November 2013 - statslausanne

Statistical Forum November 28 2013, 11:00
by Rousson Valentin 19 Nov '13

19 Nov '13

Dear all, The next seminar in our series of statistics talks for people interested in quantitative methods in biology and medicine will take place on *Thursday 28 November 2013 at 11h00* in the Salle Delachaux (Biopôle 2, premier étage, Route de la Corniche 10, Lausanne, M2: Vennes). The speaker will be *Alina Matei* (University of Neuchâtel) who will speak about "Unequal probability sampling designs, entropy and sample coordination in finite populations" The slides will be in English and the talk in French. Abstract: After a short introduction to unequal probability sampling designs, we show how the sample randomness is related to the Shannon entropy. Higher entropy of a sampling design implies more randomness in sample selection. We present afterwards the unequal probability sampling design, which maximizes the entropy criterion in the class of all possible sampling designs having the same unit inclusion probabilities and a fixed sample size (called maximum fixed-size entropy sampling or Conditional Poisson sampling). Next, we focus on sample coordination, which seeks to create a probabilistic dependence between two or more samples drawn successively or simultaneously from overlapping finite populations. This dependence leads to a maximization or a minimization of the sample overlap. The former is known as positive coordination, and the latter as negative coordination. Positive coordination is mainly employed for estimation purposes and for reasons related to data collection costs. Negative coordination is mainly used to diminish the response burden of the sampled units selected in different samples. We show two schemes to coordinate Conditional Poisson samples. The proposed methods are evaluated using the size of the expected sample overlap, and compared with their competitors, by means of the Monte Carlo simulation. Hope to see you there! Valentin (for the organizers: Jérôme Goudet, Valentin Rousson, Frédéric Schütz) ---------------------------------------------------- Valentin Rousson Unité de Statistique Institut Universitaire de Médecine Sociale et Préventive Centre Hospitalier Universitaire et Université de Lausanne Biopôle 2 Route de la Corniche 10 CH-1010 Lausanne Tél : ++41 (021) 314 73 28 Fax : ++41 (021) 314 73 73 Email: Valentin.Rousson(a)chuv.ch<mailto:Valentin.Rousson@chuv.ch>

1 0

Next statistics forum: 14 November, "Hunting for rare genetic determinants of psychiatric traits", Sina Rüeger
by Frédéric Schütz 12 Nov '13

12 Nov '13

Dear all, The next seminar in our series of statistics talks for people interested in quantitative methods in biology and medicine will take place on *Thursday 14 November 2013 at 14h00* (not 11:00!) in the Salle Delachaux (Biopôle 2, premier étage, Route de la Corniche 10, Lausanne, M2: Vennes). The speaker will be *Sina Rüeger* (UNIL) who will speak about "Hunting for rare genetic determinants of psychiatric traits" Abstract: Only little is known about the genetic architecture of psychological disorders (e.g. bipolar disorder, major depression disorder, or cocaine dependence); they are assumed to be highly complex and heterogenous traits. Very few common variants have been identified, thus it is also assumed that rare variants may play an important role. To understand the role of rare variants in psychological disorders and the relation between phenotypes we analysed the sequencing data (including rare variants) using the psychiatric arm of the Colaus study cohort. The talk will focus on difficulties arising from rare variant associations, i.e. the standard methods are not necessarily adequate in this setting. Different “variant burden tests”, designed to analyse rare variants, will be presented; regression based collapsing methods and kernel association tests (SKAT). If time permits, I will also present a strategy of how to combine multiple phenotypes to improve statistical power (some psychiatric traits might be highly correlated, and thus are assumed to have similar genetic risk factors). Hope to see you there! Frédéric (for the organizers: Jérôme Goudet, Valentin Rousson, Frédéric Schütz)

1 0

"On the consequences of dichotomization on correlations and odds-ratios", Keith Baggerly on 12 November 2013
by Frédéric Schütz 04 Nov '13

04 Nov '13

Dear all, next week, the SIB will host Keith Baggerly from the MD Anderson Cancer Center (USA), who will speak about "When is Reproducibility an Ethical Issue? Genomics, Personalized Medicine, and Human Error" The presentation will take place on *Tuesday 12 November October 2013 at 14h00* in the Auditoire B of the Génopode building (Dorigny campus of the University of Lausanne). While this presentation does not happen at the usual time for our "statistics forum", it is likely to be of interest to most statisticians around. Keith's name is associated with the terms "forensic bioinformatics" and "reproducible research", and he was one of the main actors in the "great Duke cancer disaster" saga (see http://www.nytimes.com/2011/07/08/health/research/08genes.html?_r=1& for more information). If you are looking for a more scientific reading, his paper published in the "Annals of Applied Statistics" (http://projecteuclid.org/DPubS?service=UI&version=1.0&verb=Display&handle=e…) is particularly interesting. The abstract follows: Modern high-throughput biological assays let us ask detailed questions about how diseases operate, and promise to let us personalize therapy. Careful data processing is essential, because our intuition about what the answers “should” look like is very poor when we have to juggle thousands of things at once. When documentation of such processing is absent, we must apply “forensic bioinformatics” to work from the raw data and reported results to infer what the methods must have been. We will present several case studies where simple errors may have put patients at risk. This work has been covered both in the scientific press and on the front page of the New York Times, and has prompted several journals to revisit the types of information that must accompany publications. We discuss steps we take to avoid such errors, and lessons that can be applied to large data sets more broadly. Hope to see you there! Frédéric -- Frédéric Schütz, PhD Frederic.Schutz(a)isb-sib.ch 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 1