Genome-wide association analyses of common infections in a large practice-based biobank
BMC Genomics volume 23, Article number: 672 (2022)
Infectious diseases are common causes of morbidity and mortality worldwide. Susceptibility to infection is highly heritable; however, little has been done to identify the genetic determinants underlying common infectious diseases. One GWAS was performed using 23andMe information about self-reported infections; we set out to confirm previous loci and identify new ones using medically diagnosed infections.
We used the electronic health record (EHR)-based biobank at Vanderbilt and diagnosis codes to identify cases of 12 infectious diseases in white patients: urinary tract infection, pneumonia, chronic sinus infections, otitis media, candidiasis, streptococcal pharyngitis, herpes zoster, herpes labialis, hepatitis B, infectious mononucleosis, tuberculosis (TB) or a positive TB test, and hepatitis C. We selected controls from patients with no diagnosis code for the candidate disease and matched by year of birth, sex, and calendar year at first and last EHR visits. We conducted GWAS using SAIGE and transcriptome-wide analysis (TWAS) using S-PrediXcan. We also conducted phenome-wide association study to understand associations between identified genetic variants and clinical phenotypes.
We replicated three 23andMe loci (p ≤ 0.05): herpes zoster and rs7047299-A (p = 2.6 × 10–3) and rs2808290-C (p = 9.6 × 10–3;); otitis media and rs114947103-C (p = 0.04). We also identified 2 novel regions (p ≤ 5 × 10–8): rs113235453-G for otitis media (p = 3.04 × 10–8), and rs10422015-T for candidiasis (p = 3.11 × 10–8). In TWAS, four gene-disease associations were significant: SLC30A9 for otitis media (p = 8.06 × 10–7); LRP3 and WDR88 for candidiasis (p = 3.91 × 10–7 and p = 1.95 × 10–6); and AAMDC for hepatitis B (p = 1.51 × 10–6).
We conducted GWAS and TWAS for 12 infectious diseases and identified novel genetic contributors to the susceptibility of infectious diseases.
Infections are among the most common causes of morbidity and mortality worldwide, resulting in millions of deaths [1, 2]. Complications of serious infection in the U.S. contribute to 1 in 3 hospital deaths and ~ 250,000 deaths annually . Susceptibility to infection is highly heritable, likely due to major selection pressure over millennia, when infection was the leading cause of death and no effective antimicrobials existed . More than 300 rare Mendelian disorders resulting from mutations predominantly in genes regulating immune response predispose individuals to infection [4, 5] and provide extreme proof of the critical importance of host genetic variation in susceptibility to infection. However, such variants do not account for the high heritability of susceptibility to infection seen in other studies. In a landmark twin study, adults who had been adopted as children had a 5.8-fold increased risk of dying from infection if one of their biological parents had died from infection before the age of 50 years . Other twin studies have shown high heritability for traits such as infection (h2 = 0.43) , staphylococcal infection (h2 = 0.7) , and death due to infection (h2 = 0.4) .
Despite high heritability, the genetics of susceptibility to infection is poorly defined and is recognized as a neglected area of research: only 4% of the catalog of genome wide association studies (GWAS) relates to the broad area of infectious disease . Many attempts to identify the genetic determinants underlying common infections have major limitations. First, associations have been sought in small candidate gene studies; second, few GWAS studies have been broadly relevant to patients in the U.S. One of the largest GWAS was performed using 23andMe data with self-reported health history for 23 infections . In that study, Tian et al. identified genes that play key roles in immune response and inflammatory processes associated with susceptibility to infections. However, the identified associations have not been tested in a real-world setting with infections diagnosed by physicians, and relatively few loci have been identified.
The COVID-19 pandemic resulted in urgent work to expand our understanding of the genetic mechanisms underlying severe respiratory viral infection and its complications. A recent meta-analysis of 46 independent GWASs identified loci that contribute to susceptibility or severity of COVID-19 infection  — supporting the critical role of host genetics in infectious diseases. However, whether the identified COVID-19 loci are also involved in susceptibility to other respiratory infections is unclear.
Biobanks linked to patients’ electronic health records (EHRs) provide an unprecedented opportunity to perform genetic studies and understand infectious disease. The biobank at Vanderbilt (BioVU) is one of the largest practice-based biobanks in the U.S. We set out to replicate the observations from the previous 23andMe GWAS and test the associations between the identified variants and clinical phenotypes using phenome-wide association studies (PheWAS) to identify additional associated infections as well as co-morbidities that could predispose to infection. One of our primary objectives was to replicate the earlier findings from a GWAS study that used self-reported history of various infections as the phenotypes of interest with those of a GWAS study that used the more objective outcomes of medically diagnosed infections. Then, we conducted GWAS and transcriptome-wide association study (TWAS) to further define the role of host genetics in common infections. Last, we tested if previously identified COVID-19 loci also associated with susceptibility to pneumonia in our BioVU cohort .
Data were obtained from the Synthetic Derivative (SD) and BioVU at Vanderbilt University Medical Center (VUMC) that contains a de-identified copy of the EHR for every patient and has genome-wide genotyping available for > 100,000 patients [13,14,15]. The BioVU follows the declaration of Helsinki. The study followed the declaration of Helsinki. The study was exempted by Vanderbilt University Medical Center Institutional Review Board.
We included individuals whose race was identified as white in the de-identified EHR and who had genome-wide genotyping available. We identified patients with the infectious diseases of interest using the International Classification of Disease Clinical Modification, Ninth Revision (ICD9CM) and Tenth Revision (ICD10CM) codes (Supplement Table 1).
We set out to replicate associations with common infections in Tian’s 23andMe GWAS study  which included 23 phenotypes; of those, we studied phenotypes which could be defined by ICD codes and for which we had more than 100 cases (Supplementary Table 2). These were urinary tract infection (UTI), pneumonia, chronic sinus infections, otitis media, candidiasis, streptococcal pharyngitis, herpes zoster, herpes labialis, hepatitis B, infectious mononucleosis, and tuberculosis (TB) or a positive TB test. We also included hepatitis C, a common infection that was not included in Tian’s report. The ICD diagnosis codes included in each phenotype are shown in Supplementary Table 1. For each candidate infectious disease, individuals with 2 or more codes for the phenotype on different days were considered as cases for the disease . Individuals with only 1 mention of ICD code related to the disease were excluded from the analysis of that candidate infectious disease. We selected controls from individuals with no ICD codes for the candidate disease and matched these with cases of the infectious disease using year of birth, sex, and years of first and most recent EHR. We chose the matching factors to minimize important imbalances that could occur between case and control groups and thus reduce potential confounding; for example, we matched cases and controls for age and length of EHR because younger individuals and those with shorter EHRs have less time in which to accumulate clinical diagnoses, and we matched for sex because for some illnesses (.e.g., UTI) there are marked differences in prevalence among men and women. We matched controls to cases 5:1 for UTI, pneumonia, candidiasis, chronic sinus infection, otitis media, and hepatitis C. For infections with less than 1000 cases (streptococcal pharyngitis, herpes zoster, hepatitis B, infectious mononucleosis, TB or a positive TB test, and herpes labialis), we matched controls to cases 10:1(Table 1). For phenotypes with more than 1000 cases, we chose 1:5 case–control ratios based on statistical power calculations. For phenotypes with fewer than 1000 cases, we chose a 1:10 case–control ratio to take advantage of the additional small increase in power this provided for less frequent phenotypes .
In preliminary analyses, and as reported by others , we found that patients with cystic fibrosis (CF) contributed a strong genetic signal to pneumonia and chronic sinus infection; thus, to limit confounding by a CF genetic signal, we removed individuals with CF diagnosis codes (Supplementary Table 1) from the analyses of pneumonia and chronic sinus infection.
Genotyping and SNP imputation
Genotyping was performed on the Infinium Multi-Ethnic Genotyping Array (MEGAchip). We took necessary technical measure to control genotyping quality and excluded DNA samples with (1) per-individual call rate < 95%; (2) mismatch between reported gender and X-chromosome zygosity; or (3) unexpected duplication. We performed whole genome imputation using the Michigan Imputation Server  with the Haplotype Reference Consortium, version r1.1 [20, 21] as reference. Principal components for ancestry (PCs) were calculated using common variants (MAF > 1%) with high variant call rate (> 98%); we excluded variants in linkage and regions known to affect PCs [HLA region on chromosome 6, inversion on chromosome 8 (8,135,000–12,000,000), and inversion on chromosome 17 (40,900,000–45,000,000); GRCh37 build]. Tian previously reported 28 genetic variants significantly associated with the infections we tested in BioVU; of these, 23 were directly available in our dataset or had another variant (within 500 kb) in high linkage disequilibrium (LD) using information for European ancestry population in the 1000 genomes database (R2 > 0.9, except for rs73015965 which was in LD with rs73027818 with R2 = 0.7) [22, 23].
Genome-wide association study
We used SAIGE  to test associations between genotypes and risk of candidate infectious diseases using logistic regression assuming additive allelic effects and adjustment for sex, year of birth, year of first clinical visit, EHR length, and 10 PCs of ancestry to account for residual population structure . Then, we conducted post-analysis quality control using EASYQC  to exclude (1) poorly imputed variants with r2 value of < 0.3, (2) variants with minor allele frequency (MAF) < 0.5%, (3) variants with MAF different from the HRC reference panel (MAF differences > 0.3), and (4) variants significantly derived from Hardy–Weinberg equilibrium (HWE, p < 1 × 10–6). As we consider each infection an independent phenotype, we applied the standard GWAS Bonferroni correction cut-off and considered a P-value of less than 5 × 10–8 as significant.
Transcriptome wide association study (TWAS)
We conducted transcriptome analysis using PrediXcan (https://github.com/hakyimlab/PrediXcan)  with summary statistics from GWAS analyses. We leveraged all 49 available reference tissues from GTEx version 8. One approach would be to use organ- or tissue-specific prediction models, such as lung for pneumonia. However, because of the strong correlations across tissues in the genetic architecture for the regulation of gene expression (largely a function of the cell types making up that tissue), it is statistically powerful and thus we chose to utilize information from the tissues with the highest quality prediction performance or construct cross-tissue model. We also conducted cross-tissue transcriptomic analyses using MultiXcan and meta-analyzed all available tissue-based tests . P-values of less than 2.5 × 10–6 (0.05/20000 genes) were considered significant.
Phenome-wide association studies (PheWAS)
PheWAS was conducted to identify clinical phenotypes that associate with infection-related genetic variants either reported by Tian et al. (variants in Table 2) or identified in current study [29, 30]. Specifically, we grouped each individual’s ICD codes into PheCodes following an established protocol [31, 32]. To be a case for each PheCode, an individual needs to have relevant ICD codes on at least 2 different days. Controls were individuals with no relevant ICD codes. Individuals with only one occurrence of a relevant code were excluded from the analyses. In a cohort of 65,592 white individuals, we analyzed a total of 1739 PheCodes with more than 20 cases. P-values of less than 2.9 × 10–5 (0.05/1739) were considered significant.
We conducted a post-analysis power calculation to evaluate our ability to detect the odds ratios detected in the case–control phenotypes from Tian et al.’s. report, including herpes zoster (OR 1.07–1.14), herpes labialis (OR 1.08), infectious mononucleosis (OR 1.08), hepatitis B (OR 1.32), pneumonia (OR 1.1), and otitis media (OR 1.06 – 1.43). We could not run the power calculation for (1) continuous traits in Tian’s report, such as streptococcal pharyngitis, candidiasis, and UTI (because we applied a case–control study design); and (2) associations with variants unavailable in our cohort, such as tuberculosis (or a positive TB test). We used Genetic Association Study (GAS) power calculator .
Replication of top infection hits with other clinical phenotypes
We also searched GWAS hits from the current study in the PheWeb database (http://pheweb.sph.umich.edu/) to test whether the identified top hits were associated with other clinical phenotypes from existing GWAS and PheWAS . In addition, we investigated whether the identified GWAS hits for COVID-19 susceptibility or severity also contributed to susceptibility to pneumonia by querying our analysis of patients with pneumonia (none of whom had COVID-19).
We identified cases and matched controls for 12 common infections, including 11 infections included in the Tian paper . The number of cases ranged from 102 (TB or positive TB test) to 9359 (UTI) Table 1.
Replication of previous GWAS of common infections
We replicated 3 associations with p < = 0.05 and the same direction of effect as Tian’s report: herpes zoster with the A allele of rs7047299(IFNA21 gene, odds ratio [OR], 1.18; 95% confidence interval [CI], [1.06–1.32]; p = 0.0026) and the C allele of rs2808290 (close to MKX gene, OR, 1.09; 95% CI [1.02–1.16]; p = 0.0096); and otitis media with the C allele of rs114947103 (CDHR3 gene, OR, 1.09; 95% CI [1.00–1.18]; p = 0.0407) (Table 2).
Phenome-wide association studies (PheWAS) of previous GWAS hits of common infections
We conducted PheWAS for the genetic variants in Tian et al.’s report and found 92 significant associations with clinical phenotypes (Supplementary Table 3, p < 2.9 × 10–5). Relating to infections, rs3131623 in HLA gene region was associated with chronic hepatitis infection (p = 5.07 × 10–7), and rs600038 in ABO gene region was associated with candidiasis (p = 2.35 × 10–5). Furthermore, 43 out of the 92 associations related to diabetes or diabetes related phenotypes and several in the HLA region associated with autoimmune diseases (Supplementary Table 3, Supplementary Fig. 1).
New associations between genetic variants and the risk of common infections
We identified 3 new loci significantly associated with infections. (Table 3, Fig. 1, Supplementary Fig. 2) Two variants in nucleotide binding protein like (NUBPL) gene, the G allele of rs113235453 (OR, 1.50; 95% CI [1.30–1.73]; p = 3.04 × 10–8) and the A allele of rs74633202 (OR, 1.50; 95% CI [1.30–1.73]; p = 3.05 × 10–8) were associated with increased risk of otitis media. The T allele of rs10422015 in WD repeat-containing protein 88 (WDR88) was associated with the increased risk of candidiasis (OR, 1.31; 95% [1.19–1.44]; p = 3.11 × 10–8) (Table 3).
Associations between the risk of common infections and the genetically predicted gene expression
In TWAS for the 12 infections studied, we found significant associations between elevated risk of (1) otitis media and genetically predicted increased expression of solute carrier family 30 member 9 gene (SLC30A9, zscore = 4.93, p = 8.06 × 10–7) in brain nucleus accumbens basal ganglia; (2) candidiasis and the genetically predicted increased expression of LDL receptor related protein 3 gene (LRP3, largest zscore 5.68, smallest p-value = 1.34 × 10–8) in tissues including esophagus mucosa, brain spinal cord cervical, artery, spleen, prostate, adrenal gland, and minor salivary gland; (3) candidiasis and the genetically predicted increased expression of WDR88 (largest z-score 5.54, smallest p-value = 3.11 × 10–8) in liver and brain cortex; (4) hepatitis B and the genetically predicted decreased expression of adipogenesis associated Mth938 domain containing gene (AAMDC, smallest z-score -4.89, smallest p-value = 1.02 × 10–6) in heart atrial appendage and skin (not sun exposed, Table 4). Additionally, several of these four disease-transcriptome associations were nominally significant (p < 10–5) in several other tissues (Supplementary Table 4). In the cross-tissue analysis, only the association between increased risk of candidiasis and the genetically predicted increased expression of WDR88 was significant (p-value = 1.83 × 10–6).
Associations between lead GWAS hits and other clinical phenotypes
We searched PheWeb and conducted PheWAS in BioVU for the lead GWAS hits in the current study (rs113235453 for otitis media and rs10422015 for candidiasis) for their associations with other clinical phenotypes. Both variants were significantly associated with non-infectious conditions: rs113235453 with non-traumatic intracranial hemorrhage (p = 6.4 × 10–7) and rs10422015 with heel bone mineral density T-score (p = 1.1 × 10–15). For infection-related phenotypes there were a few suggestive associations: (1) rs113235453 was associated with use of antibiotics for bacterial infections (co-Amoxiclav) (p = 2.4 × 10–4), and (2) rs10422015 with cough (p = 4.2 × 10–4) or postoperative infection (p = 4.4 × 10–4). In the PheWAS using BioVU samples, there were no significant associations with these two variants; however, leading associations included infection-related phenotypes such as hepatitis, candidiasis and abnormal findings on the examinations of urine. (Supplementary Table 5).
Associations between top COVID-19 hits and the risk of pneumonia
When we examined 13 loci associated with COVID-19  and susceptibility to pneumonia in our cohort we found an association between the C allele of rs13050728 in IFNAR2 (Interferon Alpha and Beta Receptor Subunit 2) gene and lower risk of developing pneumonia (OR 0.94, 95%CI [0.90–0.98], p = 0.0028, Table 5), an observation directionally similar to that for severity of COVID-19 .
The current study of the genetics of 12 common infections replicated 3 associations from previous 23andMe GWAS findings. Additionally, 2 new loci (from GWAS) and altered genetically predicted expression of 4 genes (from TWAS) were associated with altered susceptibility to infection. Last, one of the alleles identified with reduced severity risk of COVID-19 was associated with reduced risk of pneumonia.
The link between the innate immune response and infection is well established . Thus the replicated association between a variant in IFNA21 and herpes zoster previously reported by Tian et al., is of interest. IFNA21 encodes a type I interferon, which binds to interferon alpha receptor and activates innate immune responses. Further indication of the importance of this pathway is the association between an IFNAR2 variant and susceptibility to pneumonia. This variant was reported as one of the top hits associated with both COVID-19 susceptibility and severity . By leveraging summary statistics from a COVID-19 GWAS and a Mendelian randomization approach, a recent drug repurposing study prioritized IFNAR2 as one of top two candidate drug targets for early management of COVID-19 . Indeed, interferon and drugs that target interferon receptors have been used to treat infectious diseases [38,39,40]. Currently, there are phase II clinical trials testing interferons for COVID-19 infection, and the results of clinical trials are awaited [41,42,43]. In PheWAS analyses of the 23andMe variants reported to be significantly associated with infection , we observed associations 92 significant PheWAS associations with 6 SNPs (rs885950, rs2523591, rs2596465, rs3131623, rs9268652, rs9270656) associated with 43 clinical phenotypes related to diabetes. These SNPs are located in genes that associated with type 1 or type 2 diabetes in previous GWAS (Supplementary Table 3). Impaired glucose regulation is associated with an elevated risk of many infections, including hepatitis [44, 45], and SARS-CoV-2. Future studies will need to determine if variants predispose to infection directly or through associations with co-morbidities that increase risk of infection.
Additional to replicating variants from the 23andMe study, we identified several novel variants within NUBPL gene region associated with otitis media. The lead hit, rs113235453, has previously been associated with heart rate in patients with heart failure and reduced ejection fraction . NUBPL encodes nucleotide binding protein-like on chromosome 14q12, and functional variants in the gene are associated with mitochondrial complex I deficiency and linked to leukoencephalopathy and Parkinson’s disease [48, 49]. Infection is a common cause of morbidity in children with mitochondrial diseases; however, it is unclear if variation in NUBPL could influence the risk of infection through its role in mitochondrial complex I deficiency.
An additional new observation was an association between the WDR88/LRP3 region and the risk of candidiasis; further, TWAS also showed that the genetically determined expression of WDR88 and LRP3 in a variety of tissues associated with altered risk of candidiasis. The underlying mechanisms are not obvious. WDR88 has previously been associated with schizophrenia ; however, the function of the gene remains unclear. The association of LRP3 expression with candida infection in esophageal mucosa was interesting because the esophagus is a well-described site of candida infection. LRP3 encodes LDL receptor related protein 3, which is involved in the internalization of lipophilic molecules , but whether LRP3 could affect the risk of candida infection through this mechanism is not known.
Another new observation was the association between genetically predicted expression of SLC30A9 and altered risk of otitis media. SLC30A9 encodes solute carrier family 30 member 9, which acts as a zinc transporter involved in intracellular zinc homeostasis . In vitro experiments suggested that SLC30A9 interacted with human influenza A virus ; therefore, SLC30A9 might alter the risk of infection through its role in recognition and binding to pathogens. Although many of the HLA/infection associations reported by Tian et al., did not replicate, there was a close-to-significant association between HLA-DQB1 and the risk of infectious mononucleosis (p = 2.59 × 10–6, Supplementary Table 4). The HLA region is critical for host response to infection. Future studies using large cohorts are needed to better understand the role of the HLA region.
The study has many strengths: the use of diagnoses made by providers to identify cases of infection in a large EHR database; matching of cases and controls to limit confounding; performance of transcriptome analysis using GWAS summary statistics to further understand the associations between host genetics and common infections; and an ability to test the associations between known loci affecting COVID-19 and the risk of pneumonia. There are also limitations. First, while power was good for most infections, there was limited power to detect small odds ratios for low-frequency variants and less common phenotypes, such as TB/positive TB tests (N of cases, 102) and mononucleosis (N of cases, 116). Additional studies will be required for less common infections. Second, ICD codes serve primarily billing purposes and are not recorded by clinicians to facilitate research; misclassification or under or over coding of conditions may occur. Also, the study was conducted in White patients. For many infections the number of cases in Black patients was too small for GWAS and will require additional studies. Third, we matched controls to cases on age, sex, and year of first and last clinical visits. However, the potential for misclassification of controls remains. There is always a possibility that the control population was enriched for some co-segregating factors of infections. Future study is needed to validate our observations. Fourth, in TWAS analyses, the gene expression predicted by a single SNP may be less robust than those predicted by multiple variants. However, many examples show that a single SNP can contribute significantly to gene expression (e.g., LPA and rs10455872, CETP and rs18000777 etc.). In Table 3, although LRP3 gene and WDR88 gene expressions were both predicted using one SNP, it is worth noting that the same significant association was observed in multiple tissues. The replication of this LRP3/candidiasis and WDR88/candidiasis association in various tissues suggests that there may be mechanisms common across tissues. Response to infection can affect multiple organs, thus we presented data from all available tissues for readers. Lastly, the novel loci we identified were not detected in Tian’s study;  several study design factors may account for these differences among studies. For example, we studied a population obtaining medical care in a large hospital whereas Tian studied a presumably healthier population who sought a genetic test; we matched controls to cases whereas Tian did not; and the studies employed different disease phenotype definitions (diagnosis billing codes vs. self-report) that may vary in sensitivity and specificity. Also, environmental, social, and economic factors vary among populations, and neither Tian’s report nor our study included these potentially important factors as covariates. As the All of Us (AoU) project develops and collects information about those factors and links them to EHR and genetic data, such studies will be possible.
In conclusion, we conducted GWAS and TWAS for 12 common infectious diseases and identified novel genetic contributors to the susceptibility of infection diseases.
Availability of data and materials
The summary statistics is available in GWAS catalog (https://www.ebi.ac.uk/gwas/, GCP ID GCP000359).
The top 10 causes of death [Internet]. [cited 2019 Apr 22]. Available from: https://www.who.int/news-room/fact-sheets/detail/the-top-10-causes-of-death
Cecconi M, Evans L, Levy M, Rhodes A. Sepsis and septic shock. Lancet Lond Engl. 2018;392:75–87.
Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315:801–10.
Burgner D, Jamieson SE, Blackwell JM. Genetic susceptibility to infectious diseases: big is beautiful, but will bigger be even better? Lancet Infect Dis. 2006;6:653–63.
van de Vosse E, van Dissel JT, Ottenhoff THM. Genetic deficiencies of innate immune signalling in human infectious disease. Lancet Infect Dis. 2009;9:688–98.
Sørensen TIA, Nielsen GG, Andersen PK, Teasdale TW. Genetic and Environmental Influences on Premature Death in Adult Adoptees. N Engl J Med. 1988;318:727–32.
Polderman TJC, Benyamin B, de Leeuw CA, Sullivan PF, van Bochoven A, Visscher PM, et al. Meta-analysis of the heritability of human traits based on fifty years of twin studies. Nat Genet. 2015;47:702.
Lakhani CM, Tierney BT, Manrai AK, Yang J, Visscher PM, Patel CJ. Repurposing large health insurance claims data to estimate genetic and environmental contributions in 560 phenotypes. Nat Genet. 2019;51:327.
Obel N, Christensen K, Petersen I, Sørensen TIA, Skytthe A. Genetic and Environmental Influences on Risk of Death due to Infections Assessed in Danish Twins, 1943–2001. Am J Epidemiol. 2010;171:1007–13.
Mozzi A, Pontremoli C, Sironi M. Genetic susceptibility to infectious diseases: Current status and future perspectives from genome-wide approaches. Infect Genet Evol J Mol Epidemiol Evol Genet Infect Dis. 2018;66:286–307.
Tian C, Hromatka BS, Kiefer AK, Eriksson N, Noble SM, Tung JY, et al. Genome-wide association and HLA region fine-mapping studies identify susceptibility loci for multiple common infections. Nat Commun. 2017;8:599.
Pairo-Castineira E, Clohisey S, Klaric L, Bretherick AD, Rawlik K, Pasko D, et al. Genetic mechanisms of critical illness in Covid-19. Nature. Nature Publishing Group; 2020;1–1.
Wei W-Q, Denny JC. Extracting research-quality phenotypes from electronic health records to support precision medicine. Genome Med. 2015;7:41.
Xu H, Stenner SP, Doan S, Johnson KB, Waitman LR, Denny JC. MedEx: a medication information extraction system for clinical narratives. J Am Med Inf Assoc. 2010;17:19–24.
Denny JC, Bastarache L, Ritchie MD, Carroll RJ, Zink R, Mosley JD, et al. Systematic comparison of phenome-wide association study of electronic medical record data and genome-wide association study data. Nat Biotechnol. 2013;31:1102–10.
Wei W-Q, Teixeira PL, Mo H, Cronin RM, Warner JL, Denny JC. Combining billing codes, clinical notes, and medications from electronic health records provides superior phenotyping performance. J Am Med Inform Assoc JAMIA. 2016;23:e20–7.
Hennessy S, Bilker WB, Berlin JA, Strom BL. Factors influencing the optimal control-to-case ratio in matched case-control studies. Am J Epidemiol. 1999;149:195–7.
Chen H-H, Shaw DM, Petty LE, Graff M, Bohlender RJ, Polikowsky HG, et al. Host genetic effects in pneumonia. Am J Hum Genet. 2021;108:194–201.
Das S, Forer L, Schönherr S, Sidore C, Locke AE, Kwong A, et al. Next-generation genotype imputation service and methods. Nat Genet. 2016;48:1284.
Consortium the HR, McCarthy S, Das S, Kretzschmar W, Delaneau O, Wood AR, et al. A reference panel of 64,976 haplotypes for genotype imputation. Nat Genet. 2016;48:1279.
Do R, Willer CJ, Schmidt EM, Sengupta S, Gao C, Peloso GM, et al. Common variants associated with plasma triglycerides and risk for coronary artery disease. Nat Genet. 2013;45:1345–52.
Machiela MJ, Chanock SJ. LDlink: a web-based application for exploring population-specific haplotype structure and linking correlated alleles of possible functional variants. Bioinforma Oxf Engl. 2015;31:3555–7.
Alexander TA, Machiela MJ. LDpop: an interactive online tool to calculate and visualize geographic LD patterns. BMC Bioinformatics. 2020;21:14.
Zhou W, Nielsen JB, Fritsche LG, Dey R, Gabrielsen ME, Wolford BN, et al. Efficiently controlling for case-control imbalance and sample relatedness in large-scale genetic association studies. Nat Genet. 2018;50:1335–41.
Pearce N. Analysis of matched case-control studies. BMJ [Internet]. British Medical Journal Publishing Group; 2016 [cited 2020 Sep 29];352. Available from: https://www.bmj.com/content/352/bmj.i969
Winkler TW, Day FR, Croteau-Chonka DC, Wood AR, Locke AE, Mägi R, et al. Quality control and conduct of genome-wide association meta-analyses. Nat Protoc. 2014;9:1192–212.
Barbeira AN, Dickinson SP, Bonazzola R, Zheng J, Wheeler HE, Torres JM, et al. Exploring the phenotypic consequences of tissue specific gene expression variation inferred from GWAS summary statistics. Nat Commun. 2018;9:1825.
Barbeira AN, Pividori M, Zheng J, Wheeler HE, Nicolae DL, Im HK. Integrating predicted transcriptome from multiple tissues improves association detection. PLOS Genet. Public Library of Science; 2019;15:e1007889.
Denny JC, Ritchie MD, Basford MA, Pulley JM, Bastarache L, Brown-Gentry K, et al. PheWAS: demonstrating the feasibility of a phenome-wide scan to discover gene-disease associations. Bioinformatics. 2010;26:1205–10.
Carroll RJ, Bastarache L, Denny JC. R PheWAS: data analysis and plotting tools for phenome-wide association studies in the R environment. Bioinforma Oxf Engl. 2014;30:2375–6.
Wei W-Q, Bastarache LA, Carroll RJ, Marlo JE, Osterman TJ, Gamazon ER, et al. Evaluating phecodes, clinical classification software, and ICD-9-CM codes for phenome-wide association studies in the electronic health record. PLoS ONE. 2017;12: e0175508.
Wu P, Gifford A, Meng X, Li X, Campbell H, Varley T, et al. Mapping ICD-10 and ICD-10-CM Codes to Phecodes: Workflow Development and Initial Evaluation. JMIR Med Inform. 2019;7: e14325.
Home | GAS Power Calculator [Internet]. [cited 2021 Oct 21]. Available from: https://csg.sph.umich.edu/abecasis/gas_power_calculator/
Gagliano Taliun SA, VandeHaar P, Boughton AP, Welch RP, Taliun D, Schmidt EM, et al. Exploring and visualizing large-scale genetic associations by using PheWeb. Nat Genet Nature Publishing Group. 2020;52:550–2.
COVID-19 Host Genetics Initiative. Mapping the human genetic architecture of COVID-19. Nature. 2021;1–8.
Tosi MF. Innate immune responses to infection. J Allergy Clin Immunol. 2005;116:241–9; quiz 250.
Gaziano L, Giambartolomei C, Pereira AC, Gaulton A, Posner DC, Swanson SA, et al. Actionable druggable genome-wide Mendelian randomization identifies repurposing opportunities for COVID-19. Nat Med. 2021;27:668–76.
Wiegand J, Buggisch P, Boecher W, Zeuzem S, Gelbmann CM, Berg T, et al. Early monotherapy with pegylated interferon alpha-2b for acute hepatitis C infection: The HEP-NET acute-HCV-II study. Hepatology. 2006;43:250–6.
Maughan A, Ogbuagu O. Pegylated interferon alpha 2a for the treatment of hepatitis C virus infection. Expert Opin Drug Metab Toxicol. 2018;14:219–27.
Palumbo E. Pegylated Interferon and Ribavirin Treatment for Hepatitis C Virus Infection. Ther Adv Chronic Dis. 2011;2:39–45.
Monk PD, Marsden RJ, Tear VJ, Brookes J, Batten TN, Mankowski M, et al. Safety and efficacy of inhaled nebulised interferon beta-1a (SNG001) for treatment of SARS-CoV-2 infection: a randomised, double-blind, placebo-controlled, phase 2 trial. Lancet Respir Med. 2021;9:196–206.
Davoudi-Monfared E, Rahmani H, Khalili H, Hajiabdolbaghi M, Salehi M, Abbasian L, et al. A Randomized Clinical Trial of the Efficacy and Safety of Interferon β-1a in Treatment of Severe COVID-19. Antimicrob Agents Chemother. 2020;64:e01061-e1120.
Jagannathan P, Andrews JR, Bonilla H, Hedlin H, Jacobson KB, Balasubramanian V, et al. Peginterferon Lambda-1a for treatment of outpatients with uncomplicated COVID-19: a randomized placebo-controlled trial. Nat Commun. 2021;12:1967.
White DL, Ratziu V, El-Serag HB. Hepatitis C infection and risk of diabetes: a systematic review and meta-analysis. J Hepatol. 2008;49:831–44.
Guo X, Jin M, Yang M, Liu K, Li J. Type 2 diabetes mellitus and the risk of hepatitis C virus infection: a systematic review. Sci Rep. 2013;3:2981.
Roy S, Demmer RT. Impaired glucose regulation, SARS-CoV-2 infections and adverse COVID-19 outcomes. Transl Res J Lab Clin Med. 2022;241:52–69.
Evans KL, Wirtz HS, Li J, She R, Maya J, Gui H, et al. Genetics of heart rate in heart failure patients (GenHRate). Hum Genomics. 2019;13:22.
Eis PS, Huang N, Langston JW, Hatchwell E, Schüle B. Loss-of-Function NUBPL Mutation May Link Parkinson’s Disease to Recessive Complex I Deficiency. Front Neurol. 2020;11: 555961.
Friederich MW, Perez FA, Knight KM, Van Hove RA, Yang SP, Saneto RP, et al. Pathogenic variants in NUBPL result in failure to assemble the matrix arm of complex I and cause a complex leukoencephalopathy with thalamic involvement. Mol Genet Metab. 2020;129:236–42.
Richards AL, Leonenko G, Walters JT, Kavanagh DH, Rees EG, Evans A, et al. Exome arrays capture polygenic rare variant contributions to schizophrenia. Hum Mol Genet. 2016;25:1001–7.
Ishii H, Kim DH, Fujita T, Endo Y, Saeki S, Yamamoto TT. cDNA cloning of a new low-density lipoprotein receptor-related protein and mapping of its gene (LRP3) to chromosome bands 19q12-q13. 2. Genomics. 1998;51:132–5.
Perez Y, Shorer Z, Liani-Leibson K, Chabosseau P, Kadir R, Volodarsky M, et al. SLC30A9 mutation affecting intracellular zinc homeostasis causes a novel cerebro-renal syndrome. Brain J Neurol. 2017;140:928–39.
Generous A, Thorson M, Barcus J, Jacher J, Busch M, Sleister H. Identification of putative interactions between swine and human influenza A virus nucleoprotein and human host proteins. Virol J. 2014;11:228.
We want to acknowledge Synthetic Derivative (SD) and biobank at Vanderbilt University Medical Center (BioVU).
This study was supported by GM120523 (Q.F.), R01HL163854 (Q.F), HL133786 (W.W.), 1K01HL157755-01 (V.E.K.), CSR&D CDA IK2 CX001269 and Merit award I01CX002356 from the US Department of Veterans Affairs (M.J.O.), and Vanderbilt Faculty Research Scholar Fund (Q.F.). The dataset(s) used for the analyses described were obtained from Vanderbilt University Medical Center’s BioVU which is supported by institutional funding, the 1S10RR025141-01 instrumentation award, and by the CTSA grant UL1TR0004from NCATS/NIH. Additional funding provided by the NIH through grants P50GM115305 and U19HL065962. The authors wish to acknowledge the expert technical support of the VANTAGE and VANGARD core facilities, supported in part by the Vanderbilt-Ingram Cancer Center (P30 CA068485) and Vanderbilt Vision Center (P30 EY08126).
Role of the Funder/Sponsor: The funders had no role in design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Ethics approval and consent to participate
The BioVU follows the declaration of Helsinki. Participants of BioVU have signed consent when they agreed to donate their blood sample (DNA) to BioVU biobank. BioVU had then de-identifies those samples and prohibited re-identification as part of its regulations. The current project using BioVU data was approved by the IRB and exempted as “non-human subjects” research. The study was exempted by Vanderbilt University Medical Center Institutional Review Board. The need for informed consent was waived by the ethics committee/Institutional Review Board of Vanderbilt University Medical Center because of the non-human subject nature of the study.
Consent for publication
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Table 1. ICD codes used to definephenotypes.
Supplementary Table 2. List of phenotypes studied from 23andMe paper  and ICD codes. Supplementary Table 3. PheWAS of previous GWAS associations from Tian et al. report. Supplementary Table 4. Associations between genetically predicted gene expression and altered risk of common infections (p<1×10-5). Supplementary Table 5. PheWAS of genetic variants that were associated with common infections in BioVU (suggestive p-value cutoff, 0.001).
Supplementary Figure 1. Manhattan plots of Phenome-wide associations studies.
Supplementary Figure 2. Manhattan plots and Q-Q plots of GWAS results.
About this article
Cite this article
Jiang, L., Kerchberger, V.E., Shaffer, C. et al. Genome-wide association analyses of common infections in a large practice-based biobank. BMC Genomics 23, 672 (2022). https://doi.org/10.1186/s12864-022-08888-9