The proliferation of microbiome research over the past several years has spawned nearly as many questions as it has answered and revealed myriad new frontiers to analyze and understand. One of the most intriguing fields of study is the link between the human microbiome and the pathogenesis of diseases, including diabetes, liver disease, and cancer.
While microbial sequencing provides ample amounts of complex data to study, existing statistical models are ill-equipped to analyze between-subject attributes necessary in many practical applications. As Dr. Tanya Nguyen, an assistant clinical professor of psychiatry and member of the Center for Microbiome Innovation (CMI) at the University of California San Diego (UC San Diego), found while studying the effect of microbiome outcomes on mental health outcomes, a new statistical paradigm was necessary.
Born of this need, Nguyen and her peers at CMI collaborated with a team from the School of Public Health at UC San Diego, led by Xin Tu, Jinyuan Liu, and Xinlian Zhang, as well as researchers at other institutions, to create a statistical modeling framework suitable for analyzing beta-diversity. The group revealed their work in an article entitled “A Semiparametric Model for Between-Subject Attributes: Applications to Beta-diversity of Microbiome Data,” published online by Biometrics.
In the course of their research, they developed a novel approach to analyze the beta-diversity within a regression setting, which they then evaluated on simulated data in comparison with existing models. They also applied their proposed model to a study on alcoholic liver disease, which supported the advantages of the new statistical analysis framework.
“The significance of the research is profound, as it lays a new modeling framework for a new and burgeoning outcome to tackle outcomes of astronomical dimensions in the era of high-throughput data due to scientific and technological advances in biomedical and psychosocial research,” according to Dr. Tu. “The development requires addressing not only various technical issues but also interpretations of models and results as they relate to original research questions.”
Building upon this work, the researchers have several other projects in progress to develop and refine the statistical paradigm necessary to analyze between-subject attributes. Extending the approach to longitudinal studies is a particular focus, but the team sees a wide array of applications in clustered data and general networks.
“As the between-subject attribute is an effective approach to deal with high-dimensional data, which permeate in virtually all scientific domains in the digital age, this new paradigm will play an important role in aiding biomedical and psychosocial research for years to come,” Dr. Tu added while discussing their ongoing work.
In order to facilitate collaborative research, all of the relevant code was published on the Biometrics website, as well as GitHub.
Additional co-authors include Tian Chen at the University of Toledo; Tsung-Chin Wu, Tuo Lin, Lingjing Jiang, Sonja Lang, Lin Liu, Loki Natarajan, Tomasz Kosciolek, Bernd Schnabl, and Rob Knight, all at UC San Diego; Justin Tu at the University of Virginia Health System; James T. Morton at the Simons Foundation; Changyong Feng at the University of Rochester; and Yingchao Zhong at the University of Michigan.
The Center for Microbiome Innovation is proud to include Tanya Nguyen and Bernd Schnabl as faculty members, as well as Rob Knight on its leadership team.
