Wave genome ra is an innovative new technique for identifying genetic variants linked to diseases. Combining GWAS, gene expression analysis, and permutation tests to identify pathways associated with disease, this approach uses permutations tests as an additional feature for identification.
Previous studies employed threshold methods to select SNPs and genes associated with pathways, however this limits identification of moderate-effect pathways.
GWAS
Geneticists use genome-wide association studies (GWASs) as an efficient way of identifying susceptibility variants with significant effects. Yet interpreting their functional significance can be challenging. This is especially relevant for complex diseases, where their biological mechanisms remain obscure. Prioritising and integrating GWAS regions with epigenomic mark and/or gene expression data can assist with identifying key regulatory elements associated with disease risk, such as cis-regulatory motifs, tissue-specific histone marks/chromatin states/transcription start sites that contribute to disease aetiology – evidenced by studies on FTO loci and complement component 4 loci. This approach has proven helpful for uncovering biological processes responsible for disease aetiology as shown by studies on FTO locus and complement component 4 locus studies.
GWAS remains an effective means for discovering novel genes and phenotypes despite these challenges, thanks in large part to researchers’ efforts in creating consortia, collecting large samples in multiethnic populations, as well as employing stringent statistical assumptions when designing meta-analyses and meta-regression analyses based on them. Such measures help prevent effect size inflation that occurs when multiple groups are tested simultaneously.
The initial wave of genome-wide association studies (GWASs) in human health has provided numerous significant discoveries across several disease areas, such as obesity and mental illness. Unfortunately, its success has been hindered by several limitations; for example, many significant associations can be explained away as due to confounding. As a result of this problem, new approaches that aim to identify functionally relevant genes through larger sample sizes while simultaneously determining independent P-values have emerged.
FORGE2 utilizes a quantile-quantile plot (Extended Data Fig. 1) to identify association results for 10 million SNPs using its method, showing that their test statistics are highly inflated, possibly due to stratification or relatedness biases; using an LD score regression intercept, confounding accounts for approximately 7% of this inflation.
Although GWAS in samples with European genetic ancestry has demonstrated impressive predictive power, its use is unlikely to produce similar results among non-European populations. Therefore, it is imperative to develop and use methods capable of applying in non-European populations.
Pathway analysis
Pathway analysis is an integral step in the interpretation of GWAS results, as it allows researchers to investigate associations between pathways and disease, reduce false positives and gain more insight into underlying biology. Furthermore, pathway analysis helps narrow down searches for causal variants by identifying biologically meaningful peaks within large genomic data sets as well as regions needing further examination in order to confirm associations signaled by GWAS data sets.
However, applying this approach in practice can be challenging due to the large number of variables and interactions to analyze. To overcome these hurdles, new methods have been devised that enable for more comprehensive and efficient gene-based analyses – including pathway enrichment analysis and merging GWAS results with gene-based analyses.
This study used a novel approach to identify pathways associated with RA using gene expression data. Specifically, this method employs a pre-selected list of genes and pathways and calculates risk scores by comparing each path with associated phenotypes; pathways with P-values less than 0.05 were considered significant indicators of association with RA.
To conduct this pathway analysis, gene expression data were normalized using R software’s Affy package and then mapped onto their corresponding genomic intervals using PLINK function – this created 1000Mb intervals used by INRICH algorithm which generated 234 pathways enriched for RA that were then examined against predefined lists of genes and pathways for analysis.
Investigation of the enriched pathways was then performed for potential cis-acting regulatory regions, with 60 out of 77 pathways showing at least one significant regulatory region and 16 having more than two significant ones. These findings suggest that these pathways have the potential to be controlled by functionally relevant genetic modulators and may serve as guides for future GWAS/WGS studies. Furthermore, many peaks identified through this analysis were closer to transcription start sites of target genes than expected.
Gene expression analysis
High-density DNA arrays allow scientists to quickly and precisely measure expression levels for large sections of the genome. This represents a revolutionary advance in biology; however, its vast amounts of data require sophisticated storage tools in order to manage and interpret effectively. Gene expression analysis has opened many research avenues such as cardiovascular biology and immunology research.
To identify genes associated with RA-related vascular pathophysiology, we performed gene expression analysis on groups who either improved or worsened during baseline (T0) testing using WGCNA to create co-expression modules that positively or negatively correlated with T0 CDAI score and then tested these modules’ ability to predict subsequent improvement or worsening during pregnancy – the result being an association between gene expression profiles and likelihood of improved/worsened RA during gestation.
All protein-coding and lncRNA genes with CPM > 10 across any of the 6 samples were included in the analysis. Raw counts were loaded into R and normalized using Kallisto (v 0.46.1) and TxImport (v 1.18), to produce gene-level data. An annotation file was then deconvolved using CIBERSORTx with its accompanying LM22 signature matrix in order to estimate cell type proportions within each sample, before being aggregated using featureCounts into gene-level count data that produced normalized counts per gene for analysis by featureCounts analysis of 22 estimated cell types for analysis by featureCounts analysis for normalized counts per gene analysis.
We discovered that many transcripts significantly altered their expression between RAimproved and RAworsened conditions, with 874 genes increasing expression and 1,152 decreasing it (Additional file 3: Table S1). A Manhattan plot of the paired T statistic, represented in Figure 1a; density plot of transcript Pearson correlations between ages 70 and 76 is shown in Figure 1b with correlations for Bonferroni-corrected transcripts that either increased or decreased their expression being displayed while those without any change being shown as black curves.
Gene Ontology (GO) enrichment analysis was performed on IMT-associated genes that displayed at least a twofold change in expression, and this analysis identified multiple overrepresented functional categories including positive regulation of immune response, glucose transport and Golgi vesicle budding. These results indicate that some of these genes may play a role in inflammation processes that contribute to atherosclerosis development among RA patients.
Integrated analysis
Integrative analysis is an efficient method for combining the results of gene expression analysis with those from genome wide association studies (GWAS). By taking an integrative approach to research on RA, it allows researchers to detect pathways related to it; 28 such pathways were discovered this way (FDR0.05). Most are related to immune system functions like natural killer cell mediated cytotoxicity, primary immunodeficiency, Hematopoietic cell lineage development and Fc gamma R-mediated phagocytosis while other paths include B cell receptor signaling pathway or osteoclast differentiation.
This tool uses both genetic and epigenetic information to predict functional impact of variants. To do this, it utilizes SNP Annotation Information List R package (SAILR). Furthermore, an annotation table based on MAFs from 1000 Genomes Project EUR population has also been provided as inputs.
