February 2014 - statslausanne

(Another) Seminar by Terry Speed on Fri 7 March 2014 (15:00) : "Removing unwanted variation: from principal components to random effects ?"
by Frédéric Schütz 19 Feb '14

19 Feb '14

Following yesterday's email, Terry Speed will give a second seminar on 7 March afternoon, at EPFL. See the full description below. *EPFL Statistics Seminar* Friday, 7 March 2014 - 15h00 (sharp). CE2 (EPFL) *Removing unwanted variation: from principal components to random effects.* Professor Terry Speed UC Berkeley & Walter and Eliza Hall Institute (Hosts: Anthony Davison, Darlene Goldstein, Victor Panaretos) Abstract: *Ordinary least-squares is a venerable tool for the analysis of scientific data originating in the work of A-M. Legendre and C. F. Gauss around 1800. Gauss used the method extensively in astronomy and geodesy. Generalized least squares is more recent, originating with A. C. Aitken in 1934, though weighted least squares was widely used long before that. At around the same time (1933) H. Hotelling introduced principal components analysis to psychology. Its modern form is the singular value decomposition. In 1907, motivated by social science, G. U. Yule presented a new notation and derived some identities for linear regression and correlation. Random effects models date back to astronomical work in the mid-19th century, but it was through the work of C. R. Henderson and others in animal science in the 1950s that their connexion with generalized least squares was firmly made. These are the diverse origins of our story, which concerns the removal of unwanted variation in high-dimensional genomic and other "omic" data using negative controls. We start with a linear model that Gauss would recognize, with ordinary least squares in mind, but we add unobserved terms to deal with unwanted variation. A singular value decomposition, one of Yule's identities, and negative control measurements (here genes) permit the identification of our model. In a surprising twist, our initial solution turns out to be equivalent to a form of generalized least squares. This is the starting point for much of our recent work. In this talk I will try to explain how a rather eclectic mix of familiar statistical ideas can combine with equally familiar notions from biology (negative and positive controls) to give a useful new set of tools for omic data analysis. Other statisticians have come close to the same endpoint from a different perspectives, including Bayesian, sparse linear and random effects models. Joint work with Johann Gagnon-Bartsch and Laurent Jacob.*

1 0

SIB Workshop on "Reproducible Research", April 4 2014
by Frédéric Schütz 19 Feb '14

19 Feb '14

Dear all, the SIB Swiss Institute of Bioinformatics is organizing a new 1-day workshop on "Reproducible Research" which will be held in Lausanne on the *4 April 2014*. It is likely to be of interest to several readers of this mailing-list -- "reproducible research" was the topic of our first meeting two years ago, and at the time, several participants indicated their interest in the organization of such a workshop. Full description: This course will focus on the concept and techniques that allow statisticians/data analysts to ensure as much as possible that their analysis performed using R is reproducible. Reproducible research in statistical analysis is the idea that it should be possible for other people to reproduce a published data analysis and obtain the same results ("other people" also include the person who did the initial data analysis, 6 months later). While all data analysts agree that the concept of reproducible research is important, very few of them actually take active steps to ensure that their work is reproducible. Participants should have experience in using R for analyzing data. Basic knowledge of LaTeX or LyX is an advantage. Participants should bring a laptop and be able to install custom applications. A list of required applications including R, LaTeX and GIT will be provided before the course. The course is organized by the Bioinformatics Core Facility (BCF) of the SIB Swiss Institute of Bioinformatics. Further information and registration are available from http://edu.isb-sib.ch/course/view.php?id=155 Please note that the number of seats is limited. Hope to see you there! Frédéric -- Frédéric Schütz, PhD Frederic.Schutz(a)isb-sib.ch Maître d'Enseignement et de Recherche, Center for Integrative Genomics Senior Statistician, Bioinformatics Core Facility (Delorenzi group) Swiss Institute of Bioinformatics http://www.isb-sib.ch/ University of Lausanne http://www.unil.ch/ Batiment Genopode Quartier Sorge Phone: +41 21 692 40 94 CH-1015 Lausanne Fax: +41 21 692 40 65

1 0

Seminar by Terry Speed on Fri 7 March 2014: "Normalization of RNA-Seq Data: Are the ERCC Spike-In Controls Reliable ?"
by Frédéric Schütz 18 Feb '14

18 Feb '14

Terry Speed from the Walter and Eliza Hall Institute, Melbourne, AUS and the University of California, Berkeley, USA, will talk about: "Normalization of RNA-Seq Data: Are the ERCC Spike-In Controls Reliable?" Friday, March 7th, 2014 at 11:00 Génopode Auditorium B (UNIL, Dorigny campus) Abstract: Joint work with Sandrine Dudoit, Davide Risso and John Ngai, all from UC Berkeley. The External RNA Control Consortium (ERCC) developed a set of 92 synthetic polyadenylated RNA standards that mimic natural eukaryotic mRNA (Jiang et al., 2011). The standards are designed to have a wide range of lengths (250-2,000 nucleotides) and GC-contents (5-51%). The ERCC standards can be spiked into RNA at various concentrations prior to the library preparation step and serve as negative and positive controls in RNA-Seq. Ambion commercializes spike-in control mixes, ERCC ExFold RNA Spike-in Control Mix 1 and 2, each containing the same set of 92 standards, but at different concentrations. We investigate the use of the ERCC spike-in controls for two main purposes: (a) Quality assessment/quality control (QA/QC) of RNA-Seq data and benchmarking of normalization and differential expression (DE) methods, and (b) Direct inclusion in between-sample normalization procedures. We have two RNA-seq data sets which make use of the ERCC controls: a local one concerning treated and untreated zebrafish tissue, and some of the SEQC samples. A variety of normalization methods will be compared, both using and not using the ERCC controls.

1 0