Imputing genotypes using regularized generalized linear regression models.

As genomic sequencing technologies continue to advance, researchers are furthering their understanding of the relationships between genetic variants and expressed traits. However, missing data can significantly limit the power of a genetic study. Here, the use of a regularized generalized linear model, denoted by GLMNET, is proposed to impute missing genotypes. The method aims to address certain limitations of earlier regression approaches in regards to genotype imputation, particularly the specification of the number of neighboring SNPs to be included for imputing the missing genotype. The performance of GLMNET-based method is compared to the conventional multinomial regression method and two phase-based methods: fastPHASE and BEAGLE. Two simulation scenarios are evaluated: a sparse-missing model, and a small-panel expansion model. The sparse-missing model simulates a scenario where SNPs were missing in a random fashion across the genome. In the small-panel expansion model, a set of individuals is only genotyped at a subset of the SNPs of the large panel. Each imputation method is tested in the context of two data-sets: Canadian Holstein cattle data and human HapMap CEU data. Results show that the proposed GLMNET method outperforms the other methods in the small panel expansion scenario and fastPHASE performs slightly better than the GLMNET method in the sparse-missing scenario.

Improving monitoring of fish health in the oil sands region using regularization techniques and water quality variables.

Trout-perch are sampled from the Athabasca River in Alberta, Canada, as a sentinel species for environmental health. The performance of trout-perch populations is known to be influenced by the quality of the water in which they reside. Using climate, environmental, and water quality variables measured in the Athabasca River near trout-perch sampling locations is found to improve model fitting and the predictability of models for the adjusted body weight, adjusted gonad weight, and adjusted liver weight of trout-perch. Given a large number of covariables, three variable selection techniques: stepwise regression, the lasso, and the elastic net (EN) are considered for selecting a subset of relevant variables. The models selected by the lasso and EN are found to outperform the models selected by stepwise regression in general, and little difference is observed between the models selected by the lasso and EN. Uranium, tungsten, tellurium, pH, molybdenum, and antimony are selected for at least one fish response.

A Parallel Sliding Region Algorithm to Make Agent-Based Modeling Possible for a Large-Scale Simulation: Modeling Hepatitis C Epidemics in Canada.

Agent-based models (ABMs) are computer simulation models that define interactions among agents and simulate emergent behaviors that arise from the ensemble of local decisions. ABMs have been increasingly used to examine trends in infectious disease epidemiology. However, the main limitation of ABMs is the high computational cost for a large-scale simulation. To improve the computational efficiency for large-scale ABM simulations, we built a parallelizable sliding region algorithm (SRA) for ABM and compared it to a nonparallelizable ABM. We developed a complex agent network and performed two simulations to model hepatitis C epidemics based on the real demographic data from Saskatchewan, Canada. The first simulation used the SRA that processed on each postal code subregion subsequently. The second simulation processed the entire population simultaneously. It was concluded that the parallelizable SRA showed computational time saving with comparable results in a province-wide simulation. Using the same method, SRA can be generalized for performing a country-wide simulation. Thus, this parallel algorithm enables the possibility of using ABM for large-scale simulation with limited computational resources.

The LASSO and sparse least square regression methods for SNP selection in predicting quantitative traits.

Recent work concerning quantitative traits of interest has focused on selecting a small subset of single nucleotide polymorphisms (SNPs) from amongst the SNPs responsible for the phenotypic variation of the trait. When considered as covariates, the large number of variables (SNPs) and their association with those in close proximity pose challenges for variable selection. The features of sparsity and shrinkage of regression coefficients of the least absolute shrinkage and selection operator (LASSO) method appear attractive for SNP selection. Sparse partial least squares (SPLS) is also appealing as it combines the features of sparsity in subset selection and dimension reduction to handle correlations amongst SNPs. In this paper we investigate application of the LASSO and SPLS methods for selecting SNPs that predict quantitative traits. We evaluate the performance of both methods with different criteria and under different scenarios using simulation studies. Results indicate that these methods can be effective in selecting SNPs that predict quantitative traits but are limited by some conditions. Both methods perform similarly overall but each exhibit advantages over the other in given situations. Both methods are applied to Canadian Holstein cattle data to compare their performance.