Group sparse reduced rank regression for neuroimaging genetic study

Abstract

The neuroimaging genetic study usually needs to deal with high dimensionality of both brain imaging data and genetic data, so that often resulting in the issue of curse of dimensionality. In this paper, we propose a group sparse reduced rank regression model to take the relations of both the phenotypes and the genotypes for the neuroimaging genetic study. Specifically, we propose designing a graph sparsity constraint as well as a reduced rank constraint to simultaneously conduct subspace learning and feature selection. The group sparsity constraint conducts feature selection to identify genotypes highly related to neuroimaging data, while the reduced rank constraint considers the relations among neuroimaging data to conduct subspace learning in the feature selection model. Furthermore, an alternative optimization algorithm is proposed to solve the resulting objective function and is proved to achieve fast convergence. Experimental results on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset showed that the proposed method has superiority on predicting the phenotype data by the genotype data, than the alternative methods under comparison.

Coupling environment and physiology to predict effects of climate change on the taxonomic and functional diversity of fish assemblages in the Murray-Darling Basin, Australia

“…As the bioclimatic variable used was air temperature, we used the model relating air temperature and water temperature proposed by [50] to estimate the latter in each grid cell. The lethal maximum temperature for each species was obtained from [51–53].…”

Machine learning-based personalized composite score dissects risk and protective factors for cognitive and motor function in elderly

“…RRR is a computationally efficient method which increases statistical power in settings where the number of dimensions is large compared to the number of examples. In such m ≫ j settings RRR is nowadays a state-of-the-art method in fields with high dimensional data such as genetics and imaging 21,22 . Given j observations of m predictors and n outcome measures, standard multivariate regression requires fitting m · n coefficients Y = XC + E , with Y being the response matrix of size j × n, X being the j × m predictor matrix, C the m × n coefficient matrix and E being the error term matrix of size j × n .…”

“…RRR is a computationally efficient method which increases statistical power in settings where the number of dimensions is large compared to the number of examples. In such 𝑚 ≫ 𝑗 settings RRR is nowadays a state-of-the-art method in fields with high dimensional data such as genetics and imaging 21,22 This decomposition allows for interpretations of 𝐴 and 𝐵. 𝐴 is a mapping from the predictor matrix 𝑋 to a latent representation of dimension 𝑘. 𝐵 is a mapping from the latent scores to the responses 𝑌.…”

Machine learning-based personalized composite score dissects risk and protective factors for cognitive and motor function in elderly

“…RRR is a computationally efficient method which increases statistical power in settings where the number of dimensions is large compared to the number of examples. In such m ≫ j settings RRR is nowadays a state-of-the-art method in fields with high dimensional data such as genetics and imaging (19,20) The latent scores XA display the low-dimensional predictor variability that is predictive of the response variability.…”