Band 16, Heft 5-6

In many population genetic problems, parameter estimation is obstructed by an intractable likelihood function. Therefore, approximate estimation methods have been developed, and with growing computational power, sampling-based methods became popular. However, these methods such as Approximate Bayesian Computation (ABC) can be inefficient in high-dimensional problems. This led to the development of more sophisticated iterative estimation methods like particle filters. Here, we propose an alternative approach that is based on stochastic approximation. By moving along a simulated gradient or ascent direction, the algorithm produces a sequence of estimates that eventually converges to the maximum likelihood estimate, given a set of observed summary statistics. This strategy does not sample much from low-likelihood regions of the parameter space, and is fast, even when many summary statistics are involved. We put considerable efforts into providing tuning guidelines that improve the robustness and lead to good performance on problems with high-dimensional summary statistics and a low signal-to-noise ratio. We then investigate the performance of our resulting approach and study its properties in simulations. Finally, we re-estimate parameters describing the demographic history of Bornean and Sumatran orang-utans.

We consider the assessment of DNA methylation profiles for sequencing-derived data from a single cell type or from cell lines. We derive a kernel smoothed EM-algorithm, capable of analyzing an entire chromosome at once, and to simultaneously correct for experimental errors arising from either the pre-treatment steps or from the sequencing stage and to take into account spatial correlations between DNA methylation profiles at neighbouring CpG sites. The outcomes of our algorithm are then used to (i) call the true methylation status at each CpG site, (ii) provide accurate smoothed estimates of DNA methylation levels, and (iii) detect differentially methylated regions. Simulations show that the proposed methodology outperforms existing analysis methods that either ignore the correlation between DNA methylation profiles at neighbouring CpG sites or do not correct for errors. The use of the proposed inference procedure is illustrated through the analysis of a publicly available data set from a cell line of induced pluripotent H9 human embryonic stem cells and also a data set where methylation measures were obtained for a small genomic region in three different immune cell types separated from whole blood.

RNA-Seq is a developing technology for generating gene expression data by directly sequencing mRNA molecules in a sample. RNA-Seq data consist of counts of reads recorded to a particular gene that are often used to identify differentially expressed (DE) genes. A common statistical method used to analyze RNA-Seq data is Significance Analysis of Microarray with emphasis on RNA-Seq data (SAMseq). SAMseq is a nonparametric method that uses a resampling technique to account for differences in sequencing depths when identifying DE genes. We propose a modification of this method that takes into account asymmetry in the distribution of the effect sizes by taking into account the sign of the test statistics. Through simulation studies, we showthat the proposed method, comparedwith the traditional SAMseqmethod and other existing methods provides better power for identifying truly DE genes or more sufficiently controls FDR in most settings where asymmetry is present. We illustrate the use of the proposed method by analyzing an RNA-Seq data set containing C57BL/6J (B6) and DBA/2J (D2) mouse strains samples.

We introduce a new method to test efficiently for cospeciation in tritrophic systems. Our method utilises an analogy with electrical circuit theory to reduce higher order systems into bitrophic data sets that retain the information of the original system. We use a sophisticated permutation scheme that weights interactions between two trophic layers based on their connection to the third layer in the system. Our method has several advantages compared to the method of Mramba et al. [Mramba, L. K., S. Barber, K. Hommola, L. A. Dyer, J. S. Wilson, M. L. Forister and W. R. Gilks (2013): “Permutation tests for analyzing cospeciation in multiple phylogenies: applications in tri-trophic ecology,” Stat. Appl. Genet. Mol. Biol., 12, 679–701.]. We do not require triangular interactions to connect the three phylogenetic trees and an easily interpreted p -value is obtained in one step. Another advantage of our method is the scope for generalisation to higher order systems and phylogenetic networks. The performance of our method is compared to the methods of Hommola et al. [Hommola, K., J. E. Smith, Y. Qiu and W. R. Gilks (2009): “A permutation test of host–parasite cospeciation,” Mol. Biol. Evol., 26, 1457–1468.] and Mramba et al. [Mramba, L. K., S. Barber, K. Hommola, L. A. Dyer, J. S. Wilson, M. L. Forister and W. R. Gilks (2013): “Permutation tests for analyzing cospeciation in multiple phylogenies: applications in tri-trophic ecology,” Stat. Appl. Genet. Mol. Biol., 12, 679–701.] at the bitrophic and tritrophic level, respectively. This was achieved by evaluating type I error and statistical power. The results show that our method produces unbiased p -values and has comparable power overall at both trophic levels. Our method was successfully applied to a dataset of leaf-mining moths, parasitoid wasps and host plants [Lopez-Vaamonde, C., H. Godfray, S. West, C. Hansson and J. Cook (2005): “The evolution of host use and unusual reproductive strategies in achrysocharoides parasitoid wasps,” J. Evol. Biol., 18, 1029–1041.], at both the bitrophic and tritrophic levels.

Using publicly-available data from the Alzheimer’s Disease Neuroimaging Initiative, we investigate the joint association between single-nucleotide polymorphisms (SNPs) in previously established linkage regions for Alzheimer’s disease (AD) and rates of decline in brain structure. In an initial, discovery stage of analysis, we applied a weighted RV test to assess the association between 75,845 SNPs in the Alzgene linkage regions and rates of change in structural MRI measurements for 56 brain regions affected by AD, in 632 subjects. After confirming association, we selected refined lists of 1694 and 22 SNPs via a bootstrap-enhanced sparse canonical correlation analysis. In a final, validation stage, we confirmed association between the refined list of 1694 SNPs and the imaging phenotypes in an independent data set. Genes corresponding to priority SNPs having the highest contribution in the validation data have previously been implicated or hypothesized to be implicated in AD, including GCLC , IDE , and STAMBP1andFAS . Though the effect sizes of the 1694 SNPs in the priority set are likely small, further investigation within this set may advance understanding of the missing heritability in AD. Our analysis addresses challenges in current imaging-genetics studies such as biased sampling designs and high-dimensional data with low association signal.

Next generation sequencing allows the identification of genes consisting of differentially expressed transcripts, a term which usually refers to changes in the overall expression level. A specific type of differential expression is differential transcript usage (DTU) and targets changes in the relative within gene expression of a transcript. The contribution of this paper is to: (a) extend the use of cjBitSeq to the DTU context, a previously introduced Bayesian model which is originally designed for identifying changes in overall expression levels and (b) propose a Bayesian version of DRIMSeq, a frequentist model for inferring DTU. cjBitSeq is a read based model and performs fully Bayesian inference by MCMC sampling on the space of latent state of each transcript per gene. BayesDRIMSeq is a count based model and estimates the Bayes Factor of a DTU model against a null model using Laplace’s approximation. The proposed models are benchmarked against the existing ones using a recent independent simulation study as well as a real RNA-seq dataset. Our results suggest that the Bayesian methods exhibit similar performance with DRIMSeq in terms of precision/recall but offer better calibration of False Discovery Rate.

Genomic studies of plants often seek to identify genetic factors associated with desirable traits. The process of evaluating genetic markers one by one (i.e. a marginal analysis) may not identify important polygenic and environmental effects. Further, confounding due to growing conditions/factors and genetic similarities among plant varieties may influence conclusions. When developing new plant varieties to optimize yield or thrive in future adverse conditions (e.g. flood, drought), scientists seek a complete understanding of how the factors influence desirable traits. Motivated by a study design that measures rice yield across different seasons, fields, and plant varieties in Indonesia, we develop a regression method that identifies significant genomic factors, while simultaneously controlling for field factors and genetic similarities in the plant varieties. Our approach develops a Bayesian maximum a posteriori probability (MAP) estimator under a generalized double Pareto shrinkage prior. Through a hierarchical representation of the proposed model, a novel and computationally efficient expectation-maximization (EM) algorithm is developed for variable selection and estimation. The performance of the proposed approach is demonstrated through simulation and is used to analyze rice yields from a pilot study conducted by the Indonesian Center for Rice Research.