Please activate JavaScript!
Please install Adobe Flash Player, click here for download

ISCB2014_abstract_book

ISCB 2014 Vienna, Austria • Abstracts - Poster Presentations 113Tuesday, 26th August 2014 • 10:30-11:00 Monday25thAugustTuesday26thAugustThursday28thAugustAuthorIndexPostersWednesday27thAugustSunday24thAugust P2.5.194 A non-homogeneous hidden Markov-model for gene mapping based on whole‑genome sequencing data F Zamanzad Ghavidel1 , J Claesen1 , T Burzykowski1 1 I-Biostat, Hasselt University, Diepenbeek, Belgium   The analysis of polygenetic characteristics for mapping quantitative trait loci (QTL) remains an important challenge. QTL analysis requires two or more strains of organisms that differ substantially in the (poly-)genetic trait of interest, resulting in a heterozygous offspring. The offspring with the trait of interest is selected and subsequently screened for genetic markers such as single nucleotide polymorphisms (SNPs) with next gen- eration sequencing (NGS). Gene mapping relies on the principle of co- segregation, the tendency for closely linked genes and genetic markers to be inherited together. For each marker, observed mismatch frequencies between the reads of the offspring and the parental reference strains can be modeled by a mul- tinomial distribution with the probabilities depending on the state of an unobserved (hidden) Markov process (Claesen and Burzykowski, 2014). After fitting the model to data, the Viterbi algorithm can be used to pre- dict the most probable state for each of the SNPs. The predicted states can be used to infer whether a SNP is located in a (vicinity of a) QTL or not. Consequently, genomic loci associated with the QTL can be discovered by analyzing hidden states along the genome. The aforementioned hidden Markov‑model does not take into account the variation in the location of SNPs across the genome. To address this issue, we develop a non-homogenous hidden Markov‑model with a tran- sition matrix that depends on a set of distance-varying observed covari- ates. The application of the model is illustrated on the data from a study of ethanol tolerance in yeast.   P2.6 Software aspects of efficient statistical analyses P2.6.26 Easy-to-use R-application to evaluate bioequivalence studies M Borsos1 , M Müller1 1 AdWare Research Ltd, Balatonfüred, Hungary   Aim: The aim of our work was to develop an easy-to-use application for researchers to be able to evaluate their bioequivalency study data from the calculation of pharmacokinetic parameters to obtain a raw statistical report with easily understandable tables and graphs. Methods: Our goal was to apply only free softwares and still provide a user-friendly solution that doesn’t require high programming or statisti- cal knowledge from the researcher. We developed two R macros: one for calculating the pharmacokinetic parameters using non-compartmental methods, and one for performing the standard statistical analysis required by the FDA for 2x2 crossover bioequivalence studies. Then we validated our programs with the help of previously programmed and validated SAS codes by comparing the results on numerous simulated databases. Finally, we integrated our R programs with Sweave to provide the required tables and graphs automatically in a pdf report. Results: We obtained two LaTeX codes with R programs that only needs the study data given in easy pre-defined formats and after running the ap- plications, it results in dynamically changing pdf reports including tables, graphs and standard texts fitted to the actual results.   P2.6.30 Boosting diagnosis performance of biomarkers with nonparametric logistic type classification functions Y-CI Chang1 1 Institute of Statistical Science, Academia Sinica, Taipei, Republic of China (Taiwan)   The binary classification task is very common in medical diagnosis, where subjects are classified into one of two groups based on observed values of variables. As there are usually many variables of interest in a study, optimiz- ing the combination of these biomarkers is an important problem. A linear combination is usually preferred because of its ease of interpretation. In the literature, there are already numerous published reports on achieving the best linear combination of biomarkers that maximizes the area under a receiver operating characteristic (ROC) curve, which is a popular tool for measuring classification performance of a classifier. However, there is of- ten a lack of information about the relationship between disease status and the value of each biomarker. Hence, an improved method is always in demand. Here we propose a nonparametric classification function based on a general additive logistic model. It is proved that the proposed meth- od gives a greater area under ROC curve than that of a linear combination. Moreover, because of the property of the method of additive functions, the proposed method retains the information of the relationship between biomarker and the disease status in the sense of a general additive model. Numerical results based on both synthesized and real data are reported.   P2.6.73 Repeated observations design analysed with ANOVA tools in MS-Excel K Hrach1 1 J.E.Purkyne University, Faculty of Health Studies, Usti nad Labem, Czech Republic   Repeated measures design (i.e. the situation, when each subject in the study is exposed to each level of the factor or factors) was discussed in a similar contribution last year in Munchen. Now, repeated observations de- sign (i.e. the situation, when each subject in the study is measured several times under the same conditions) is in focus. Such designs are often used in practical trials. The aim of this contribution is to show, how to perform the analysis of such a data, with the help of the common Excel analytical tools (ANOVA 1-factor and ANOVA 2-factors with replications, namely). This contribution proves the validity of the approach and performs a practical manual, how to do it. The situation concerning balanced single-factor repeated obser- vations data, is discussed. Concrete example is solved as an illustration. These findings were formed as a by-product of the grant project n. NT 14448-3/2013 of the Czech Ministry of Health.

Pages Overview