Health risk factors such as body mass index (BMI) and serum cholesterol are associated with many common diseases. It often remains unclear whether the risk factors are cause or consequence of disease, or whether the associations are the result of confounding. We develop and apply a method (called GSMR) that performs a multi-SNP Mendelian randomization analysis using summary-level data from genome-wide association studies to test the causal associations of BMI, waist-to-hip ratio, serum cholesterols, blood pressures, height, and years of schooling (EduYears) with common diseases (sample sizes of up to 405,072). We identify a number of causal associations including a protective effect of LDL-cholesterol against type-2 diabetes (T2D) that might explain the side effects of statins on T2D, a protective effect of EduYears against Alzheimer’s disease, and bidirectional associations with opposite effects (e.g., higher BMI increases the risk of T2D but the effect of T2D on BMI is negative).

Endometriosis is a heritable hormone-dependent gynecological disorder, associated with severe pelvic pain and reduced fertility; however, its molecular mechanisms remain largely unknown. Here we perform a meta-analysis of 11 genome-wide association case-control data sets, totalling 17,045 endometriosis cases and 191,596 controls. In addition to replicating previously reported loci, we identify five novel loci significantly associated with endometriosis risk (P<5 × 10⁻⁸), implicating genes involved in sex steroid hormone pathways (FN1, CCDC170, ESR1, SYNE1 and FSHB). Conditional analysis identified five secondary association signals, including two at the ESR1 locus, resulting in 19 independent single nucleotide polymorphisms (SNPs) robustly associated with endometriosis, which together explain up to 5.19% of variance in endometriosis. These results highlight novel variants in or near specific genes with important roles in sex steroid hormone signalling and function, and offer unique opportunities for more targeted functional research efforts.

Precise prediction for genetic architecture of complex traits is impeded by the limited understanding on genetic effects of complex traits, especially on gene-by-gene (GxG) and gene-by-environment (GxE) interaction. In the past decades, an explosion of high throughput technologies enables omics studies at multiple levels (such as genomics, transcriptomics, proteomics and metabolomics). The analyses of large omics data, especially two-loci interaction analysis, are very time intensive. Integrating the diverse omics data and environmental effects in the analyses also remain challenges. We proposed mixed linear model approaches using GPU (Graphic Processing Unit) computation to simultaneously dissect various genetic effects. Analyses can be performed for estimating genetic main effects, GxG epistasis effects and GxE environment interaction effects on large-scale omics data for complex traits and for estimating heritability of specific genetic effects. Both mouse data analyses and Monte Carlo simulations demonstrated that genetic effects and environment interaction effects could be unbiasedly estimated with high statistical power by using the proposed approaches.

The critical barrier in interaction analysis for rare variants is that most traditional statistical methods for testing interactions were originally designed for testing the interaction between common variants and are difficult to apply to rare variants because of their prohibitive computational time and poor ability. The great challenges for successful detection of interactions with next-generation sequencing (NGS) data are (1) lack of methods for interaction analysis with rare variants, (2) severe multiple testing, and (3) time-consuming computations. To meet these challenges, we shift the paradigm of interaction analysis between two loci to interaction analysis between two sets of loci or genomic regions and collectively test interactions between all possible pairs of SNPs within two genomic regions. In other words, we take a genome region as a basic unit of interaction analysis and use high-dimensional data reduction and functional data analysis techniques to develop a novel functional regression model to collectively test interactions between all possible pairs of single nucleotide polymorphisms (SNPs) within two genome regions. By intensive simulations, we demonstrate that the functional regression models for interaction analysis of the quantitative trait have the correct type 1 error rates and a much better ability to detect interactions than the current pairwise interaction analysis. The proposed method was applied to exome sequence data from the NHLBI’s Exome Sequencing Project (ESP) and CHARGE-S study. We discovered 27 pairs of genes showing significant interactions after applying the Bonferroni correction (P-values < 4.58 × 10⁻¹⁰) in the ESP, and 11 were replicated in the CHARGE-S study.

Gene co-expression networks comprise one type of valuable biological networks. Many methods and tools have been published to construct gene co-expression networks; however, most of these tools and methods are inconvenient and time consuming for large datasets. We have developed a user-friendly, accelerated and optimized tool for constructing gene co-expression networks that can fully harness the parallel nature of GPU (Graphic Processing Unit) architectures. Genetic entropies were exploited to filter out genes with no or small expression changes in the raw data preprocessing step. Pearson correlation coefficients were then calculated. After that, we normalized these coefficients and employed the False Discovery Rate to control the multiple tests. At last, modules identification was conducted to construct the co-expression networks. All of these calculations were implemented on a GPU. We also compressed the coefficient matrix to save space. We compared the performance of the GPU implementation with those of multi-core CPU implementations with 16 CPU threads, single-thread C/C++ implementation and single-thread R implementation. Our results show that GPU implementation largely outperforms single-thread C/C++ implementation and single-thread R implementation, and GPU implementation outperforms multi-core CPU implementation when the number of genes increases. With the test dataset containing 16,000 genes and 590 individuals, we can achieve greater than 63 times the speed using a GPU implementation compared with a single-thread R implementation when 50 percent of genes were filtered out and about 80 times the speed when no genes were filtered out.