Search

Abstract Background Uncovering the functional relevance underlying verbal declarative memory (VDM) genome-wide association study (GWAS) results may facilitate the development of interventions to reduce age-related memory decline and dementia. Methods We performed multi-omics and pathway enrichment analyses of paragraph (PAR-dr) and word list (WL-dr) delayed recall GWAS from 29,076 older non-demented individuals of European descent. We assessed the relationship between single-variant associations and expression quantitative trait loci (eQTLs) in 44 tissues and methylation quantitative trait loci (meQTLs) in the hippocampus. We determined the relationship between gene associations and transcript levels in 53 tissues, annotation as immune genes, and regulation by transcription factors (TFs) and microRNAs. To identify significant pathways, gene set enrichment was tested in each cohort and meta-analyzed across cohorts. Analyses of differential expression in brain tissues were conducted for pathway component genes. Results The single-variant associations of VDM showed significant linkage disequilibrium (LD) with eQTLs across all tissues and meQTLs within the hippocampus. Stronger WL-dr gene associations correlated with reduced expression in four brain tissues, including the hippocampus. More robust PAR-dr and/or WL-dr gene associations were intricately linked with immunity and were influenced by 31 TFs and 2 microRNAs. Six pathways, including type I diabetes, exhibited significant associations with both PAR-dr and WL-dr. These pathways included fifteen MHC genes intricately linked to VDM performance, showing diverse expression patterns based on cognitive status in brain tissues. Conclusions VDM genetic associations influence expression regulation via eQTLs and meQTLs. The involvement of TFs, microRNAs, MHC genes, and immune-related pathways contributes to VDM performance in older individuals.

Genome-wide studies (GWS) of SNP associations and differential gene expressions have generated abundant results; next-generation sequencing technology has further boosted the number of variants and genes identified. Effective interpretation requires massive annotation and downstream analysis of these genome-wide results, a computationally challenging task. We developed the snpGeneSets package to simplify annotation and analysis of GWS results. Our package integrates local copies of knowledge bases for SNPs, genes, and gene sets, and implements wrapper functions in the R language to enable transparent access to low-level databases for efficient annotation of large genomic data. The package contains functions that execute three types of annotations: (1) genomic mapping annotation for SNPs and genes and functional annotation for gene sets; (2) bidirectional mapping between SNPs and genes, and genes and gene sets; and (3) calculation of gene effect measures from SNP associations and performance of gene set enrichment analyses to identify functional pathways. We applied snpGeneSets to type 2 diabetes (T2D) results from the NHGRI genome-wide association study (GWAS) catalog, a Finnish GWAS, and a genome-wide expression study (GWES). These studies demonstrate the usefulness of snpGeneSets for annotating and performing enrichment analysis of GWS results. The package is open-source, free, and can be downloaded at: https://www.umc.edu/biostats_software/.

Seven gene sets were significantly enriched for SNP associations with diabetes, and considered as potential diabetes pathways in a previous meta-analysis of diabetes GWAS. This study aims to examine if these gene sets also have expression associations with diabetes. The analysis was conducted using pooled data from 23 diabetes gene expression studies. Gene associations were examined using linear modeling with an empirical Bayes approach, and pathway associations were investigated by testing enrichment for significant genes. Meta-analyses were performed to investigate gene and pathway associations in all studies and tissue types. The analysis showed that six gene sets and three member genes of ACADSB, RASSF2, and KLF12 had significant associations with diabetes traits. The findings suggest that these gene sets are related to diabetes regulation, and their functions tend to be tissue non-specific.

We performed expression studies to identify tissue non-specific genes and pathways of diabetes by meta-analysis. We searched curated datasets of the Gene Expression Omnibus (GEO) database and identified 13 and five expression studies of diabetes and insulin responses at various tissues, respectively. We tested differential gene expression by empirical Bayes-based linear method and investigated gene set expression association by knowledge-based enrichment analysis. Meta-analysis by different methods was applied to identify tissue non-specific genes and gene sets. We also proposed pathway mapping analysis to infer functions of the identified gene sets, and correlation and independent analysis to evaluate expression association profile of genes and gene sets between studies and tissues. Our analysis showed that PGRMC1 and HADH genes were significant over diabetes studies, while IRS1 and MPST genes were significant over insulin response studies, and joint analysis showed that HADH and MPST genes were significant over all combined data sets. The pathway analysis identified six significant gene sets over all studies. The KEGG pathway mapping indicated that the significant gene sets are related to diabetes pathogenesis. The results also presented that 12.8% and 59.0% pairwise studies had significantly correlated expression association for genes and gene sets, respectively; moreover, 12.8% pairwise studies had independent expression association for genes, but no studies were observed significantly different for expression association of gene sets. Our analysis indicated that there are both tissue specific and non-specific genes and pathways associated with diabetes pathogenesis. Compared to the gene expression, pathway association tends to be tissue non-specific, and a common pathway influencing diabetes development is activated through different genes at different tissues. [ABSTRACT FROM AUTHOR]

Background: Genetic heritability and expression study have shown that different diabetes traits have common genetic components and pathways. A computationally efficient pathway analysis of GWAS results will benefit post- GWAS study of SNP associations and identification of common genetic pathways from diabetes GWAS can help to improve understanding of the disease pathogenesis. Results: We proposed a uniform-score gene-set analysis (USGSA) with implemented package to unify different gene measures by a uniform score for identifying pathways from GWAS data, and use a pre-generated permutation distribution table to quickly obtain multiple-testing adjusted p-value. Simulation studies of uniform score for four gene measures (minP, 2ndP, simP and fishP) have shown that USGSA has strictly controlled family-wise error rate. The power depends on types of gene measure. USGSA with a two-stage study strategy was applied to identify common pathways associated with diabetes traits based on public dbGaP GWAS results. The study identified 7 gene sets that contain binding motifs at promoter region of component genes for 5 transcription factors (TFs) of FOXO4, TCF3, NFAT, VSX1 and POU2F1, and 1 microRNA of mi-218. These gene sets include 25 common genes that are among top 5% of the gene associations over genome for all GWAS. Previous evidences showed that nearly all of these genes are mainly expressed in the brain. Conclusions: USGSA is a computationally efficient approach for pathway analysis of GWAS data with promoted interpretability and comparability. The pathway analysis suggested that different diabetes traits share common pathways and component genes are potentially regulated by common TFs and microRNA. The result also indicated that the central nervous system has a critical role in diabetes pathogenesis. The findings will be important in formulating novel hypotheses for guiding follow-up studies. [ABSTRACT FROM AUTHOR]