Poster M07 |
Automatic classification of P-type ATPases using Structured Logistic Regression |
Poul Liboriussen- Aarhus University |
Bjørn Panyella Pedersen (Aarhus University, Centre for Membrane Pumps in Cells and Disease (PUMPKIN)); Poul Nissen (Aarhus University, Centre for Membrane Pumps in Cells and Disease (PUMPKIN)); Christian Nørgaard Storm Pedersen (Aarhus University, Bioinformatics Research Center (BiRC)); |
Short Abstract: P-type ATPases are a very large family of ATP-driven membrane pumps involved in transmembrane transport of charged substrates. We have constructed a classifier that can distinguish between the 11 subfamilies with high accuracy. The classified it applied to Swiss-Prot/TrEMBL, and finds 6.624 P-Type ATPases. |
Long Abstract: Click Here |
|
Poster M09 |
Combining evidence from ranked gene lists |
Raivo Kolde- University of Tartu |
Sven Laur (University of Tartu, Institute of Computer Science); Priit Adler (University of Tartu, Institute of Molecular and Cell Biology); Jaak Vilo (University of Tartu, Institute of Computer Science); |
Short Abstract: We propose a strategy for combining evidence from ranked lists of genes. In addition to the ranking of genes, the algorithm assigns significance probability for each gene. The method can be applied in network reconstruction, meta-analysis of microarray studies, etc. |
Long Abstract: Click Here |
|
Poster M10 |
Missing Value Imputation for Epistasis Maps |
Colm Ryan- University College Dublin |
Derek Greene (University College Dublin, School of Computer Science and Informatics); Nevan Krogan (University of California, San Francisco, Quantitiative Biology Institute); Gerard Cagney (University College Dublin, Conway Institute of Biomolecular and Biomedical Research); Pádraig Cunningham (University College Dublin, School of Computer Science and Informatics); |
Short Abstract: We introduce the problem of missing value imputation for Epistasis miniarray profiles(E-MAPS) and show the results of adapting two existing techniques to address the problem. In doing so we highlight some unique aspects of the problem – the pairwise nature of the data and the high percentage of missing values. |
Long Abstract: Click Here |
|
Poster M11 |
KIRMES: Kernel-based Identification of Regulatory Modules in Euchromatic Sequences |
Sebastian Schultheiss- Friedrich Miescher Laboratory of the Max Planck Society |
Wolfgang Busch (Duke University, Biology Department); Jan Lohmann (University of Heidelberg, Center for Organismal Studies); Oliver Kohlbacher (University of Tuebingen, Wilhelm Schickard Institute for Computer Science); Gunnar Raetsch (Friedrich Miescher Laboratory of the Max Planck Society, Machine Learning in Biology); |
Short Abstract: We identify genes regulated by the same transcription factor by analyzing sets of co-expressed genes from microarrays. KIRMES infers all genes regulated by the same mechanism as the ones in the input set. KIRMES makes use of motif sampling and newly developed kernel methods for this task. |
Long Abstract: Click Here |
|
Poster M14 |
Using support vector machines for the evaluation of computationally developed lipoxygenase structures |
Aditya Jitta- University of Hyderabad |
Aparoy P (University of Hyderabad, School of Life Sciences); Reddanna P (University of Hyderabad, School of Life Sciences); |
Short Abstract: Lipoxygenases are a group of structurally related family of non-heme, iron-containing dioxygenases,the geometry and composition at metal binding site in 3D models of lipoxygenases is very important.Based on these features,a tool was developed using support vector machines to evaluatecomputationally developed lipoxygenase structures. |
Long Abstract: Click Here |
|
Poster M15 |
I/NI-calls: a novel unsupervised feature selection criterion |
Sepp Hochreiter - Johannes Kepler University Linz |
Djork-Arné Clevert (Johannes Kepler University Linz, Institute of Bioinformatics); Willem Talloen (Johnson & Johnson Pharmaceutical Research & Development, Pharmaceutical Research & Development); Hinrich Göhlmann (Johnson & Johnson Pharmaceutical Research & Development, Pharmaceutical Research & Development); Sepp Hochreiter (Johannes Kepler University Linz, Institute of Bioinformatics); |
Short Abstract: We propose a novel unsupervised gene selection criterion that is based on a probabilistic latent variable model that takes probe level information -- probe correlations that cannot be explained by noise -- into account to filter out inconsistent probe sets. |
Long Abstract: Click Here |
|
Poster M16 |
An integrative pipeline for automated data analysis and gene function annotation for genome wide high content RNAi screening |
Stephen Wong- Center for Biotechnology and Informatics, The Methodist Hospital |
Xiaobo Zhou (Center for Biotechnology and Informatics, The Methodist Hospital, The Methodist Hospital Research Institute and Department of Radiology); |
Short Abstract: We propose an integrated pipeline of automated data analysis for high-content screening of genome-wide RNA interference on Drosophila cell assays. Millions of cells are efficiently segmented, and previously un-scored phenotypes are identified. This image bioinformatics pipeline is especially helpful in predicting the roles of genes in complex biological processes. |
Long Abstract: Click Here |
|
Poster M17 |
Computational Linguistic Analyses of Unknown Metagenome Sequences |
Victor Seguritan- San Diego State University |
Anca Segall (San Diego State University, Biology); Rob Edwards (San Diego State University, Computer Science); Forest Rohwer (San Diego State University, Biology); |
Short Abstract: A method is needed to assign functions to unknown sequences which does not rely on sequence homology alone. The linguistic elements, syntax and semantics, of several model proteins will be used to assign functions to unknown metagenomes in a manner similar to the concept of understanding human language. |
Long Abstract: Click Here |
|
Poster M18 |
Neural Network Pairwise Interaction Fields for protein model quality assessment |
Alberto Jesus Martin- University College |
Gianluca Pollastri (Complex and Adaptive Systems Laboratory, University College Dublin, School of Computer Science and Informatics); Alessandro Vullo (Complex and Adaptive Systems Laboratory, University College Dublin, School of Computer Science and Informatics); |
Short Abstract: We present a new knowledge-based Model Quality Assessment Program (MQAP) at the residue level which evaluates single protein structure models. We use a tree representation of the C-alpha trace to train a novel Neural Network Pairwise Interaction Field (NN-PIF) to predict the global quality of a model. |
Long Abstract: Click Here |
|
Poster M21 |
A method for analyzing gene expression profiles based on the underlying structures |
Shigeto Seno- Dept. Bioinfo. Eng., Grad. Sch. Info. Sci. Tech., Osaka Univ. |
Yoichi Takenaka (Graduate School of Information Science and Technology, Osaka University, Bioinfomatic Engineering); Hideo Matsuda (Graduate School of Information Science and Technology, Osaka University, Bioinfomatic Engineering); |
Short Abstract: Clustering is a powerful tool for elucidating relationships among genes, and one of the first steps in analysis. Meanwhile choice of suitable method for a given dataset is still difficult. Our approach discovers the underlying structure of a gene expression profile and provides a more intuitive understanding. |
Long Abstract: Click Here |
|
Poster M22 |
Monte Carlo-Based Bayesian Prediction of Gene Regulatory Networks with Zipf Distribution: Mouse Nuclear Receptor Superfamily |
Haruka Miyachika- Waseda University |
Yusuke Kitamura (Waseda University, Electrical Engineering and Bioscience); Tomomi Kimiwada (National Center of Neurology and Psychiatry, Neurosurgery); Jun Maruyama (Waseda University, Electrical Engineering and Bioscience); Takashi Kaburagi (Waseda University, Electrical Engineering and Bioscience); Takashi Matsumoto (Waseda University, Electrical Engineering and Bioscience); Keiji Wada (National Center of Neurology and Psychiatry, Neurosurgery); |
Short Abstract: We present a Monte Carlo-based algorithm to predict gene regulatory network structure within a Bayesian framework. The algorithm assumes that prior distribution follows the Zipf law, and is implemented using the Exchange Monte Carlo method. We applied the algorithm to a mouse nuclear receptor superfamily. |
Long Abstract: Click Here |
|
Poster M24 |
Two-way Analysis of High-Dimensional Metabolomic Datasets |
Ilkka Huopaniemi- Helsinki University of Technology |
Tommi Suvitaival (Helsinki University of Technology, Department of Information and Computer Science); Janne Nikkilä (Helsinki University of Technology, Department of Information and Computer Science); Matej Oresic (VTT Technical Research Centre of Finland, Quantitative Biology and Bioinformatics); Samuel Kaski (Helsinki University of Technology, Department of Information and Computer Science); |
Short Abstract: We present a Bayesian machine learning method for multivariate two-way ANOVA-type analysis ofhigh-dimensional, small sample-size metabolomic datasets. The method assumes clustered metabolites and presents confidence intervals of main and interaction up/down-regulation effects of the clusters. |
Long Abstract: Click Here |
|
Poster M26 |
Bioinformatic analyses of mammalian 5'-UTR sequence properties of mRNAs predicts alternative translation initiation sites |
Jill Wegrzyn- University of California at San Diego |
Thomas Drudge (University of California at San Diego, Skaggs School of Pharmacy and Pharmaceutical Sciences); Farmarz Valafar (San Diego State University, Bioinformatics and Medical Informatics Research Center (BMIRC)); Vivian Hook (University of California at San Diego, Skaggs School of Pharmacy and Pharmaceutical Sciences); |
Short Abstract: This study conducted a bioinformatic evaluation of the 5'-UTR of mammalian mRNA sequences. Machine learning techniques were applied for the classification and identification of non-AUG initiation sites in a group of mRNAs that have been experimentally demonstrated to utilize alternative sites for protein translation. |
Long Abstract: Click Here |
|
Poster M28 |
The evaluation of common 1H-NMR metabolomics data preprocessing procedures reveals unanticipated side-effects |
Tim De Meyer- Ghent University |
Bjorn Van Gasse (Ghent University, Dept. Organic Chemistry); Davy Sinnaeve (Ghent University, Dept. Organic Chemistry); Sofie Bekaert (Ghent University, Dept. Molecular Biotechnology); José Martins (Ghent University, Dept. Organic Chemistry); Wim Van Criekinge (Ghent University, Dept. Molecular Biotechnology); |
Short Abstract: 1H-NMR metabolomics provides a high-throughput methodology capable of acquiring high-resolution profiles of low-molecular weight metabolites. However, the complicated data-analysis forms a major drawback, requiring numerous data preprocessing procedures (particularly normalization, reduction and scaling steps). Here, we evaluate the most common procedures and demonstrate several unanticipated side-effects. |
Long Abstract: Click Here |
|