- Open Access
Filtering genetic variants and placing informative priors based on putative biological function
© Friedrichs et al. 2016
Published: 3 February 2016
High-density genetic marker data, especially sequence data, imply an immense multiple testing burden. This can be ameliorated by filtering genetic variants, exploiting or accounting for correlations between variants, jointly testing variants, and by incorporating informative priors. Priors can be based on biological knowledge or predicted variant function, or even be used to integrate gene expression or other omics data. Based on Genetic Analysis Workshop (GAW) 19 data, this article discusses diversity and usefulness of functional variant scores provided, for example, by PolyPhen2, SIFT, or RegulomeDB annotations. Incorporating functional scores into variant filters or weights and adjusting the significance level for correlations between variants yielded significant associations with blood pressure traits in a large family study of Mexican Americans (GAW19 data set). Marker rs218966 in gene PHF14 and rs9836027 in MAP4 significantly associated with hypertension; additionally, rare variants in SNUPN significantly associated with systolic blood pressure. Variant weights strongly influenced the power of kernel methods and burden tests. Apart from variant weights in test statistics, prior weights may also be used when combining test statistics or to informatively weight p values while controlling false discovery rate (FDR). Indeed, power improved when gene expression data for FDR-controlled informative weighting of association test p values of genes was used. Finally, approaches exploiting variant correlations included identity-by-descent mapping and the optimal strategy for joint testing rare and common variants, which was observed to depend on linkage disequilibrium structure.
With the availability of very dense genetic marker data sets, such as sequence data, even large association studies can become underpowered. This raises the need to filter, or prioritize, or jointly test genetic variants.
Filters or priors on genes may be derived from methylation or expression data if available in the same individuals. Alternatively, one may use external information. Recently, multiple annotation tools have become available using several databases and algorithms that predict functional effects of genetic variants. Commonly used are, for example, ANNOVAR (Annotate Variation) , VariantTools , PolyPhen , SIFT (Sorting Intolerant From Tolerant) , ENCODE (Encyclopedia of DNA Elements) , RegulomeDB , CADD (Combined Annotation-Dependent Depletion) , or Gerp++ . Tools like ANNOVAR additionally provide variant annotation to genes and to regions such as conserved regions among species, predicted transcription factor binding sites, and segmental duplication regions. Many of the above-listed tools also provide information on regulatory elements that control gene activity. This article demonstrates that functional scores can contribute to the success of association studies. Simultaneously, functional scores may differ substantially between databases and prediction tools as they can be based on different functional aspects. Additionally, variant annotations to chromosomal positions continue to be updated with the National Center for Biotechnology Information (NCBI)  human genome build as standard. Furthermore, variants can be annotated to genes based on different sources, such as ENSEMBL , Vega , GENCODE , and many more. Researchers also use a variety of definitions of flanking regions. Finally, genes may be grouped by function or biological pathway, again with substantial variability between data bases such as KEGG , Biocarta , or Pathway Interaction Database . This article discusses approaches that filtered or prioritized genetic variants, regions, or genes. Pathway-based approaches, although also incorporating filters or priors, are discussed separately by Kent .
Many researchers filter genetic variants. The simplest forms of filters are minor allele frequency (MAF), candidate genes or variants, or considering the exome. Filters and statistical models are chosen to increase the power under a hypothetical disease model. The advent of sequencing renewed interest in disease mechanisms less frequent but more penetrant than common single nucleotide polymorphisms (SNPs) of genome-wide association studies (GWAS). This led, for example, to screening for recessive variants by examining runs of homozygosity , . When multiple rare causal variants cluster within a gene, identity-by-descent (IBD) mapping may be more powerful than single-locus association testing . IBD mapping can be used in 2-step approaches. For example, Balliu et al  identified regions where hypertension cases shared more segments of IBD than controls in one part of the sample. They modeled aggregate effects of each of these regions on blood pressure (BP) in the sample remainder. Aggregation tests are used especially for testing rare single-nucleotide variants (SNVs). Aggregation tests are burden tests, variance-component tests, or a combination of both, such as SKAT-O (optimal unified sequence kernel association test) (see, eg, Lee et al  for a review). Kernel-based approaches (see Schaid  for a review) such as the sequence kernel association test (SKAT)  are variance-component tests. Examples of genetic burden tests are T5, combined multivariate collapsing (CMC) , or C-α ; see also Santorico et al . Aggregation tests can prioritize SNVs by weighting minor allele dosages in the test statistic. Typical weights account for MAF, but may also incorporate putative functional relevance of SNVs , . Moreover, weights may be used to combine aggregation test statistics , , , and one may weight p values while controlling the false discovery rate (FDR) , . For example, GWAS p values may be weighted based on functional annotations. For aggregation tests on genes, p value weights can be utilized to integrate gene expression or other omics data .
Statistical tests and analyzed data
Almeida et al 
Single-variant regression in SOLAR
Smoking, BP medication, PC1-3, sex, age, age2, sex*age, sex*age2
Real SBP and DBP at first time point, own simulated trait for H0
Liu et al 
Chr3: GWASmp and sequence
Unrelated individuals (from family study)
Regress pairwise DBP residual difference and sum on IBD sharing status; sequence data analyses by SKAT-O
Sex, age, smoking, PC 1-3
Real DBP at first time point
Kim and Wei 
Informative SNV weights in burden test T5 and SKAT; with R: seqMeta
Age, sex, smoking, BP medication
Real SBP at earliest available measurement
Zhang et al 
Unrelated individuals (large Hispanic sample)
LRT, C-, CMC on informatively weighted SNV burden
Simulated HT status; real SBP, DBP with cutoffs for case-control status
Malzahn et al 
Sequence and GWASmp
SKAT with R (coxme, kinship2, QuadCompForm); strategies for joint testing of rare and common SNVs
Sex, age, sex*age; subjects not on BP medication
Real and simulated SBP at first time point
Ho et al 
Sequence and GWASmp
Family study, including gene expression data
Seq-aSum-VS burden test; regression on gene expression data; gene set enrichment analysis
Average real SBP and DBP
Filters, priors, and findings
Almeida et al 
Functional annotation, LD-corrected effective number of tests
LD-correction in WGS reduces multiple-testing burden by 85 %, significant associations: PFH14 with SBP, MAP4 with DBP
Location: ANNOVAR; functional annotation: PolyPhen, SIFT
Liu et al 
No significances, ZPLD1 had strongest evidence
IBD mapping: BEAGLE; functional annotation: CADD
Kim and Wei 
Sliding window on MAF ≤5 % SNVs
SNV-weights: based on MAF or regulatory importance
Significant association: SNUPN
Functional annotation: ENCODE, RegulomeDB, PolyPhen2
Zhang et al 
SNV-weights: up-weight protein binding sites, apply direction weights
Top-ranked genes differ between weighted burden tests LRT, C-α, CMC; but good overlap with literature
ANNOVAR, variant tools; random forest classifiers assign SNVs to protein binding sites; DSSP, PSAIA, DOMINO
Malzahn et al 
Gene covering LD-blocks
SNV-weights: using MAF
SKAT: power depends on SNV weights, exploiting LD is very beneficial, optimal strategy for joint testing rare and common SNVs depends on LD structure
Haploview with HapMap data for LD-calculation
Overall weight: on rare SNV variance component in SKAT
Ho et al 
Rare SNVs in genes with >1 and <50 rare SNVs (MAF < 0.01)
p value weights: improve gene ranking
Power of burden tests improved by incorporating phenotype associated gene expression into p value weights
Genes: hg19; GO biological process categories
Analyzed data were provided by GAW 19 and included a family sample (n = 959) with extended pedigrees of Mexican Americans from the San Antonio Family Heart Study (SAFHS) and the San Antonio Family Diabetes/Gallbladder Study (SAFDS/ SAFGS) . The family sample also contained 103 unrelated sequenced subjects; 259 subjects had gene expression data. This study was designed to identify low-frequency or rare variants influencing susceptibility to type 2 diabetes (T2D) as part of the T2D Genetic Exploration by Next-generation sequencing in Ethnic Samples (T2D-GENES) Consortium. Phenotypes included real and simulated longitudinal systolic (SBP) and diastolic blood pressure (DBP) and hypertension (HT) status. Available were sequence for 464 pedigree members and GWAS SNPs for all 959 subjects. Additionally, all subjects were imputed to sequence based on original genotypes and familial relationships . Approaches described herein mostly analyzed imputed dosages to avoid missing genotypes and to maximize sample size. Zhang et al  analyzed the GAW19 sample of 1943 independent Hispanic subjects with whole exome sequence. This sample had been ascertained by T2D status. However, GAW19 provided real and simulated cross-sectional BP traits instead , using the same trait-simulation model as for the family study.
All approaches described herein are nonlongitudinal analyses of BP traits (SBP, DBP, or HT) in relation to minor allele dosages of sequence SNVs or genome-wide SNPs.
Statistical methods employed by this group (see Table 1) to incorporate filters or informative priors are mostly based on regression models , , , , ; one is also based on counting methods . Analyses of family data adjusted for familial dependence based on the kinship matrix. They included the familial covariance in a linear mixed model , ,  or transformed the trait to a conditionally independent surrogate variable . Analyses of independent subjects accounted for population structure (cryptic relatedness and admixture)  by using the programs Eigensoft  and Admixture .
Annotating genetic variants for location and function
A variety of freely available genetic databases and highly developed software tools support annotation of location and biological function of SNVs. In our group, SNV locations were obtained by ANNOVAR ,  or determined based on reference data, for example, from the Genome Reference Consortium  or the International Haplotype Map (HapMap) Consortium  , . Reference data were also used to determine linkage disequilibrium (LD) blocks  with Haploview .
Kim and Wei  and Almeida et al  used functional annotations from ENCODE, PolyPhen or PolyPhen2, and SIFT, while Liu et al  used CADD. In contrast, Zhang et al  annotated putative protein binding sites based on 2 different algorithms using random forest classifiers .
Filtering genetic variants
Not all areas of the genome were studied. Some researchers filtered the data prior to analyses. Zhang et al  investigated exome sequence and Almeida et al  molecularly functional nonsynonymous SNVs predicted by PolyPhen and SIFT. Liu et al  examined IBD sharing regions on chromosome 3. Malzahn et al  considered gene-containing LD blocks for selected candidate genes. Ho et al  analyzed rare SNV burden in genes containing less than 50 and more than 1 rare SNV (MAF <0.01).
Accounting for correlations between genetic variants
An important difference between methods is that variant correlations can either be a nuisance or may be used to increase power. For example, IBD mapping exploits variant correlations. IBD mapping can be more powerful than single-locus association testing when multiple causal rare variants cluster within a gene . Therefore, Liu et al  tested the relationship between IBD sharing status and trait differences and sums for pairs of individuals. Moreover, the power of kernel methods such as SKAT may be increased through the exploitation of variant correlations . This ability can be utilized fully by analyzing LD blocks . On the other hand, single-locus methods need to account for variant correlations to appropriately correct the significance level for multiple testing. Hence, Almeida et al  determined the effective number of independent tests by extreme value theory based on replicates of a simulated unassociated trait.
Correcting the significance level for the number of independent tests
The significance level used with multiple testing is always an issue as too conservative a correction will cause false negatives and not correcting enough will cause false positives.
Almeida et al  adjusted the significance level for single locus analyses by estimating the number of independent tests . A total of 1000 replicates of a quantitative phenotype with no genetic effects were simulated and tested on whole genome sequence data, using linear mixed models in SOLAR (Sequential Oligogenic Linkage Analysis Routines) . The smallest p value per simulation run was extracted. The density of these 1000 extremely small p values was fitted to a theoretical beta distribution beta(1,n e ) where n e is the effective number of independent tests ; yielding the adjusted significance level . This procedure was applied to both whole genome sequence and functional nonsynonymous SNVs.
IBD mapping aims to detect loci sharing ancestral segments in unrelated individuals. In particular, unrelated subject-pairs with smaller trait differences are expected to share significantly more rare causative variants than pairs with larger trait differences. Liu et al  estimated IBD sharing segments with BEAGLE . The squared trait difference (D) and squared trait sum (S) for trait DBP between pairs of unrelated subjects was regressed on IBD sharing status. This yielded parameter estimates for slopes and variances (σ S 2 , σ D 2 ), which were combined into an overall slope estimate . Linkage was tested with test statistic under the null hypothesis of an overall slope of zero . The significance threshold for nonindependent pairs was estimated by permutation procedure.
Priors on genes and variants
Genetic priors can be incorporated by variant weights in aggregation tests such as burden tests or SKAT . Burden tests collapse minor allele dosages x ik of a set of i = 1, …, m variants into a burden score s k = ∑ i = 1 m ω i x ik per individual k using a priori specified variant weights ω i . One tests trait association with genetic burden s k . Although burden tests are powerful when causal SNVs have the same effect direction, SKAT is more powerful when effect directions differ or if many noncausal SNVs are included in testing , . SKAT is based on an underlying Bayesian model that estimates a random effect per SNV . Specified is a kernel matrix of genetic between-subject similarity and this kernel constitutes a prior on genetic model space . SNV weights are incorporated in the kernel (see, eg, Malzahn et al ).
Typically, rarer SNVs get assigned more weight to counterbalance their reduced power compared to more frequent SNVs. Used are, for example, weights , inverse MAF weights , or beta-weights such as ωj = b(MAFi) , where b is the probability density function of a beta(1, 25) random variable. Malzahn et al  compared the power of SKAT when using different SNV weights and different kernel functions that either allow or do not allow for SNV interactions in the genetic model. Alternatively, SNV weights may be based on regulatory importance  or protein binding effects .
Incorporating functional information into variant weights
Kim and Wei  categorized SNVs according to RegulomeDB and PolyPhen2 functional relevance scores. SNV weights were defined based on f(s) = S 2 where s equaled the reverse order of categories, namely s = 6, 5, 4, 3, 2, 1 for category 1 (“most likely affecting binding and expression”) to category 6 (“not functionally relevant”). Kim and Wei  tested rare SNVs jointly, in sets defined by sliding windows of 4 kb size, for association with SBP. They compared the power of SNV weighting schemes in SKAT ( versus ω j = b(MAFj)), and burden test T5 (ω j = f(s j ) versus ). SKAT and T5 provide analytical asymptotically exact p values with good small sample size behavior.
Zhang et al  used a likelihood ratio test (LRT)  to test if the proportion of subjects with an informatively weighted minor allele burden exceeding a given threshold differed between HT cases and controls. P values were obtained by permutation procedure. SNV weights ω i accounted for putative effect direction and distinguished between functional SNVs in binding-sites (|ω i | = 10), not in binding-sites (|ω i | = 5), and nonfunctional SNVs (|ω i | = 1). The informatively weighted LRT was compared with C-α and CMC burden tests.
Optimal joint testing of rare and common variants
weights the rare SNV variance-component by overall a priori weight (1-λ) relative to the common SNV variance-component (see Ionita-Laza et al  and Malzahn et al  for choices of λ). The weighted sum test (1) is another way of structuring a prior in SKAT. Note that Qrare and Qcommon may use different kernel functions or different SNV weights. Malzahn et al  compared this form of joint testing of rare and common SNVs with the default choice of entering all SNVs with appropriate weights into a single kernel. Exact p values for SKAT and weighted sum test (1) were obtained by Davies method . Another investigated alternative was Fisher pooling of the correlated p values resulting from the separate rare SNV and common SNV SKAT statistics. Fisher pooling accounted for correlations by Satterthwaite approximation and Brown’s method (; see also , ).
Informed p value weighting for genes
Results and discussion
The results for this GAW19 working group varied widely as a result of the different objectives of each contributor. Table 2 provides a brief summary of specific results.
Under H 0, extreme p values follow a beta distribution . Almeida et al  reported that the beta distribution provided an excellent fit to determine the effective number of independent tests n e for n single-locus tests. For whole genome sequence, ; that is, accounting for LD reduced the multiple-testing burden by 85 %. However, significant associations could only be found when LD-correcting the significance level after a priori reducing sequence data based on functional annotations. Then 2 SNPs were detected: rs218966 in gene PHF14 associated with SBP and rs9836027 in MAP4 associated with DBP.
Liu et al  scanned chromosome 3 (GWAS data) for IBD sharing segments that associated with DBP. No genome-wide significance was found. However, several risk variants were detected in the region of gene ZPLD1 by using CADD functional scores and sequence for the most promising region at 3q12.3.
When using gene expression data to informatively weight gene-wise p values for association of rare SNV burden with BP , 153 genes (out of 6118) reached nominal significance (weighted p ≤0.05). P value weights were determined such that evidence for phenotype associated gene expression lowered burden test p values. As no gene reached multiple-testing adjusted significance, Ho et al  used gene set enrichment analysis as aggregation test to relate the 153 top genes to biological pathways.
All analyses presented herein used a cross-sectional design by analyzing trait data of the first examination, the first available examination, or longitudinally averaged traits. This mainly contributed to differences in sample size and trait variability. Furthermore, analyzing trait values at different time points may affect the marginal effect of genes that interact with age.
Including biological knowledge increased the power of association studies performed in our GAW group; especially filtering variants based on putative functional relevance. Prior weights can be included at different stages of the testing procedure. They can be incorporated into the test statistic of SKAT or burden tests, used when combining test statistics, or applied to association test p values. Selecting variant-sets also should take genetic structures into consideration, such as LD or IBD sharing. Moreover, the effective number of independent tests can be determined relatively easily by extreme value theory. This enables appropriate adjustment of the significance level for multiple testing to avoid an overly conservative approach. Ideally, variant grouping and selection, inclusion of biological information, and significance level adjustment can be applied simultaneously. Strategies like these are useful in increasing power in analyses of highly dense genetic data sets.
Filtering variants clearly boosted power in the discussed studies. However, filtering might also lose information. Functional scores such as PolyPhen2, SIFT, CADD, or RegulomeDB differ as they focus on different information about SNVs. Moreover, appropriateness of functional scores for a considered trait is a priori unknown. Hence, one is well advised to use and combine multiple functional annotations into a single filter or prior. This is feasible as functional annotations yield strong filters that greatly reduce the SNV space.
The authors would like to thank Zheyang Wu and Peng Wei for their comments and suggestions, as well as the GAW organizers for all their efforts. SF and DM were supported by the Deutsche Forschungsgemeinschaft (DFG, grant Research Training Group “Scaling Problems in Statistics” RTG 1644; grant Klinische Forschergruppe (KFO) 241: TP5, BI 576/5-1). EWP and JNB acknowledge support by National Institutes of Health (NIH) grants (HHSN268201200008I, R01 NS055057). XQL was supported by the University of Manitoba start-up funds. T2D-GENES is supported by NIH grants U01 DK085524, U01 DK085501, U01 DK085526, U01 DK085584 and U01 DK085545, the SAFHS by grant P01 HL045222, the SAFDS by grant R01 DK047482, and the SAFGS by grant R01 DK053889. Genetic Analysis Workshop 19 was supported by NIH grant R01 GM031575.
This article has been published as part of BMC Genetics Volume 17 Supplement 2, 2016: Genetic Analysis Workshop 19: Sequence, Blood Pressure and Expression Data. Summary articles. The full contents of the supplement are available online at www.biomedcentral.com/bmcgenet/supplements/17/S2. Publication of the proceedings of Genetic Analysis Workshop 19 was supported by National Institutes of Health grant R01 GM031575.
- Wang K, Li M, Hakonarson H: ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 2010, 38: e164-e164. 10.1093/nar/gkq603.View ArticlePubMedPubMed CentralGoogle Scholar
- San Lucas FA, Wang G, Scheet P, Peng B: Integrated annotation and analysis of genetic variants from next-generation sequencing studies with variant tools. Bioinformatics. 2012, 28: 421-2. 10.1093/bioinformatics/btr667.View ArticlePubMedGoogle Scholar
- Adzhubei I, Jordan DM, Sunyaev SR: Predicting functional effect of human missense mutations using PolyPhen-2. Curr Protoc Hum Genet. 2013, Chapter 7: Unit 7.20-Google Scholar
- Kumar P, Henikoff S, Ng PC: Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm. Nat Protoc. 2009, 4: 1073-81. 10.1038/nprot.2009.86.View ArticlePubMedGoogle Scholar
- Kellis M, Wold B, Snyder MP, Bernstein BE, Kundaje A, Marinov GK, et al: Defining functional DNA elements in the human genome. Proc Natl Acad Sci U S A. 2014, 111: 6131-8. 10.1073/pnas.1318948111.View ArticlePubMedPubMed CentralGoogle Scholar
- Boyle AP, Hong EL, Hariharan M, Cheng Y, Schaub MA, Kasowski M, et al: Annotation of functional variation in personal genomes using RegulomeDB. Genome Res. 2012, 22: 1790-7. 10.1101/gr.137323.112.View ArticlePubMedPubMed CentralGoogle Scholar
- Kircher M, Witten DM, Jain P, O’Roak BJ, Cooper GM, Shendure J: A general framework for estimating the relative pathogenicity of human genetic variants. Nat Genet. 2014, 46: 310-5. 10.1038/ng.2892.View ArticlePubMedPubMed CentralGoogle Scholar
- Davydov EV, Goode DL, Sirota M, Cooper GM, Sidow A, Batzoglou S: Identifying a high fraction of the human genome to be under selective constraint using GERP++. PLoS Comput Biol. 2010, 6: e1001025-10.1371/journal.pcbi.1001025.View ArticlePubMedPubMed CentralGoogle Scholar
- NCBI: National center for biotechnology information search database. http://www.ncbi.nlm.nih.gov/
- Cunningham F, Amode MR, Barrell D, Beal K, Billis K, Brent S, et al: Ensembl 2015. Nucleic Acids Res. 2015, 43 (Database issue): D662-9. 10.1093/nar/gku1010.View ArticlePubMedGoogle Scholar
- Harrow JL, Steward CA, Frankish A, Gilbert JG, Gonzalez JM, Loveland JE, et al: The vertebrate genome annotation browser 10 years on. Nucleic Acids Res. 2014, 42: D771-9. 10.1093/nar/gkt1241.View ArticlePubMedGoogle Scholar
- Harrow J, Frankish A, Gonzalez JM, Tapanari E, Diekhans M, Kokocinski F, et al: GENCODE: the reference human genome annotation for the ENCODE project. Genome Res. 2012, 22: 1760-74. 10.1101/gr.135350.111.View ArticlePubMedPubMed CentralGoogle Scholar
- Kanehisa M, Goto S: KEGG: Kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 2000, 28: 27-30. 10.1093/nar/28.1.27.View ArticlePubMedPubMed CentralGoogle Scholar
- Nishimura D: BioCarta. Biotech Softw Internet Rep. 2001, 2: 117-20. 10.1089/152791601750294344.View ArticleGoogle Scholar
- Schaefer CF, Anthony K, Krupa S, Buchoff J, Day M, Hannay T, et al: PID: the pathway interaction database. Nucleic Acids Res. 2009, 37: D674-9. 10.1093/nar/gkn653.View ArticlePubMedGoogle Scholar
- Kent JW: Pathway-based analyses. BMC Genet. 2015, 16 Suppl 3: S5-Google Scholar
- Gibson J, Morton NE, Collins A: Extended tracts of homozygosity in outbred human populations. Hum Mol Genet. 2006, 15: 789-95. 10.1093/hmg/ddi493.View ArticlePubMedGoogle Scholar
- Hildebrandt F, Heeringa SF, Rüschendorf F, Attanasio M, Nürnberg G, Becker C, et al: A systematic approach to mapping recessive disease genes in individuals from outbred populations. PLoS Genet. 2009, 5: e1000353-10.1371/journal.pgen.1000353.View ArticlePubMedPubMed CentralGoogle Scholar
- Browning SR, Thompson EA: Detecting rare variant associations by identity-by-descent mapping in case-control studies. Genetics. 2012, 190: 1521-31. 10.1534/genetics.111.136937.View ArticlePubMedPubMed CentralGoogle Scholar
- Balliu B, Uh HW, Tsonaka R, Boehringer S, Helmer Q, Houwing-Duistermaat JJ: Combining information from linkage and association mapping for next-generation sequencing longitudinal family data. BMC Proc. 2014, 8 (Suppl 1): S34-10.1186/1753-6561-8-S1-S34.View ArticlePubMedPubMed CentralGoogle Scholar
- Lee S, Abecasis GR, Boehnke M, Lin X: Rare-variant association analysis: study designs and statistical tests. Am J Hum Genet. 2014, 95: 5-23. 10.1016/j.ajhg.2014.06.009.View ArticlePubMedPubMed CentralGoogle Scholar
- Schaid DJ: Genomic similarity and kernel methods I: advancements by building on mathematical and statistical foundations. Hum Hered. 2010, 70: 109-31. 10.1159/000312641.View ArticlePubMedGoogle Scholar
- Wu MC, Lee S, Cai T, Li Y, Boehnke M, Lin X: Rare variant association testing for sequencing data using the sequence kernel association test (SKAT). Am J Hum Genet. 2011, 89: 82-93. 10.1016/j.ajhg.2011.05.029.View ArticlePubMedPubMed CentralGoogle Scholar
- Li B, Leal SM: Methods for detecting associations with rare variants for common diseases: application to analysis of sequence data. Am J Hum Genet. 2008, 83: 311-21. 10.1016/j.ajhg.2008.06.024.View ArticlePubMedPubMed CentralGoogle Scholar
- Neale BM, Rivas MA, Voight BF, Altshuler D, Devlin B, Orho-Melander M, et al: Testing for an unusual distribution of rare variants. PLoS Genet. 2011, 7: e1001322-10.1371/journal.pgen.1001322.View ArticlePubMedPubMed CentralGoogle Scholar
- Santorico SA, Hendricks AE: Progress in methods for rare variant association. BMC Genet. 2015, 16 Suppl 3: S7-Google Scholar
- Kim T, Wei P: Incorporating ENCODE information into association analysis of whole genome sequencing data. BMC Proc. 2015, 9 (Suppl 8): S34-Google Scholar
- Zhang D, Cui H, Korkin D, Wu Z: Incorporation of protein binding effects into likelihood ratio test for exome sequencing data. BMC Proc. 2015, 9 Suppl 8: S37-Google Scholar
- Ionita-Laza I, Lee S, Makarov V, Buxbaum JD, Lin X: Sequence kernel association tests for the combined effect of rare and common variants. Am J Hum Genet. 2013, 92: 841-53. 10.1016/j.ajhg.2013.04.015.View ArticlePubMedPubMed CentralGoogle Scholar
- Malzahn D, Friedrichs S, Bickeböller H: Comparing strategies for combined testing of rare and common variants in whole sequence and genome-wide genotype data. BMC Proc. 2015, 9 Suppl 8: S36-Google Scholar
- Genovese CR, Roeder K, Wasserman L: False discovery control with p-value weighting. Biometrika. 2006, 93: 509-24. 10.1093/biomet/93.3.509.View ArticleGoogle Scholar
- Roeder K, Wasserman L: Genome-wide significance levels and weighted hypothesis testing. Stat Sci. 2009, 24: 398-413. 10.1214/09-STS289.View ArticlePubMedPubMed CentralGoogle Scholar
- Ho YY, Guan W, Basu S: Powerful association test combining rare variant and gene expression using family data from genetic analysis workshop 19. BMC Proc. 2015, 9 Suppl 8: S33-Google Scholar
- Almasy L, Dyer TD, Peralta JM, Jun G, Wood AR, Fuchsberger C, et al: Data for Genetic Analysis Workshop 18: human whole genome sequence, blood pressure, and simulated phenotypes in extended pedigrees. BMC Proc. 2014, 8 (Suppl 1): S2-10.1186/1753-6561-8-S1-S2.View ArticlePubMedPubMed CentralGoogle Scholar
- Blangero J, Teslovich TM, Sim X, Almeida MA, Jun G, Dyer TD, et al: Omics-squared: human genomic, transcriptomic and phenotypic data for Genetic Analysis Workshop 19. BMC Proc. 2015, 9 Suppl 8: S2-Google Scholar
- Almeida M, Blondell L, Peralta J, Kent JW, Jun G, Teslovich TM, et al: Independent test assessment using the extreme value distribution theory. BMC Proc. 2015, 9 Suppl 8: S32-Google Scholar
- Liu X-Q, Fazio J, Hu PZ, Paterson AD: Identity-by-descent mapping for diastolic blood pressure in unrelated Mexican Americans. BMC Proc. 2015, 9 Suppl 8: S35-Google Scholar
- Patterson N, Price AL, Reich D: Population structure and eigenanalysis. PLoS Genet. 2006, 2: e190-10.1371/journal.pgen.0020190.View ArticlePubMedPubMed CentralGoogle Scholar
- Alexander DH, Novembre J, Lange K: Fast model-based estimation of ancestry in unrelated individuals. Genome Res. 2009, 19: 1655-64. 10.1101/gr.094052.109.View ArticlePubMedPubMed CentralGoogle Scholar
- GRC: The Genome Reference Consortium. http://www.ncbi.nlm.nih.gov/projects/genome/assembly/grc/
- The International HapMap Consortium: The international HapMap project. Nature. 2003, 426: 789-96. 10.1038/nature02168.View ArticleGoogle Scholar
- Barrett JC, Fry B, Maller J, Daly MJ: Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005, 21: 263-5. 10.1093/bioinformatics/bth457.View ArticlePubMedGoogle Scholar
- Sikić M, Tomić S, Vlahovicek K: Prediction of protein-protein interaction sites in sequences and 3D structures by random forests. PLoS Comput Biol. 2009, 5 (1): e1000278-10.1371/journal.pcbi.1000278.View ArticlePubMedPubMed CentralGoogle Scholar
- Schifano ED, Epstein MP, Bielak LF, Jhun MA, Kardia SL, Peyser P, et al: SNP set association analysis for familial data. Genet Epidemiol. 2012, 36: 797-810.PubMedPubMed CentralGoogle Scholar
- Dudbridge F, Gusnanto A: Estimation of significance thresholds for genome wide association scans. Genet Epidemiol. 2008, 32: 227-34. 10.1002/gepi.20297.View ArticlePubMedPubMed CentralGoogle Scholar
- Almasy L, Blangero J: Multipoint quantitative-trait linkage analysis in general pedigrees. Am J Hum Genet. 1998, 62: 1198-211. 10.1086/301844.View ArticlePubMedPubMed CentralGoogle Scholar
- Sidak Z: Rectangular confidence regions from means of multivariate normal distributions. J Am Stat Assoc. 1967, 62: 626-33.Google Scholar
- Browning SR, Browning BL: Rapid and accurate haplotype phasing and missing data inference for whole genome association studies by use of localized haplotype clustering. Am J Hum Genet. 2007, 81: 1084-97. 10.1086/521987.View ArticlePubMedPubMed CentralGoogle Scholar
- Chen H, Malzahn D, Balliu B, Li C, Bailey JN: Testing genetic association with rare and common variants in family data. Genet Epidemiol. 2014, 38 (Suppl 1): S37-43. 10.1002/gepi.21823.View ArticlePubMedPubMed CentralGoogle Scholar
- Liu D, Lin X, Ghosh G: Semiparametric regression of multidimensional genetic pathway data: least-squares kernel machines and linear mixed models. Biometrics. 2007, 63: 1079-88. 10.1111/j.1541-0420.2007.00799.x.View ArticlePubMedPubMed CentralGoogle Scholar
- Rasmussen CE, Williams CKI: Gaussian processes for machine learning. 2006, MIT Press, CambridgeGoogle Scholar
- Madsen BE, Browning SR: A groupwise association test for rare mutations using a weighted sum statistic. PLoS Genet. 2009, 5: e1000384-10.1371/journal.pgen.1000384.View ArticlePubMedPubMed CentralGoogle Scholar
- Chen Y-C, Carter H, Parla J, Kramer M, Goes FS, Pirooznia M, et al: A hybrid likelihood model for sequence-based disease association studies. PLoS Genet. 2013, 9: e1003224-10.1371/journal.pgen.1003224.View ArticlePubMedPubMed CentralGoogle Scholar
- Davies RB: Algorithm as 155: the distribution of a linear combination of chi-2 random variables. J R Stat Soc: Ser C: Appl Stat. 1980, 29: 323-33.Google Scholar
- Brown MB: A method for combining non-independent, one-sided tests of significance. Biometrics. 1975, 31: 987-92. 10.2307/2529826.View ArticleGoogle Scholar
- Lee S, Wu MC, Lin X: Optimal tests for rare variant effects in sequencing association studies. Biostatistics. 2012, 13: 762-75. 10.1093/biostatistics/kxs014.View ArticlePubMedPubMed CentralGoogle Scholar
- Basu S, Pan W: Comparison of statistical tests for disease association with rare variants. Genet Epidemiol. 2011, 35: 606-19. 10.1002/gepi.20609.View ArticlePubMedPubMed CentralGoogle Scholar
This article is published under license to BioMed Central Ltd. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.