Qiyao Qin

B.D. in Statistics

Renmin University of China

Student

PLCO (Learn more about this study)

PLCO-943

Mar 22, 2022

Large-Scale Group Testing in Causal Mediation Effects Analysis

As an interdisciplinary field of computer science, biology and statistics, biostatistics is currently one of the most significant fields in technology. In biostatistical studies, human health has increasingly become a vital topic, ranging from laboratory experiments to widespread commercial use. With the progress in genome-wide epigenetic studies, it is of great scientific significance to find out the potential genetic causes of complex diseases. Testing in causal mediation effects analysis is challenging due to a large number of potential mediators. Additionally, considering the natural subgroups of potential mediators, it is important to consider adopting powerful group testing methods. This proposal reviews the traditional structures for testing and the existing test procedures. On this basis, the paper proposes a new group testing procedure that changes the second step of the TS method using the JC method and the adjusted DACT’s statistic.

1. To improve the power of existing group testing algorithms;
2. To extend the powerful single variable testing into group testing to deal with natural sets of biomarkers and take the correlations among mediators into account;
3. To offer the theoretical and mathematical foundations for further mediation effect study;
4. To help biologists find large quantities of meaningful mediators in high-dimensional data to do further research.

And I have already accomplished the modeling and related experiments. So I really need this data to accomplish a practical application study. After this, I will also publish this work!