Volume 16, Issue 4

The systematic study of transcriptional responses to genetic and chemical perturbations in human cells is still in its early stages. The largest available dataset to date is the newly released L1000 compendium. With its 1.3 million gene expression profiles of treated human cells it offers many opportunities for biomedical data mining, but also data normalization challenges of new dimensions. We developed a novel and practical approach to obtain accurate estimates of fold change response profiles from L1000, based on the RUV (Remove Unwanted Variation) statistical framework. Extending RUV to a big data setting, we propose an estimation procedure, in which an underlying RUV model is tuned by feedback through dataset specific statistical measures, reflecting p -value distributions and internal gene knockdown controls. Applying these metrics – termed evaluation endpoints – to disjoint data splits and integrating the results to select an optimal normalization, the procedure reduces bias and noise in the L1000 data, which in turn broadens the potential of this resource for pharmacological and functional genomic analyses. Our pipeline and normalization results are distributed as an R package (nelanderlab.org/FC1000.html).

An important topic in bioinformatics is the protein structure alignment. Some statistical methods have been proposed for this problem, but most of them align two protein structures based on the global geometric information without considering the effect of neighbourhood in the structures. In this paper, we provide a Bayesian model to align protein structures, by considering the effect of both local and global geometric information of protein structures. Local geometric information is incorporated to the model through the partial Procrustes distance of small substructures. These substructures are composed of β -carbon atoms from the side chains. Parameters are estimated using a Markov chain Monte Carlo (MCMC) approach. We evaluate the performance of our model through some simulation studies. Furthermore, we apply our model to a real dataset and assess the accuracy and convergence rate. Results show that our model is much more efficient than previous approaches.

Heritability is the proportion of phenotypic variance in a population that is attributable to individual genotypes. Heritability is considered an important measure in both evolutionary biology and in medicine, and is routinely estimated and reported in genetic epidemiology studies. In population-based genome-wide association studies (GWAS), mixed models are used to estimate variance components, from which a heritability estimate is obtained. The estimated heritability is the proportion of the model’s total variance that is due to the genetic relatedness matrix (kinship measured from genotypes). Current practice is to use bootstrapping, which is slow, or normal asymptotic approximation to estimate the precision of the heritability estimate; however, this approximation fails to hold near the boundaries of the parameter space or when the sample size is small. In this paper we propose to estimate variance components via a Haseman-Elston regression, find the asymptotic distribution of the variance components and proportions of variance, and use them to construct confidence intervals (CIs). Our method is further developed to obtain unbiased variance components estimators and construct CIs by meta-analyzing information from multiple studies. We demonstrate our approach on data from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL).

Genomic imprinting is an epigenetic mechanism that leads to differential contributions of maternal and paternal alleles to offspring gene expression in a parent-of-origin manner. We propose a novel test for detecting the parent-of-origin effects (POEs) in genome wide genotype data from related individuals (twins) when the parental origin cannot be inferred. The proposed method exploits a finite mixture of linear mixed models: the key idea is that in the case of POEs the population can be clustered in two different groups in which the reference allele is inherited by a different parent. A further advantage of this approach is the possibility to obtain an estimation of parental effect when the parental information is missing. We will also show that the approach is flexible enough to be applicable to the general scenario of independent data. The performance of the proposed test is evaluated through a wide simulation study. The method is finally applied to known imprinted genes of the MuTHER twin study data.