Finding the molecular basis of complex genetic variation in humans and mice

Effect of genotype imputation on genome-enabled prediction of complex traits: an empirical study with mice data

Background

Genotype imputation is an important tool for whole-genome prediction as it allows cost reduction of individual genotyping. However, benefits of genotype imputation have been evaluated mostly for linear additive genetic models. In this study we investigated the impact of employing imputed genotypes when using more elaborated models of phenotype prediction. Our hypothesis was that such models would be able to track genetic signals using the observed genotypes only, with no additional information to be gained from imputed genotypes.

Results

For the present study, an outbred mice population containing 1,904 individuals and genotypes for 1,809 pre-selected markers was used. The effect of imputation was evaluated for a linear model (the Bayesian LASSO - BL) and for semi and non-parametric models (Reproducing Kernel Hilbert spaces regressions – RKHS, and Bayesian Regularized Artificial Neural Networks – BRANN, respectively). The RKHS method had the best predictive accuracy. Genotype imputation had a similar impact on the effectiveness of BL and RKHS. BRANN predictions were, apparently, more sensitive to imputation errors. In scenarios where the masking rates were 75% and 50%, the genotype imputation was not beneficial. However, genotype imputation incorporated information about important markers and improved predictive ability, especially for body mass index (BMI), when genotype information was sparse (90% masking), and for body weight (BW) when the reference sample for imputation was weakly related to the target population.

Conclusions

In conclusion, genotype imputation is not always helpful for phenotype prediction, and so it should be considered in a case-by-case basis. In summary, factors that can affect the usefulness of genotype imputation for prediction of yet-to-be observed traits are: the imputation accuracy itself, the structure of the population, the genetic architecture of the target trait and also the model used for phenotype prediction.

Prediction of body mass index in mice using dense molecular markers and a regularized neural network

Bayesian regularization of artificial neural networks (BRANNs) were used to predict body mass index (BMI) in mice using single nucleotide polymorphism (SNP) markers. Data from 1896 animals with both phenotypic and genotypic (12 320 loci) information were used for the analysis. Missing genotypes were imputed based on estimated allelic frequencies, with no attempt to reconstruct haplotypes based on family information or linkage disequilibrium between markers. A feed-forward multilayer perceptron network consisting of a single output layer and one hidden layer was used. Training of the neural network was done using the Bayesian regularized backpropagation algorithm. When the number of neurons in the hidden layer was increased, the number of effective parameters, γ, increased up to a point and stabilized thereafter. A model with five neurons in the hidden layer produced a value of γ that saturated the data. In terms of predictive ability, a network with five neurons in the hidden layer attained the smallest error and highest correlation in the test data although differences among networks were negligible. Using inherent weight information of BRANN with different number of neurons in the hidden layer, it was observed that 17 SNPs had a larger impact on the network, indicating their possible relevance in prediction of BMI. It is concluded that BRANN may be at least as useful as other methods for high-dimensional genome-enabled prediction, with the advantage of its potential ability of capturing non-linear relationships, which may be useful in the study of quantitative traits under complex gene action.

Genome and proteome annotation: organization, interpretation and integration

Recent years have seen a huge increase in the generation of genomic and proteomic data. This has been due to improvements in current biological methodologies, the development of new experimental techniques and the use of computers as support tools. All these raw data are useless if they cannot be properly analysed, annotated, stored and displayed. Consequently, a vast number of resources have been created to present the data to the wider community. Annotation tools and databases provide the means to disseminate these data and to comprehend their biological importance. This review examines the various aspects of annotation: type, methodology and availability. Moreover, it puts a special interest on novel annotation fields, such as that of phenotypes, and highlights the recent efforts focused on the integrating annotations.

Performance of genomic selection in mice

Selection plans in plant and animal breeding are driven by genetic evaluation. Recent developments suggest using massive genetic marker information, known as "genomic selection." There is little evidence of its performance, though. We empirically compared three strategies for selection: (1) use of pedigree and phenotypic information, (2) use of genomewide markers and phenotypic information, and (3) the combination of both. We analyzed four traits from a heterogeneous mouse population (http://gscan.well.ox.ac.uk/), including 1884 individuals and 10,946 SNP markers. We used linear mixed models, using extensions of association analysis. Cross-validation techniques were used, providing assumption-free estimates of predictive ability. Sampling of validation and training data sets was carried out across and within families, which allows comparing across- and within-family information. Use of genomewide genetic markers increased predictive ability up to 0.22 across families and up to 0.03 within families. The latter is not statistically significant. These values are roughly comparable to increases of up to 0.57 (across family) and 0.14 (within family) in accuracy of prediction of genetic value. In this data set, within-family information was more accurate than across-family information, and populational linkage disequilibrium was not a completely accurate source of information for genetic evaluation. This fact questions some applications of genomic selection.

Genetic variation in a member of the laminin gene family affects variation in body composition in Drosophila and humans

Background

The objective of the present study was to map candidate loci influencing naturally occurring variation in triacylglycerol (TAG) storage using quantitative complementation procedures in Drosophila melanogaster. Based on our results from Drosophila, we performed a human population-based association study to investigate the effect of natural variation in LAMA5 gene on body composition in humans.

Results

We identified four candidate genes that contributed to differences in TAG storage between two strains of D. melanogaster, including Laminin A (LanA), which is a member of the α subfamily of laminin chains. We confirmed the effects of this gene using a viable LanA mutant and showed that female flies homozygous for the mutation had significantly lower TAG storage, body weight, and total protein content than control flies. Drosophila LanA is closely related to human LAMA5 gene, which maps to the well-replicated obesity-linkage region on chromosome 20q13.2-q13.3. We tested for association between three common single nucleotide polymorphisms (SNPs) in the human LAMA5 gene and variation in body composition and lipid profile traits in a cohort of unrelated women of European American (EA) and African American (AA) descent. In both ethnic groups, we found that SNP rs659822 was associated with weight (EA: P = 0.008; AA: P = 0.05) and lean mass (EA: P= 0.003; AA: P = 0.03). We also found this SNP to be associated with height (P = 0.01), total fat mass (P = 0.01), and HDL-cholesterol (P = 0.003) but only in EA women. Finally, significant associations of SNP rs944895 with serum TAG levels (P = 0.02) and HDL-cholesterol (P = 0.03) were observed in AA women.

Conclusion

Our results suggest an evolutionarily conserved role of a member of the laminin gene family in contributing to variation in weight and body composition.

Differential sensitivity of mouse strains to an N-alkylated imino sugar: glycosphingolipid metabolism and acrosome formation

This review deals with the pharmacological properties of an alkylated monosaccharide mimetic, N-butyldeoxynojirimycin (NB-DNJ). This compound is of pharmacogenetic interest because one of its biological effects in mice - impairment of spermatogenesis, leading to male infertility - depends greatly on the genetic background of the animal. In susceptible mice, administration of NB-DNJ perturbs the formation of an organelle, the acrosome, in early post-meiotic male germ cells. In all recipient mice, irrespective of reproductive phenotype, NB-DNJ has a similar biochemical effect: inhibition of the glucosylceramidase beta-glucosidase 2 and subsequent elevation of glucosylceramide, a glycosphingolipid. The questions that we now need to address are: how can glucosylceramide specifically affect early acrosome formation, and why is this contingent on genetic factors? Here we discuss relevant aspects of reproductive biology, the metabolism and cell biology of sphingolipids, and complex trait analysis; we also present a speculative model that takes our observations into account.

Management, presentation and interpretation of genome scans using GSCANDB

MOTIVATION

Advances in high-throughput genotyping have made it possible to carry out genome-wide association studies using very high densities of genetic markers. This has led to the problem of the storage, management, quality control, presentation and interpretation of results. In order to achieve a successful outcome, it may be necessary to analyse the data in different ways and compare the results with genome annotations and other genome scans.

RESULTS

We created GSCANDB, a database for genome scan data, using a MySQL backend and Perl-CGI web interface. It displays genome scans of multiple phenotypes analysed in different ways and projected onto genome annotations derived from EnsMart. The current version is optimized for analysis of mouse data, but is customizable to other species.

AVAILABILITY

Source code and example data are available under the GPL, in versions tailored to either human or mouse association studies, from http://gscan.well.ox.ac.uk/software.

An integrated in silico gene mapping strategy in inbred mice

In recent years in silico analysis of common laboratory mice has been introduced and subsequently applied, in slightly different ways, as a methodology for gene mapping. Previously we have demonstrated some limitation of the methodology due to sporadic genetic correlations across the genome. Here, we revisit the three main aspects that affect in silico analysis. First, we report on the use of marker maps: we compared our existing 20,000 SNP map to the newly released 140,000 SNP map. Second, we investigated the effect of varying strain numbers on power to map QTL. Third, we introduced a novel statistical approach: a cladistic analysis, which is well suited for mouse genetics and has increased flexibility over existing in silico approaches. We have found that in our examples of complex traits, in silico analysis by itself does fail to uniquely identify quantitative trait gene (QTG)-containing regions. However, when combined with additional information, it may significantly help to prioritize candidate genes. We therefore recommend using an integrated work flow that uses other genomic information such as linkage regions, regions of shared ancestry, and gene expression information to obtain a list of candidate genes from the genome.