF variants and diseases. LOF variants can identify drivers of genetic risk, novel disease genes, and therapeutic targets [ 16 ]. Also, large-scale sequencing can evaluate the disease prevalence and carrier frequencies of rare genetic variants [ 18 ]. A recent study focused on the effect of PIEZO1 on altered VV risk and identified the association of VV with rare protein-truncating variants in PIEZO1 [ 19 ]. A multi-phenotype exome-wide association study (ExWAS) from the UK Biobank (UKB) involving 49,960 participants [ 16 ] identified novel LOF variants with significant disease impact, including PIEZO1 on VV (P = 3.2 × 10 −8 ). In addition, a phenome-wide level ExWAS [ 20 ] exploring the contribution of rare variants to human disease also found the association between PIEZO1 and VV (P = 3.24 × 10 −24 ). In contrast to their studies, we conducted a more comprehensive ExWAS of VV using the updated exome sequencing data of more than 350,000 UKB individuals, with additional validation and exploration of the findings and biological functions. We identified 19 known and 17 novel causal genes significantly associated with VV, with most of these associations validated in FinnGen [ 21 ]. In addition, our study explored the phenotypic associations between VV genes and multiple traits, showing that cardiovascular disease, height, biochemistry, and inflammatory indicators might contribute to the development of VV.

Varicose veins (VV) are the most common chronic condition of the venous system, often affecting the lower extremities and manifesting as dilated, stretched, or tortuous superficial veins [ 1 ]. The majority of VV patients suffer from complications such as pain, swelling, hyperpigmentation, and ulcers. About 23% of adults aged 40–80 in the United States develop VV, including 22 million women and 11 million men [ 2 ]. Today, endovenous laser ablation is considered clinically the first-line treatment option for VV but has a post-operative recurrence rate of up to 20% [ 3 , 4 ]. Therefore, it is particularly important to define the etiology of VV. Previous epidemiological studies have shown that VV are associated with several risk factors, including advanced age, being female, pregnancy, obesity, height, and history of deep vein thrombosis [ 5 – 8 ]. A positive family history is one of the important risk factors for VV, which suggests that VV are likely to be modulated by genetic factors [ 9 – 12 ]. Most of the previous genetic studies on VV were genome-wide association studies (GWAS) [ 13 – 15 ]. Within the last decade, dozens of genomic loci associated with VV have been identified. However, GWAS are more limited in scope and mostly focused on common variants (minor allele frequency (MAF > 1%)), which usually have a small effect size and cannot directly identify the causal gene.

In addition, we tried to find genes with similar continuous trait fingerprints to these rare variant genes on the Gene-SCOUT website [ 25 ]. SCUBE3 and IGF2BP2 were found to be the most similar to the PIEZO1, respectively ( Fig 6A ). The module of gene signatures showed that both these similar genes and PIEZO1 were negatively associated with height related traits ( Fig 6B ). Recent evidence using machine learning methods [ 15 ] has identified that height is a risk factor for VV. Our findings further validated the effect of height on the risk of developing VV.

To further explore the association between VV-related genes and other phenotypes, we performed phenome-wide association analysis (Phe-WAS) for multiple quantitative traits and binary phenotypes, including biochemical indicators, cardiovascular disease, body size measures, inflammatory indicators, and pulmonary function ( S18 Table ). As expected, the majority of 53 traits showed significant associations with the identified VV-related genes (P < 2.77 × 10 −5 ), especially common variant genes such as RNF5 (27/53), PGBD1 (24/53), TRIM31 (24/53), and GNA12 (14/53) ( Fig 4 ). Furthermore, among these significant genotype-phenotype associations of independent common variant genes, 168 associations were negative and 113 were positive ( S19 Table ). Among three rare variant genes, PIEZO1 was significantly associated with a variety of curated phenotypes, such as HbA1c (P = 1.02 × 10 −75 ), standing height (P = 5.41 × 10 −23 ), sitting height (P = 7.23 × 10 −16 ), and forced expiratory volume in 1 second (FEV1 pred, P = 1.07 × 10 −12 ), and FBLN7 was mainly linked to height-related traits and pulmonary function indices ( Fig 5 and S20 Table ).

In addition, we explored the expression of VV-related independent common and rare variant genes in 30 general tissues in FUMA GTEx. We discovered that several genes were highly expressed in adipose tissue, including PIEZO1 and the novel identified ECE1 ( Fig 3C and S17 Table ). Next, single-cell expression data from human adipose tissue showed that PIEZO1 and ECE1 were most strongly expressed in the endothelial cell subpopulation, and higher expression of ECE1 was also detected in fibroblasts and neutrophils ( Fig 3D, 3E and 3F ).

(A) Results of the functional enrichment in biological pathway databases (P < 0.05). (B) Results of the functional enrichment in biological pathway databases in females. Only the top 12 significant pathways (P < 0.05) are shown, and the complete results are shown in S16 Table . (C) The heat map shows the tissue enrichment results of each VV-related gene. The H2AC6 gene was not recognized in FUMA Ensembl ID. (D) Single-cell RNA sequencing analysis results of adipose tissue. (E, F) Cell-specific expression of PIEZO1 and ECE1 in adipose tissue. Abbreviation: BP, biological process; CC, cellular component; MF, molecular function; TPM, transcripts per million; Umap, Uniform manifold approximation and projection.

We conducted a pathway enrichment analysis using the Gene Ontology (GO) Consortium [ 23 , 24 ] for all 36 genes identified in the study as being significantly associated with VV. There were 12 GO terms significantly enriched for these genes (false discovery rate (FDR) correction, corrected α = 0.05). It was found that several VV genes (including UBE2H, TRIM31, and RNF5) were significantly enriched in protein K48-linked ubiquitination (P = 7.20 × 10 −3 ) and 30 genes were enriched in protein binding (P = 2.79 × 10 −2 ) ( Fig 3A and S15 Table ). In females, GO enrichment analysis showed that the identified VV-related genes were significantly enriched in protein ubiquitination (P = 1.47 × 10 −3 ) and most genes were enriched in protein binding (P = 1.48 × 10 −2 ) ( Fig 3B and S16 Table ). However, the VV genes identified in males were not successfully enriched for the same biological processes.

To explore whether the effects of genes on VV differ by sex, we re-performed single-variant analysis and collapsing analysis after stratifying participants by sex. These analyses included 162,210 males and 188,560 females. After clumping, we identified 5 and 21 independent common variants significantly related to VV in males and females, respectively, with four variants mapping to novel genes only in females (ARHGEF26, BTN1A1, H1-5, TRIM27, S13 Table ). The PIEZO1 gene was strongly associated with VV, with different significance levels across sexes (male, P = 4.66 × 10 −14 ; female, P = 6.81 × 10 −21 ). The significant association between FBLN7 and VV was found only in females (P = 1.12 × 10 −6 ). Notably, we additionally identified a significant association between a novel gene (METTL21A, P = 1.38 × 10 −6 ) and VV in females in the model of LOF variants with MAF < 1 × 10 −4 ( S14 Table ).

We performed leave-one-variant-out (LOVO) analyses to assess the robustness of associations identified in gene-based collapsing analysis and to find the variants that affect each VV-related rare variant gene ( S11 Table ). After the removal of the variant chr1:21227980:A:G, the association between ECE1 and VV became non-significant (P = 3.81 × 10 −6 ). Similarly, the highest LOVO P value attained for the association was P = 9.90 × 10 −3 for FBLN7 and VV after removing chr2:112165024:C:G. Importantly, the association of VV and PIEZO1 remained robust in the LOVO analysis ( S5 Fig ). This suggested that the significant associations of VV with ECE1 and FBLN7 were strongly influenced by single rare variants, whereas the association with PIEZO1 were more influenced by a combined effect of multiple rare variants. We further assessed whether the significant rare variants were independent of nearby common variants by a conditional analysis. The results showed that the nearby common variants did not affect the effect size and significance of these three rare variant genes ( S12 Table ).

Then we calculated the gene-wise burden heritability of VV using the method developed by Weiner et al. [ 22 ]. In this analysis, ultra-rare variants were defined as MAF < 1 × 10 −5 and rare variants were defined as having a MAF between 1 × 10 −5 and 1 × 10 −2 . The burden heritability of rare variants focuses on variants that predict the most serious functional consequences. Among these, ultra-rare coding LOF variants explained 0.287% (se = 0.056%) of phenotypic variance, more than rare LOF variants explained. But rare missense variants accounted for more burden heritability than ultra-rare (rare, 0.053%; ultra-rare, -7.63 × 10 −7 ) ( Fig 2D and S10 Table ).

Considering the high prevalence and incidence of VV in the general population, we calculated the carrier frequencies and disease prevalence rates of the putative pathogenic variants. PIEZO1 had the highest carrier frequency of missense variants (3.77%, the probability of being loss-of-function intolerant (pLI) = 0.54), followed by FBLN7 (1.47%, pLI = 0) and ECE1 (0.20%, pLI = 1) ( Fig 2B and S8 and S9 Tables ). The high pLI of ECE1 underscores its pronounced intolerance to LOF variants, consistent with our findings of lower carrier frequency ( S4 Fig ). These variant carriers showed generally modest disease prevalence rates (< 15%), and the disease prevalence rates of missense variants were lower than those of LOF variants in carriers ( Fig 2C ).

(A) Manhattan plot of the gene-based collapsing analysis. The red dashed line indicates the significance threshold (P = 2.51 × 10 −6 ). We only showed the results of the collapsing test with four models including two MAF cutoffs (0.01 and 0.001) and two variant annotation groups (LOF and LOF + missense), and the other results are shown in S6 Table . All models were adjusted for age, sex, and the top 10 PCs. (B) Bar chart showing carrier frequencies for rare LOF and missense variants for three identified rare variant genes. (C) Disease prevalence of LOF and missense variants in VV-related rare variant genes. (D) Heritability of the burden of genetic variants in VV. Abbreviation: LOF, loss-of-function.

We performed the gene-level collapsing analysis to determine rare variant associations with VV. To elucidate the differences in genetic structures between genes, our analysis included 19,897 genes under 8 different models ( Methods ). Collapsing analysis showed 13 significant associations involving 3 genes (PIEZO1, ECE1, FBLN7) after Bonferroni correction (P < 2.51 × 10 −6 ) ( Fig 2A and S6 Table ). The corresponding Q-Q plots are shown in S3 Fig . The significance level of the genetic variant associations between these significant genes and VV was enhanced by the inclusion of missense variants. In the LOF + missense model with MAF < 1%, these 3 genes were significantly associated with an elevated risk of VV (PIEZO1, odds ratio (OR) = 1.010, 95% confidence interval (CI) = 1.008–1.013, P = 1.15 × 10 −18 ; ECE1, OR = 1.028, 95% CI = 1.016–1.039, P = 4.89 × 10 −7 ; FBLN7, OR = 1.012, 95% CI = 1.007–1.017, P = 1.93 × 10 −6 ) ( S7 Table ). The gene which had the strongest correlation with VV was PIEZO1 in the LOF model with MAF < 10 −3 (PIEZO1, OR = 1.037, 95% CI = 1.029–1.046, P = 5.01 × 10 −17 ). The associations of PIEZO1 and FBLN7 with VV have been reported in previous genetic studies [ 14 , 19 ]. The significant difference between the LOF + missense model (P = 4.89 × 10 −7 ) and the LOF-only model (P = 0.02) for ECE1, a novel identified VV-related gene, suggested its association with VV was primarily driven by the missense mutation ( Table 2 ). The associations of the three genes we identified in collapsing analysis with VV were further successfully validated in the VV genomic data from FinnGen (P ranged from 2.39 × 10 −59 to 4.93 × 10 −5 , Table 2 ).

(A) Manhattan plot of single-variant associations for common variants and VV. The red solid line indicates the significance threshold (P = 1.19 × 10 −6 ), and the red dashed line indicates the genome-wide significance threshold (P = 5.0 × 10 −8 ). The gene name corresponds to the gene closest to the variant. Black fonts indicate previously identified genes, while red fonts indicate newly identified genes. (B) The quantile-quantile (Q-Q) plot for single-variant analysis. (C) The plot of effect size (odds ratio) versus effect allele frequency of identified independent common variants. Abbreviation: OR, odds ratio.

The mixed-effects logistic models adjusted for age, sex, and the top 10 principal components (PCs) were used to assess the association between VV and common coding variants. Firstly, 169 variants and 114 variants were found to reach the exome-wide significance (P < 1.19 × 10 −6 ) and the genome-wide significance (P < 5 × 10 −8 ) in the single-variant analysis ( Fig 1A and S3 Table ). After clumping 169 variants in strong linkage disequilibrium, 36 independent common variants associated with VV were identified ( Fig 1A ), of which 34 were successfully validated in FinnGen [ 21 ] ( Table 1 and S4 Table ). Among these independent common variants, 19 were significantly protective against VV, while the other 17 were significantly associated with an elevated risk of VV ( Table 1 ). In addition, 36 independent common variants were mapped to 34 genes, of which 18 have been reported previously and the other 16 genes were novel (including TRIM10, UBE2H, TUBAL3, DUSP8, DNAH10, CAPRIN2, MSL1, ZBTB4, CIB3, SERPIND1, SHANK3, REST, CTXN3, H2AC6, PGBD1, ABHD16A). PNO1 and PIEZO1 had the most significant associations with VV, and the P-values were 9.81 × 10 −44 and 1.75 × 10 −39 , respectively. TRIM10 showed the strongest association with VV among 16 novel common variant genes (P = 5.29 × 10 −14 ). The quantile-quantile (Q-Q) plot of single common variants is shown in Fig 1B . It was noteworthy that these independent common variants were characterized by larger effect sizes at lower frequencies ( Fig 1C ). Furthermore, we conducted a conditional analysis of the identified common variants linked to VV using conditional and joint analysis (COJO). The results revealed that 27 independent common variants were retained, 25 of which were consistent with variants identified after clumping ( S5 Table ).

In this study, we performed an ExWAS by utilizing phenotypic and genetic data from UKB, including exome sequencing data and VV diagnosis data ( S1 Table ). Following stringent quality control (QC) of genotypes and samples, we identified 350,770 unrelated Caucasian participants (21,643 VV cases and 329,127 controls) with a mean age of 56.94 years at enrollment, of which 53.76% were female ( S2 Table ). The case and control groups had similar age distribution ( S1 Fig ). The clustering of the VV case and control groups in the principal component analysis is shown in S2 Fig . A total of 13,823,269 autosomal genetic variants were obtained after QC, including 100,098 common variants (MAF > 1%) and 13,723,171 rare variants (MAF < 1%).

Discussion

Previous GWAS [14,15,26] have identified multiple risk loci and pathways involved in the pathophysiology of VV, improving the understanding of the polygenic architecture of the disease. The WES data from the UKB could provide a new direction for exploring the genetic mechanisms of VV [27]. VV-related genes and genetic variants identified in population-based WES studies were more abundant and reliable. It is perhaps that whole-exome data are usually obtained directly from sequencing rather than by imputation. WES typically focuses on coding regions of the genome and is more cost-effective and widely available than whole-genome sequencing [28]. Compared to non-coding variants, WES can directly assess variants that alter protein sequence, making it easier to explain their functional consequences. It also provides a clearer pathway to deeper mechanistic insights, and can potentially be useful for therapeutic target discovery [29–32] and precision medicine [33,34].

Recently, several studies of human gene-phenotype association for rare coding variants have identified the association between PIEZO1 and VV by using exome sequencing data from UKB participants [16,20]. Compared to the studies by Van Hout et al.[16] and Wang et al.[20], our study has a larger sample size and more validation of the findings (replication in the FinnGen cohort) and exploration of biological function (e.g. functional enrichment analysis, tissue expression analysis, and single-cell RNA sequencing analysis). In addition to the replication of 19 known VV-related genes from previous studies, our study further identified 17 novel genes. Most of these genes were successfully validated in FinnGen. Tissue analysis and single-cell gene expression analysis emphasized the importance of adipose tissue in the association between these identified genes and VV risk. Our Phe-WAS revealed the strong associations of these genes with body size measures, biological indicators, and inflammatory markers.

We identified a total of 36 VV-associated genes in our large-scale exome sequencing study. First, the exome-wide single-variant analysis identified 36 VV-related independent common variants mapping to 34 genes, half of which have been previously established [14,15,26,35,36]. The significant effects found in the single-variant analysis were the effect of PNO1 on chromosome 2 and that of PIEZO1 on chromosome 16. PNO1 plays an important role in both proteasome and ribosome biogenesis [37], and PIEZO1 is required for vascular development and function [38]. PNO1 is predominantly expressed in the liver and spleen, and slightly in the thymus, testis, and ovary, but not in the heart or brain [39]. In addition, it has been suggested that PNO1 may be involved in the expression of calcineurin phosphatase, a key signaling component in angiogenesis, and calcineurin phosphatase activity is essential for vascular development [15,40]. As the most significant novel gene identified in the single-variant analysis, TRIM10 was thought to be potentially involved in the body’s immune response [41]. Phenotypic association analysis revealed that TRIM10 had associations with several inflammatory markers, such as platelets, monocytes, and neutrophils. As a protein-coding gene, UBE2H encodes an E2 ubiquitin-conjugating enzyme family protein, which is involved not only in ubiquitination but also in cytoskeletal regulation and calcium signaling [42]. Our study also indicated that UBE2H is involved in protein binding and protein K48-linked ubiquitination by functional enrichment of the gene.

The other findings in this study were the associations of VV with three rare variant genes (PIEZO1, ECE1, and FBLN7) identified by the gene-based collapsing analysis. Both single-variant analysis and gene-based collapsing analysis revealed an association between VV and PIEZO1, supporting the previously reported hypothesis that common and rare variants may involve overlapping genes [43,44]. A study by Backman et al. [43] found that missense variants in PIEZO1 might have a gain-of-function effect and reduce VV risk, which offers the possibility of treating VV by pharmacological interventions. PIEZO1 is a mechanically activated ion channel that plays a decisive role in vascular architecture [38,45]. Its integral or endothelial-specific disruption can severely damage the growing vascular system and even lead to embryonic death.

As a newly discovered VV-related gene, ECE1 is an estrogen-regulatory gene in mesenteric arteries that could catalyze the conversion of pro-endothelin into endothelin-1, a potent vasoconstrictor [46,47]. A mouse model showed that estrogen receptor stimulation suppressed ECE1 expression [48]. In addition, previous studies have found an increase in ECE1 in pathological conditions, for instance, hypertension, coronary atherosclerosis, and vascular injury models [49–52]. Our expression analysis of ECE1 revealed its high expression in adipose tissue, especially in endothelial cells, and a positive association between obesity and VV risk has been demonstrated before [15,26]. It was well known that the development of VV was associated with a variety of vascular factors, such as changes in hemodynamics, endothelial cell activation, inflammation, and hypoxia [53–55]. As a cell adhesion molecule associated with diseases and developmental abnormalities, FBLN7 plays an essential role in specialized tissues such as the placenta, blood vessels, and cartilage [56–58]. FBLN7 protein fragments have been found to regulate endothelial cells and leukocyte function [59,60]. With anti-angiogenic function [57], they can be used to treat inflammatory diseases by modulating immune cell activity [61].

Moreover, although the carrier frequencies of LOF variants in the population were low, LOF mutations confer greater disease prevalence and genetic susceptibility to VV. As for the burden heritability, ultra-rare LOF variants had the highest heritability, followed by rare LOF variants and rare missense variants, which was in line with the previous finding by Weiner et al. [22] that LOF variants explained the majority of burden heritability.

We evaluated the robustness of the three genes identified in the collapsing analysis. The rare variant association results remained stable after adjusting for nearby common variants, probably due to our strict variant filter. Our sensitivity analysis revealed a novel gene in females (METTL21A), which prompted us to speculate whether the association between VV and sex might be influenced by genetic factors. Moreover, the current study revealed an overlap between VV-related genes identified by single-variant analyses and those identified by burden tests, suggesting the robustness of the association between overlapping genes and VV risk. A previous genetic study showed that vascular diseases were tightly linked to anthropometric indices [36]. The Phe-WAS of the identified genes showed VV genes were also associated with VV-related phenotypes, such as cardiovascular disease, body size measures, and pulmonary function indices. Phe-WAS can further explore genetic contributions to disease and identify the mechanisms of sharing across diseases [62].

There were some limitations in our study. First, the UKB was generally recruited from populations of European ancestry between the ages of 37 and 73, which makes our findings potentially inapplicable across all ethnic and age groups. Second, disease diagnoses are typically confirmed based on participant self-reports, ICD diagnosis codes, surgical records, and death registries. This may lead to the misclassification of disease phenotypes. Nevertheless, previous studies have been able to replicate genetic loci findings for common variants using the phenotype definitions in GWAS [14,15,26]. Third, we focused on the coding regions of the genome captured by WES, which may lead to the neglect of variants in non-coding regulatory regions. Due to the difficulty in obtaining exome sequencing data for VV from other databases, we validated our findings using summary statistics from FinnGen. The strength of our study is that it has the largest sample size of any VV-related whole-exome association study to date. Exome-wide analyses allow for more efficient targeting of variant genes that play an essential role in complex diseases compared to GWAS [63]. We defined the VV phenotype mainly based on the study of Ahmed et al.[14], which successfully replicated most of the VV-related risk loci identified in previous studies using this definition [15,64]. We additionally validated the results in FinnGen to determine the reproducibility of our findings.

In conclusion, large-scale exome-wide association studies are required for exploring the genetic associations of diseases. In this study, we identified many known genes and several novel effector genes significantly associated with VV. We found that LOF variants explained more burden heritability than missense variants. Furthermore, investigating genotype–phenotype associations would be critical for our insight into the underlying biological mechanisms of the disease, as well as for disease prevention and treatment. The study of genetic variants and diseases helps to develop strategies to target risk genes for prevention and treatment purposes.

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1011339

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.