Sangin Lee

Korean, Japanese, and Chinese populations featured similar genes encoding drug-metabolizing enzymes and transporters: a DMET Plus microarray assessment.

Background Interethnic differences in genetic polymorphism in genes encoding drug-metabolizing enzymes and transporters are one of the major factors that cause ethnic differences in drug response. This study aimed to investigate genetic polymorphisms in genes involved in drug metabolism, transport, and excretion among Korean, Japanese, and Chinese populations, the three major East Asian ethnic groups. Methods The frequencies of 1936 variants representing 225 genes encoding drug-metabolizing enzymes and transporters were determined from 786 healthy participants (448 Korean, 208 Japanese, and 130 Chinese) using the Affymetrix Drug-Metabolizing Enzymes and Transporters Plus microarray. To compare allele or genotype frequencies in the high-dimensional data among the three East Asian ethnic groups, multiple testing, principal component analysis (PCA), and regularized multinomial logit model through least absolute shrinkage and selection operator were used. Results On microarray analysis, 1071 of 1936 variants (>50% of markers) were found to be monomorphic. In a large number of genetic variants, the fixation index and Pearson’s correlation coefficient of minor allele frequencies were less than 0.034 and greater than 0.95, respectively, among the three ethnic groups. PCA identified 47 genetic variants with multiple testing, but was unable to discriminate ethnic groups by the first three components. Multinomial least absolute shrinkage and selection operator analysis identified 269 genetic variants that showed different frequencies among the three ethnic groups. However, none of those variants distinguished between the three ethnic groups during subsequent PCA. Conclusion Korean, Japanese, and Chinese populations are not pharmacogenetically distant from one another, at least with regard to drug disposition, metabolism, and elimination.

A modified local quadratic approximation algorithm for penalized optimization problems

In this paper, we propose an optimization algorithm called the modified local quadratic approximation algorithm for minimizing various ? 1 -penalized convex loss functions. The proposed algorithm iteratively solves ? 1 -penalized local quadratic approximations of the loss function, and then modifies the solution whenever it fails to decrease the original ? 1 -penalized loss function. As an extension, we construct an algorithm for minimizing various nonconvex penalized convex loss functions by combining the proposed algorithm and convex concave procedure, which can be applied to most nonconvex penalty functions such as the smoothly clipped absolute deviation and minimax concave penalty functions. Numerical studies show that the algorithm is stable and fast for solving high dimensional penalized optimization problems.

Moderately clipped LASSO

The least absolute shrinkage and selection operator (LASSO) has been widely used in high-dimensional linear regression models. However, it is known that the LASSO selects too many noisy variables. In this paper, we propose a new estimator, the moderately clipped LASSO (MCL), that deletes noisy variables successively without sacrificing prediction accuracy much. Various numerical studies are done to illustrate superiority of the MCL over other competitors.

Sparse optimization for nonconvex group penalized estimation

We consider a linear regression model where there are group structures in covariates. The group LASSO has been proposed for group variable selections. Many nonconvex penalties such as smoothly clipped absolute deviation and minimax concave penalty were extended to group variable selection problems. The group coordinate descent (GCD) algorithm is used popularly for fitting these models. However, the GCD algorithms are hard to be applied to nonconvex group penalties due to computational complexity unless the design matrix is orthogonal. In this paper, we propose an efficient optimization algorithm for nonconvex group penalties by combining the concave convex procedure and the group LASSO algorithm. We also extend the proposed algorithm for generalized linear models. We evaluate numerical efficiency of the proposed algorithm compared to existing GCD algorithms through simulated data and real data sets.

A random-effect model approach for group variable selection

We consider regression models with a group structure in explanatory variables. This structure is commonly seen in practice, but it is only recently realized that taking the information into account in the modeling process may improve both the interpretability and accuracy of the model. In this paper, we study a new approach to group variable selection using random-effect models. Specific distributional assumptions on random effects pertaining to a given structure lead to a new class of penalties that include some existing penalties. We also develop an efficient computational algorithm. Numerical studies are provided to demonstrate better sensitivity and specificity properties without sacrificing the prediction accuracy. Finally, we present some real-data applications of the proposed approach.

Sparse pathway-based prediction models for high-throughput molecular data

Pathway-based prediction problems for high-throughput molecular data motivate the development of sparsity-constrained models with structured predictive variables. Intuitively it is desirable to incorporate the structural information into the model building procedure, potentially for improving both interpretability and prediction performances. Various random-effect models are developed for structured sparse prediction where the predictive variables/genes can be naturally grouped into overlapping groups or pathways. The hierarchical likelihood approach can be used for these random-effect models that impose sparse selection of the overlapping groups as well as further selection within the selected groups. In addition, the approach leads to a unified optimization algorithm for these random-effect models. Extensive numerical studies based on simulated and real breast-cancer data demonstrate that the proposed methods perform well against existing methods that ignore the structural information.

Sparse estimation of gene-gene interactions in prediction models.

Current assessment of gene–gene interactions is typically based on separate parallel analysis, where each interaction term is tested separately, while less attention has been paid on simultaneous estimation of interaction terms in a prediction model. As the number of interaction terms grows fast, sparse estimation is desirable from statistical and interpretability reasons. There is a large literature on sparse estimation, but there is a natural hierarchy between the interaction and its corresponding main effects that requires special considerations. We describe random-effect models that impose sparse estimation of interactions under both strong and weak-hierarchy constraints. We develop an estimation procedure based on the hierarchical-likelihood argument and show that the modelling approach is equivalent to a penalty-based method, with the advantage of the models being more transparent and flexible. We compare the procedure with some standard methods in a simulation study and illustrate its application in an analysis of gene–gene interaction model to predict body-mass index.