Lasso logistic regression based approach for extracting plants coregenes responding to abiotic stresses

Sparse methods have a significant advantage of reducing gene expression data complexity to make them comprehensible and interpretable. In this paper, based on Lasso Logistic Regression (LLR), we propose a novel approach to extract plant characteristic gene set, namely coregenes, responding to abiotic stresses. Firstly, to obtain the regression coefficients, the lasso logistic regression was performed according to the samples. Then, the regression coefficients were sorted by the absolute value of them. Finally, the corresponding genes of the nonzero entries of the coefficients are selected as the coregene. Each of coregene extracted can capture the changes of the samples belong to the same condition. The experimental results show that the proposed LLR-based method is efficient to extract the coregenes concerning straight with the stresses.