Modelling Count Data in the Era of Next Generation Sequencing Data
Presenter: Hongzhe Li
When: Monday, July 11, 2016 Time: 11:00 AM - 12:30 PM
Room: Salon A Carson Hall (Level 2)
Session Synopsis:
Covariance Estimation for Compositional Data via Composition-Adjusted Thresholding
High-dimensional compositional data arise naturally in many applications such as metagenomic data analysis. The observed data lie in a high-dimensional simplex, and conventional statistical methods often fail to produce sensible results due to the unit-sum constraint. In this article, we introduce a composition-adjusted thresholding (COAT) method for estimating the covariance structure of high-dimensional compositional data under the assumption that the basis covariance matrix is sparse. Our method is based on a decomposition relating the compositional covariance to the basis covariance, which is approximately identifiable as the dimensionality tends to infinity. The resulting procedure can be viewed as thresholding the sample centered log-ratio covariance matrix and hence is scalable for large matrices. We rigorously characterize the identifiability of the covariance parameters, derive rates of convergence under the spectral norm, and provide theoretical guarantees on support recovery. Simulation studies demonstrate that the COAT estimator outperforms some naive thresholding estimators that ignore the unique features of compositional data. We apply the proposed method to the analysis of a microbiome dataset in order to understand the complex dependence structure among bacterial taxa in the human gut.
Modelling Count Data in the Era of Next Generation Sequencing Data
Presenter: Stephane Robin
When: Monday, July 11, 2016 Time: 11:00 AM - 12:30 PM
Room: Salon A Carson Hall (Level 2)
Session Synopsis:
Modeling over-dispersion heterogeneity in differential expression analysis using mixtures
Next-generation sequencing technologies now constitute a method of choice to measure gene expression. Data to analyze are read counts, commonly modeled using Negative Binomial distributions. A relevant issue associated with this probabilistic framework is the reliable estimation of the over-dispersion parameter, reinforced by the limited number of replicates generally observable for each gene. Many strategies have been proposed to estimate this parameter, but when differential analysis is the purpose, they often result in procedures based on plug-in estimates, and we show here that this discrepancy between the estimation framework and the testing framework can lead to uncontrolled type-I errors. Instead we propose a mixture model that allows each gene to share information with other genes that exhibit similar variability. Three consistent statistical tests are developed for differential expression analysis. We show through a wide simulation study that the proposed method improves the sensitivity of detecting differentially expressed genes with respect to the common procedures, since it reaches the nominal value for the type-I error, while keeping elevate discriminative power between differentially and not differentially expressed genes. The method is finally illustrated on prostate cancer RNA-Seq data.
Modelling Count Data in the Era of Next Generation Sequencing Data
Presenter: Mark Robinson
When: Monday, July 11, 2016 Time: 11:00 AM - 12:30 PM
Room: Salon A Carson Hall (Level 2)
Session Synopsis:
Highly optimized statistical modeling of genome-scale count data
With the advent of large scale "next generation" sequencing, there is now a large amount of quantitative count data regarding molecular events (e.g., gene expression, splicing, protein-DNA interactions) being collected. Experimental units are still limiting, so there is high demand for statistical methods that make parallel inferences in small samples. In the last few years, several optimizations of negative binomial modeling have afforded flexible, powerful and robust methods. I will give an overview of the method developments for differential gene expression analysis, as well as pointers to various new features and additional extensions of the methods to other related domains.
Modelling Count Data in the Era of Next Generation Sequencing Data
Presenter: Roula Tsonaka
When: Monday, July 11, 2016 Time: 11:00 AM - 12:30 PM
Room: Salon A Carson Hall (Level 2)
Session Synopsis:
METHODOLOGICAL CHALLENGES IN THE ANALYSIS OF LONGITUDINAL RNASEQ DATA
Identification of genes with differentially expressed profiles in follow-up RNAseq experiments is crucial for understanding the transcriptional regulatory network. Experiments may involve samples repeatedly sequenced at a couple or even more occasions, and the number of samples can vary from a handful of patients assigned to two or more experimental conditions to hundreds of patients. Depending on the experimental design at hand, several complications may arise in the statistical analysis. Proper normalization, careful statistical modelling which addresses the research questions of interest and captures key features of longitudinal RNAseq data is crucial. Currently available statistical software for RNAseq experiments cannot be successfully used for the differential gene expression analysis in all cases. They may be limited to the analysis of single or at most paired measurements and testing can preserve good statistical properties only in small sample designs. For longer follow-up designs, time-dependent over-dispersion and within samples serial correlation may complicate the statistical analysis. Common mixed-effects models can be computationally intensive and fail to converge. In this talk we will discuss statistical challenges in studying the progression of RNAseq data from normalization to differential gene expression, provide an overview of state-of-the-art methods and present a recently developed approach which pairs methods for mixed-effects models with empirical Bayes methodology to stabilize estimation of differential gene expression over time.
