Catalog
- Results include
-
Gene hunting with knockoffs for hidden Markov models
Sesia, MatteoStanford, California : Department of Statistics, Stanford University, June 2017.Online statistics.stanford.edu
-
New methods for variable importance testing with applications to genetic studies
Sesia, Matteo[Stanford, California] : [Stanford University], 2020The objective of this thesis is to develop new practical and principled statistical methodology for the analysis of genome-wide association data, in order to identify, as precisely as possible, the genetic variants that affect complex phenotypes. This problem can be stated as one of testing multiple hypotheses of conditional independence between many possible explanatory variables and a response of interest, within a high-dimensional non-parametric regression setting. This dissertation builds upon previous work on knockoffs, which provides a general framework for addressing such variable importance testing problems. In particular, we study how to generate valid knockoffs for genetic variants, while taking into account the particular structure of these data and the hidden Markov models developed by geneticists to describe their distribution. As a result, we can obtain an effective and statistically rigorous tool for genetic mapping that controls the false discovery rate under minimal assumptions, while overcoming many of the limitations of the existing state-of-the-art methods. Extensive numerical experiments with genetic data confirm the empirical validity and effectiveness of our method, while applications to the analysis of large data sets lead to many new discoveries
-
Deep knockoffs
Romano, YanivStanford, Calif. : Department of Statistics, Stanford University, December 2018.Online statistics.stanford.edu
Guides
Library website
Exhibits
EarthWorks
More search tools
Tools to help you discover resources at Stanford and beyond.