Canonical correlation analysis for multi-omics: Application to cross-cohort analysis.

Integrative approaches that simultaneously model multi-omics data have gained increasing popularity because they provide holistic system biology views of multiple or all components in a biological system of interest. Canonical correlation analysis (CCA) is a correlation-based integrative method designed to extract latent features shared between multiple assays by finding the linear combinations of features-referred to as canonical variables (CVs)-within each assay that achieve maximal across-assay correlation. Although widely acknowledged as a powerful approach for multi-omics data, CCA has not been systematically applied to multi-omics data in large cohort studies, which has only recently become available. Here, we adapted sparse multiple CCA (SMCCA), a widely-used derivative of CCA, to proteomics and methylomics data from the Multi-Ethnic Study of Atherosclerosis (MESA) and Jackson Heart Study (JHS). To tackle challenges encountered when applying SMCCA to MESA and JHS, our adaptations include the incorporation of the Gram-Schmidt (GS) algorithm with SMCCA to improve orthogonality among CVs, and the development of Sparse Supervised Multiple CCA (SSMCCA) to allow supervised integration analysis for more than two assays. Effective application of SMCCA to the two real datasets reveals important findings. Applying our SMCCA-GS to MESA and JHS, we identified strong associations between blood cell counts and protein abundance, suggesting that adjustment of blood cell composition should be considered in protein-based association studies. Importantly, CVs obtained from two independent cohorts also demonstrate transferability across the cohorts. For example, proteomic CVs learned from JHS, when transferred to MESA, explain similar amounts of blood cell count phenotypic variance in MESA, explaining 39.0% ~ 50.0% variation in JHS and 38.9% ~ 49.1% in MESA. Similar transferability was observed for other omics-CV-trait pairs. This suggests that biologically meaningful and cohort-agnostic variation is captured by CVs. We anticipate that applying our SMCCA-GS and SSMCCA on various cohorts would help identify cohort-agnostic biologically meaningful relationships between multi-omics data and phenotypic traits.

Whole genome sequence analysis of pulmonary function and COPD in 19,996 multi-ethnic participants.

Chronic obstructive pulmonary disease (COPD), diagnosed by reduced lung function, is a leading cause of morbidity and mortality. We performed whole genome sequence (WGS) analysis of lung function and COPD in a multi-ethnic sample of 11,497 participants from population- and family-based studies, and 8499 individuals from COPD-enriched studies in the NHLBI Trans-Omics for Precision Medicine (TOPMed) Program. We identify at genome-wide significance 10 known GWAS loci and 22 distinct, previously unreported loci, including two common variant signals from stratified analysis of African Americans. Four novel common variants within the regions of PIAS1, RGN (two variants) and FTO show evidence of replication in the UK Biobank (European ancestry n ~ 320,000), while colocalization analyses leveraging multi-omic data from GTEx and TOPMed identify potential molecular mechanisms underlying four of the 22 novel loci. Our study demonstrates the value of performing WGS analyses and multi-omic follow-up in cohorts of diverse ancestry.

Genetic variation in the base excision repair pathway, environmental risk factors, and colorectal adenoma risk.

Cigarette smoking, high alcohol intake, and low dietary folate levels are risk factors for colorectal adenomas. Oxidative damage caused by these three factors can be repaired through the base excision repair pathway (BER). We hypothesized that genetic variation in BER might modify colorectal adenoma risk. In a sigmoidoscopy-based study, we examined associations between 182 haplotype tagging SNPs in 14 BER genes, and colorectal adenoma risk, and examined their potential role as modifiers of the effect cigarette smoking, alcohol intake, and dietary folate levels. Among all individuals, no statistically significant associations between BER SNPs and adenoma risk persisted after correction for multiple comparisons. However, among Asian-Pacific Islanders we observed two SNPs in FEN1 and one in NTHL1, and among African-Americans one SNP in APEX1 that were associated with colorectal adenoma risk. Significant associations were also observed between SNPs in the NEIL2 gene and rectal adenoma risk. Three SNPS modified the effect of smoking (MUTYH interaction p = 0.002; OGG1 interaction p = 0.013); FEN1 interaction p = 0.013)), one SNP in LIG3 modified the effect of alcohol consumption (interaction p = 0.024) and two SNPs in LIG3 modified the effect of dietary folate (interaction p = 0.001 and p = 0.08) on colorectal adenoma risk. These findings support a role for genetic variants in the BER pathway as potential modifiers of colorectal adenoma risk. Our findings strengthen the role of oxidative damage induced by key lifestyle and dietary risk factors in colorectal adenoma formation.

Genetic variation in insulin pathway genes and distal colorectal adenoma risk.

BACKGROUND: Insulin, glucose, and other insulin-related proteins that mediate insulin signaling are associated with colorectal neoplasia risk, but associations with common genetic variation in insulin axis genes are less clear. In this study, we used a comprehensive tag single-nucleotide polymorphisms (SNPs) approach to define genetic variation in six insulin axis genes (IGF1, IGF2, IGFBP1, IGFBP3, IRS1, and IRS2) and three genes associated with estrogen signaling (ESR1, ESR2, and PGR). METHODS: We assessed associations between SNPs and distal colorectal adenoma (CRA) risk in a case-control study of 1,351 subjects. Cases were individuals with one or more adenomas diagnosed during sigmoidoscopy, and controls were individuals with no adenomas at the sigmoidoscopy exam. We used unconditional logistic regression assuming an additive model to assess SNP-specific risks adjusting for multiple comparisons with P (act). RESULTS: Distal adenoma risk was significantly increased for one SNP in IGF2 [per minor allele OR = 1.41; 95 % confidence interval (CI) = 1.16, 1.67; P (act) = 0.005] and decreased for an ESR2 SNP (per minor allele OR = 0.78; 95 % CI = 0.66, 0.91; P (act) = 0.041). There was no statistically significant heterogeneity of these associations by race, sex, BMI, physical activity, or, in women, hormone replacement therapy use. Risk estimates did not differ in the colon versus rectum or for smaller (<1 cm) versus larger (>1 cm) adenomas. CONCLUSIONS: These data suggest that selected genetic variability in IGF2 and ESR2 may be modifiers of CRA risk.

Heterogenous effect of androgen receptor CAG tract length on testicular germ cell tumor risk: shorter repeats associated with seminoma but not other histologic types.

Increasing rates of testicular germ cells tumors (TGCTs) overtime suggest that environmental factors are involved in disease etiology, but familial risk and genome-wide association studies implicate genetic factors as well. We investigated whether variation in the functional CAG(n) polymorphism in the androgen receptor (AR) gene is associated with TGCT risk, using data from a population-based family study. We estimated odds ratios (OR) and 95% confidence intervals (CI) for the association of CAG repeat length and TGCT risk using matched pairs logistic regression. Analyses of 273 TGCT case-mother pairs revealed no association between AR CAG repeat length and overall TGCT risk. However, risk of seminoma was significantly associated with shorter CAG repeat length [CAG 20-21 versus CAG ≤ 19: OR = 0.82 (95% CI: 0.43-1.58), CAG 22-23 versus CAG ≤ 19: OR = 0.39 (95% CI: 0.19-0.83) and CAG ≥ 24 versus CAG ≤ 19: OR = 0.42 (95% CI: 0.20-0.86)], with a highly significant trend over these four categories of decreasing CAG repeat length (P(trend) = 0.0030). This is the first report of a statistically significant association between AR CAG repeat length and seminoma risk, suggesting that increased AR transactivation may be involved in development of seminoma and/or progression of carcinoma in situ/intratubular germ cell neoplasia unclassified to seminoma. This result provides a rationale whereby androgenic environmental compounds could contribute to increases in TGCT incidence, and identifies for the first time a potential biological pathway influencing whether TGCTs achieve seminomatous versus nonseminomatous histology, a clinically and biologically important distinction.

Variation in folate pathway genes and distal colorectal adenoma risk: a sigmoidoscopy-based case-control study.

BACKGROUND: Folate-associated one-carbon metabolism (FOCM) is an important pathway in colorectal neoplasia risk but data on genetic variation in this pathway are largely limited to studies of single SNPs in selected genes. METHODS: We used a comprehensive tagSNP approach to study the association between genetic variation in 11 genes in the FOCM pathway and risk of incident distal colorectal adenomas in a sigmoidoscopy-based case-control study. We included 655 cases (one or more adenomas) and 695 controls (no adenomas) recruited from one of two Kaiser Permanente clinics between 1991 and 1995. We assessed a total of 159 tagSNPs selected using Haploview Tagger as well as selected non-synonymous SNPs. We used unconditional logistic regression to model the association between SNPs and risk of distal adenomas, assuming a log-additive model. RESULTS: Five SNPs in the SLC19A1 (RFC1) gene: rs1051266 (G80A), rs283895, rs2236484, rs12482346, and rs2838958 were associated with adenoma risk after correction for multiple testing (all corrected p values ≤ 0.043). The non-synonymous SLC19A1 SNP G80A interacted significantly with the MTHFR C677T genotype (interaction p value = 0.018). CONCLUSION: Our data suggest that genetic variation in SLC19A1 may modify the risk of distal colorectal adenoma.