Computational method identifies genetic drivers of microbiome diversity

From the Wu Group, Public Health Sciences Division

“We are well aware that the microbiome plays a critical role in human health, but our understanding of what exactly shapes the microbiome is very incomplete,” explained Pearl (Hongjiao) Liu, a graduate student in Michael Wu’s group in the Public Health Sciences Divsion. She added, “the microbiome is widely regarded as modifiable,” which can be shaped by environmental factors like diet and medication, but “there are also genetic factors driving the composition.” These genetic factors “represent variation in the microbiome that is not really changeable,” yet how they influence one’s microbiome is poorly understood. For instance, a recent large-scale genome-wide association study (GWAS) sought to identify genes that could influence the microbiome, but found only a single locus that may dictate microbial composition. Researchers in the Wu group suspect there are likely many genes capable of  influencing a person’s microbiome that are being missed in studies such as this. In their recently published Microbiome paper, spear-headed by Liu, the research team sought “to develop strategies for more powerful mapping of genetic-microbiome relationships,” Liu noted.

Our microbiome is an integral component of our overall health and has been shown to be “involved in fundamental body functions such as metabolism and immune response” and can influence “various diseases such as obesity, type 2 diabetes and inflammatory bowel diseases,” Liu said. “As the microbiome is composed of many microbes and functions as a community, it is useful to analyze the microbiome composition from a particular body site (e.g., gut) as a whole.” A person’s gut microbiome, for example, is unique to that individual and the “difference in overall microbial profiles among individuals, as measured by beta-diversity, can be associated with differences in various health outcomes,” Liu explained. Beta-diversity can indicate how dissimilar overall microbial profiles are between individuals, where “a high level of beta-diversity between two individuals means that there is a considerable difference in their overall microbial profiles,” added Liu. “More conceptually, beta-diversity characterizes and emphasizes differences among microbiomes (in people) based on overall structure rather than just looking at individual microbes.”

Overview of kernel RV framework to identify genetic drivers of microbial composition.
Overview of kernel RV framework to identify genetic drivers of microbial composition. Image taken from original article.

To identify genetic drivers of microbial beta-diversity, the researchers needed to develop a method that would have enough statistical power to map these complex gene-microbiome relationships and for this reason, turned to a covariate-adjusted kernel RV (KRV) framework. Kernel-RV “is a statistic for measuring and testing the dependency between two groups of variables, such as a group of genetic variants and the microbiome,” Liu stated. In previous microbiome GWAS studies, researchers “studied the association between one genetic variant and one microbial taxon at a time,” however since there are millions of genetic variants and “hundreds of microbial taxa within the microbial community, such an approach will cause a huge multiple-testing burden and result in a low statistical power,” Liu noted. To increase statistical power, the Wu team chose to “instead evaluate the association between a group of genetic variants within a gene and the overall microbiome composition, measured by beta-diversity, at the community level.” This novel approach allowed the researchers to “reduce the multiple-testing burden and account for unique characteristics of genetic and microbiome data, e.g., interaction among genetic variants and phylogenetic relationships among microbial taxa. These benefits of the KRV framework can improve the statistical power and help us identify a greater number of genetic loci that are associated with the human microbiome.”

To test whether this method would have the power to do so, Liu et al. applied it to the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) cohort, “one of the most comprehensive studies of Hispanic/Latino populations in the USA,” said Liu. Narrowing in on a specific population was biologically important for identifying genes that influence microbiome composition since “different populations tend to differ systematically in their genetic and microbial profiles and can also have different genetics-microbiome relationships.” When the Wu group put their novel method into practice, the researchers were excited to discover four genes in the HCHS/SOL cohort that were “associated with the overall gut microbiome composition (or gut microbiome beta-diversity),” none of which were “identified using two other competing methods that focus on the same type of genetic features (i.e., genes associated with the overall microbiome composition). Therefore, these results confirm the power improvement of our proposed approach and contribute exciting novel findings to further understand the genetics-microbiome relationship.” While some of these associations “replicated previous, less-robust, findings,” others were previously unknown. “Three of the four genes that were identified in our microbiome GWAS analysis” were previously shown to be involved in “immune functions or immunity-related diseases, suggesting that there might be an important role of immunity-related genes in shaping the gut microbiome composition.” One of these identified genes, “IL23R, is a pro-inflammatory cytokine closely involved in autoimmunity,” which has documented associations with “both the gut microbiome and inflammatory bowel diseases (IBD). Coupled with previous studies, our finding further supports that the gut microbiome may mediate the host genetic effect on the development of inflammatory diseases like IBD,” Liu exclaimed.

The goal of the Hispanic Community Health Study/Study of Latinos cohort was to identify risk factors for health outcomes in Hispanic/Latino populations in the US and through analysis of this data, “our study is the first to investigate the genetic effect on the overall gut microbiome composition in Hispanic/Latino populations,” stated Liu. “The results from our study will help inform important genetic risk factors for gut-microbiome-related health outcomes in Hispanic/Latino individuals.” This powerful method expands the traditional approach of “studying the association between one genetic variant and one taxon at a time” to now examine “the association between variants within a gene and the overall microbiome composition,” and will be an asset for routine microbiome analysis. One of the biggest impacts of this methodological work however is “that we are developing the tools to help other researchers in their projects. Accordingly, we believe that our framework will directly enable other groups to more powerfully identify genetic features associated with microbiome composition. Thus, our impact is not necessarily our own deployment of our methods, but the eventual impact of others’ studies.”

Fred Hutch/UW/Seattle Children’s Cancer Consortium member Michael Wu contributed to this work.

This work was supported by the National Institutes of Health, the National Heart, Lung, and Blood Institute, the National Institute on Minority Health and Health Disparities, National Institute on Deaf and Other Communication Disorders, the National Institute of Dental and Craniofacial Research, National Institute of Diabetes and Digestive and Kidney Diseases, National Institute of Neurological Disorders and Stroke, and the NIH Institution-Office of Dietary Supplements.

Liu H, Ling W, Hua X, Moon JY, Williams-Nguyen JS, Zhan X, Plantinga AM, Zhao N, Zhang A, Knight R, Qi Q, Burk RD, Kaplan RC, Wu MC. 2023. Kernel-based genetic association analysis for microbiome phenotypes identifies host genetic drivers of beta-diversity. Microbiome. 11(1):80.