High-dimensional log-error-in-variable regression with applications to microbial compositional data analysis

Summary In microbiome and genomic studies, the regression of compositional data has been a crucial tool for identifying microbial taxa or genes that are associated with clinical phenotypes. To account variation in sequencing depth, classic log-contrast model is often used where read counts normalized into compositions. However, zero randomness covariates remain critical issues. We introduce surprisingly simple, interpretable efficient method estimation through lens novel high-dimensional log-error-in-variable model. The proposed provides corrections on possible overdispersion simultaneously avoids any subjective imputation counts. provide theoretical justifications matching upper lower bounds error. merit procedure illustrated real analysis simulation studies.