2016/05/27 by John Palowitch, Andrey Shabalin, Palowitch, John +9
Biochemistry, Genetics and Molecular Biology · Mathematics · #Artificial intelligence #Biology #Computational biology #Computer science #Expression quantitative trait loci #FOS: Computer and information sciences #Gene #Gene expression and cancer classification #Genetic Mapping and Diversity in Plants and Animals #Genetic and phenotypic traits in livestock #Genetics #Genotype #Linear model #Machine learning #Methodology (stat.ME) #stat.ME
paper · pdf · doi:10.48550/arxiv.1605.08799
published in arXiv (Cornell University) (Cornell University)
openalex publication_date 2016/05/27 · arxiv created 2017/09/07 · arxiv updated 2017/09/08 · openalex created_date 2022/10/01 · openalex updated_date 2026/08/05
The study of expression Quantitative Trait Loci (eQTL) is an important\nproblem in genomics and biomedicine. While detection (testing) of eQTL\nassociations has been widely studied, less work has been devoted to the\nestimation of eQTL effect size. To reduce false positives, detection methods\nfrequently rely on linear modeling of rank-based normalized or log-transformed\ngene expression data. Unfortunately, these approaches do not correspond to the\nsimplest model of eQTL action, and thus yield estimates of eQTL association\nthat can be uninterpretable and inaccurate. In this paper we propose a new,\nlog-of-linear model for eQTL action, termed ACME, that captures allelic\ncontributions to cis-acting eQTLs in an additive fashion, yielding effect size\nestimates that correspond to a biologically coherent model of cis-eQTLs. We\ndescribe a non-linear least-squares algorithm to fit the model by maximum\nlikelihood, and obtain corresponding p-values. We perform careful\ninvestigation of the model using a combination of simulated data and data from\nthe Genotype Tissue Expression (GTEx) project. Our results reveal little\nevidence for dominance effects, a parsimonious result that accords with a\nsimple biological model for allele-specific expression and supports use of the\nACME model. We show that Type-I error is well-controlled under our approach in\na realistic setting, so that rank-based normalizations are unnecessary.\nFurthermore, we show that such normalizations can be detrimental to power and\nestimation accuracy under the proposed model. We then provide summaries of ACME\neffect sizes for whole-genome cis-eQTLs in the GTEx data.\n