2019/04/30 by Aniket Biswas, Biswas, Aniket
Biochemistry, Genetics and Molecular Biology · Decision Sciences · Mathematics · #62F10 #62P10 #FOS: Mathematics #Gene expression and cancer classification #Optimal Experimental Design Methods #Statistical Methods and Inference #Statistics Theory (math.ST)
paper · pdf · doi:10.48550/arxiv.1904.13282
openalex publication_date 2019/04/30 · openalex created_date 2022/07/29 · openalex updated_date 2026/07/28
A new formulation for the proportion of true null hypotheses (\π0), based\non the sum of all p-values and the average of expected p-value under the\nfalse null hypotheses has been proposed in the current work. This formulation\nof the parameter of interest \π0 has also been used to construct a new\nestimator for the same. The proposed estimator removes the problem of choosing\ntuning parameters in the existing estimators. Though the formulation is quite\ngeneral, computation of the new estimator demands use of an initial estimate of\n\π0. The issue of choosing an appropriate initial estimator is also\ndiscussed in this work. The current work assumes normality of each gene\nexpression level and also assumes similar tests for all the hypotheses.\nExtensive simulation study shows that, the proposed estimator performs better\nthan its closest competitor, the estimator proposed in Cheng et al., 2015 over\na substantial continuous subinterval of the parameter space, under independence\nand weak dependence among the gene expression levels. The proposed method of\nestimation is applied to two real gene expression level data-sets and the\nresults are in line with what is obtained by the competing method.\n