| Accepted Posters |
| Category 'L'- Machine Learning' |
| Poster L01 |
| Nucleosome positioning signals in yeast and human genomic DNA |
| William Noble- University of Washington |
| Heather Peckham (Boston University, Bioinformatics Program); Zhiping Weng (Boston University, Bioinformatics Program); Robert Thurman (University of Washington, Medical Genetics); John Stamatoyannopoulos (University of Washington, Genome Sciences); Jonathan Dennis (Harvard University, Molecular Biology); Robert Kingston (Harvard University, Molecular Biology); Shobhit Gupta (University of Washington, Genome Sciences); Kevin Struhl (Harvard University, Biological Chemistry and Molecular Pharmacology ); |
| Short Abstract: None On File |
| Long Abstract: Click Here |
| Poster L02 |
| Structure Learning in Nested Effects Models |
| Florian Markowetz- Princeton University |
| Achim Tresch (LMU Munich, Gene Center); Olga G Troyanskaya (Princeton University, Lewis-Sigler Institute); |
| Short Abstract: None On File |
| Long Abstract: Click Here |
| Poster L03 |
| GeneX-2DrugPheno: Method to identify markers for gemcitabine cytotoxicity |
| Krishna Kalari- Mayo Clinic, Rochester |
| Liang Li (Mayo Clinic, Rochester, Molecular Pharmacology and Experimental Therapeutics); Jean-Pierre Kocher (Mayo Clinic, Rochester, Bioinformatics); Michelle Hildebrandt (M.D. Anderson Cancer Center, Epidemiology); Richard Weinshilboum (Mayo Clinic, Rochester, Molecular Pharmacology and Experimental Therapeutics); Liewei Wang (Mayo Clinic, Rochester, Molecular Pharmacology and Experimental Therapeutics); |
| Short Abstract: Selecting informative subset of genes to predict clinical outcome as biological response requires appropriate methods due to the increased dimensionality in data. In this study, we investigated several machine learning methods to identify markers that differentiate responder and non-responder groups using basal mRNA expression, and gemcitabine cytotoxicity phenotypes obtained from lymphoblastoid cell lines. |
| Long Abstract: Click Here |
| Poster L04 |
| Predicting Pathway Membership via Domain Signatures |
| Holger Froehlich- German Cancer Research Center (DKFZ) |
| No additional authors |
| Short Abstract: We present a classification model, which for a specific gene of interest can predict the mapping to a KEGG pathway, based on its domain signature. The classifier makes explicit use of the hierarchical organization of pathways in the KEGG database.The method is available as R package gen2pathway. |
| Long Abstract: Click Here |
| Poster L05 |
| Eukaryotic Transcription Start Site Recognition Involving Non-Promoter Model |
| Yasuo Matsuyama- Dept. of Computer Science and Engineering |
| Koukei Kawasaki (Cybozu, R & D); Toru Hotta (Waseda University, Computer Science and Engineering); Ryosuke Mizutani (Waseda University, Computer Science and Engineering); Mika Takata (Waseda University, Computer Science and Engineering); Ayaka Ishida (Waseda University, Computer Science and Engineering); |
| Short Abstract: Eukaryotic transcription start sites are recognized using a committee SVM machine. The committee machine comprises a spectrum kernel estimator for a sequence similarity, an HMM-based promoter recognizer, and a new FFT-based non-promoter recognizer. Top-class scores in view of ROC curves are obtained by utilizing EM, FFT, HMM and SVM. |
| Long Abstract: Click Here |
| Poster L06 |
| Application of A Genetic Algorithm in Molecular Evolution Simulation |
| Fredrick Kinyua- University of the Western Cape |
| No additional authors |
| Short Abstract: This study highlights the use of a Genetic Algorithm in simulating the evolutionary transformation of a gene to its orthologous equivalent. The algorithm also computes the Genetic Distance between the two orthologs, by enumerating the number of generations needed for the complete transformation. |
| Long Abstract: Click Here |
| Poster L07 |
| Learning to Predict Expression Efficacy of Tags in Recombinant Protein Production |
| Wen-Ching Chan- Academia Sinica |
| Cheng-Ju Kuo (Academia Sinica, Institute of Information Science); Yu-Shi Lin (Academia Sinica, Institute of Information Science); Ueng-Cheng Yang (National Yang-Ming University, INSTITUTE OF BIOMEDICAL INFORMATICS); Po-Huang Liang (Academia Sinica, Institute of Biological Chemistry); Chun-Nan Hsu (Academia Sinica, Institute of Information Science); |
| Short Abstract: The production of recombinant proteins is one of important techniques of biotechnology. It is a challenging issue to predict whether target proteins could be over-expressed by which kinds of fusion tags. We applied SVMs to the characters from DNA and protein sequences and achieved acceptable results for recall and precision. |
| Long Abstract: Click Here |
| Poster L08 |
| Interpretable SVMs for regulatory elements defining alternative and constitutive splicing |
| Murlidharan Nair- Indiana University South Bend |
| No additional authors |
| Short Abstract: In this study we try to determine the extent to which characteristic and regulatory elements are important when in differentiating constitutive and alternative splicing using support vector machines (SVMs). Since SVMs lack interpretability, we have used a combination of methods to rank the characteristic features that were critical for classification. |
| Long Abstract: Click Here |
| Poster L09 |
| A kernel-based integration framework for high-throughput data sources in clinical decision support |
| Anneleen Daemen- KULeuven |
| Olivier Gevaert (Katholieke Universiteit Leuven, Electrical Engineering (ESAT-SCD)); Annelies Debucquoy (University Hospital Gasthuisberg Leuven, Radiation Oncology); Jean-Pascal Machiels (Universite Catholique de Louvain, Medical Oncology); Karin Haustermans (University Hospital Gasthuisberg Leuven, Radiation Oncology); Bart De Moor (Katholieke Universiteit Leuven, Electrical Engineering (ESAT-SCD)); |
| Short Abstract: A kernel-based framework is presented for the development of classifiers in clinical decision support in which high dimensional data sources can be combined over time and multiple levels in the genome. |
| Long Abstract: Click Here |
| Poster L10 |
| A graph-based method for phenotype change detection from microscopic cell images |
| Yi-Hung Huang- Academia Sinica |
| Yu-Shih Lin (Academia Sinica, Institute of Information Science); Chung-Chih Lin (National Yang-Ming University, Life Sciences); Yuh-Show Tsu (Chung Yuan Cgristian University, Biomedical Engineering); |
| Short Abstract: We propose a graph based approach to drug screening. Given a set of cell images acquired under different drug influence conditions (dosage, time, etc.), our method can quantify correlations between drug influence and cellular phenotypic change, such as mitochondria segmentation, by creating a feature space that minimizes regularized graph energy. |
| Long Abstract: Click Here |
| Poster L11 |
| Prediction of Subcellular Localization for Grape Proteome based on Secondary Structure Information Using Support Vector Machine with Probability Estimate |
| Wengang Zhou- Iowa State University |
| John Van Hemert (Iowa State University, Department of Electrical and Computer Engineering); Julie Dickerson (Iowa State University, Department of Electrical and Computer Engineering); |
| Short Abstract: Two learned support vector machine classifiers are used to predict subcellular locations with secondary structural information as the feature for the entire grape proteome. More than 80% of predicted grape protein locations match with each other. 64% of 1371 grape proteins match with their orthologous Arabidopsis proteins with known locations. |
| Long Abstract: Click Here |
| Poster L12 |
| Degradome project: Insights into intracellular protein degradation |
| Guy Zinman- Carnegie Mellon University |
| Grace Huang (University of Pittsburgh, Computational Biology); Thomas S Jensen (Technical University of Denmark, Center for Biological Sequence Analysis); Soren Brunak (Technical University of Denmark, Center for Biological Sequence Analysis); |
| Short Abstract: Protein degradation is one of the most important, yet less understood intracellular processes. The Degradome project is a novel in-silico computational approach aimed at characterizing ubiquitinated and actively degraded proteins by their physical properties and features, and finding potential actively-degraded substrates. |
| Long Abstract: Click Here |
| Poster L13 |
| Machine Learning’s Techniques applied in the study of the profiles of particular and common gene expression of autoimmune diseases and cancer |
| Beatriz Pinto- University of São Paulo |
| Cristina Junta (University of São Paulo, Department of Genetics); Geraldo Passos Junior (University of São Paulo, Department of Genetics); Silvana Giuliatti (University of São Paulo, Department of Genetics); |
| Short Abstract: Autoimmune Diseases are characterized by spreading out some immune system's reactions against its own cells and tissues. These diseases are closely related to malignant immunoproliferative disorders in a bidirectional association. The purpose of this project is to study these co-associations, using gene expression data from microarrays and Machine Learning’stechniques. |
| Long Abstract: Click Here |
| Poster L14 |
| A Non-Parametric Bayesian Approach for Predicting RNA Secondary Structures |
| Kengo Sato- Japan Biological Informatics Consortium |
| Taishin Kin (Computational Biology Research Center, National Institution of Advanced Industrial Science and Technology); Yasubumi Sakakibara (Keio University, Department of Biosciences and Informatics); Kiyoshi Asai (University of Tokyo, Department of Computational Biology); |
| Short Abstract: We propose a non-parametric Bayesian approach for predicting RNA secondary structures based on hierarchical Dirichlet processes for stochastic context-free grammars (HDP-SCFGs). Our results show that HDP-SCFGs are more accurate than the MFE-based models and the other existing generative models, and comparable with CONTRAfold. |
| Long Abstract: Click Here |
| Poster L15 |
| Predicting Protein-RNA Binding Sites Using Structural Information |
| Cornelia Caragea- Iowa State University |
| Cornelia Caragea (Iowa State University, Computer Science); Michael Terribilini (Iowa State University, Computational Biology); Jivko Sinapov (Iowa State University, Computer Science); Jae-Hyung Lee (Iowa State University, Computational Biology); Fadi Towfic (Iowa State University, Computational Biology); Drena Dobbs (Iowa State University, Computational Biology); Vasant Honavar (Iowa State University, Computer Science); |
| Short Abstract: RNA molecules play diverse functional roles in cells. Understanding the molecular mechanisms by which proteins recognize and bind RNA is essential for comprehending the functional implications of these interactions. In this study, we use machine learning algorithms for training classifiers to predict protein-RNA interfaces using information derived from the sequence. |
| Long Abstract: Click Here |
| Poster L16 |
| Using Global Sequence Similarity Improves Biological Site-Specific Classifiers |
| Jivko Sinapov- Iowa State University |
| Cornelia Caragea (Iowa State University, Computer Science); Drena Dobbs (Iowa State University, Computational Biology); Vasant Honavar (Iowa State University, Computer Science); |
| Short Abstract: Many bioinformatics problems involve the prediction of class labels for each residue in a protein sequence (e.g., prediction of RNA-binding residues, post-translational modification sites, etc.). We present a Hierarchical Mixture of Experts model that considers global sequence similarity between protein sequences in addition to local features used for classification. |
| Long Abstract: Click Here |
| Poster L17 |
| Annotation of complex cellular dynamics in high-throughput live cell imaging experiments |
| Michael Held- ETH Zurich |
| Bernd Fischer (Institute of Computational Science, ETH Zurich, Computer Science); Joachim Buhmann (Institute of Computational Science, ETH Zurich, Computer Science); Michael Schmitz (Institute of Biochemistry, ETH Zurich, Biology); Daniel Gerlich (Institute of Biochemistry, ETH Zurich, Biology); |
| Short Abstract: High throughput live cell imaging provides an ideal framework to systematically address complex cellular phenotypes. Our work focuses on the implementation of computational methods for quantitative annotation of time-resolved imaging data on a single cell level. We present an automated pipeline, for detection, tracking, and classification of cells over time. |
| Long Abstract: Click Here |
| Poster L18 |
| Integrating direct and indirect interaction data using pathway kernels to predict influences on a synthetic network |
| Marc Hulsman- TU Delft |
| Marcel J.T. Reinders (TU Delft, Electrical Engineering, Mathematics and Computer Science); |
| Short Abstract: To learn if one gene can influene another, we build a classifier on (directed) interaction pathways, integrating both direct (e.g. 2-hybrid) and indirect (e.g. co-expression) data sources. This is used to determine if certain genes, used within a synthetic network, still have unwanted interactions with the rest of the biological network. |
| Long Abstract: Click Here |
| Poster L19 |
| Discovering hairpin RNA motifs using generative models |
| Hilal Kosucu- University of Toronto |
| No additional authors |
| Short Abstract: We present a probabilistic approach for characterizing the sequence and structure preferences of RNA binding proteins (RBPs). Our program uses the semi-quantitative microarray-based measurements of RBP binding affinity to particular RNA sequences and fits motif models by maximizing the likelihood of these binding preferences. |
| Long Abstract: Click Here |
| Poster L20 |
| Selecting binary classifiers for multi-class classification problems via L1 penalties |
| Yuichi Shiraishi- RIKEN Advanced Science Institute |
| Mariko Hatakeyama (RIKEN Advanced Science Institute, Advanced Computational Sciences Department); |
| Short Abstract: Combining binary classifiers for multi-class classification is popular approach. However, some binary classifiers are often irrelevant to the whole classification system. In this paper, we propose to reduce several binary classifiers for multi-class classification by using L1 type penalty terms. |
| Long Abstract: Click Here |
| Poster L21 |
| An Approach to Predict Transmembrane Protein Structure with Stochastic Dynamical Systems Using Backward Smoothing Scheme |
| Takashi Kaburagi- Waseda University |
| Takashi Matsumoto (Waseda University, Electric Engineering and Bioscience); |
| Short Abstract: A backward smoothing approach utilizing a stochastic dynamical system with two-dimensional vector trajectories is used to predict transmembrane protein structures. Given a sequence of amino acids with unknown structures, the presence/absence of each residue in a transmembrane region is predicted by a backward smoothing process. |
| Long Abstract: Click Here |
| Poster L22 |
| A two level machine learning approach to the prediction of beta sheet structures |
| Merja Oja- University of Washington |
| William Stafford Noble (University of Washington, Department of Genome Sciences); |
| Short Abstract: We treat prediction of beta strand interactions as a structured output machine learning problem. An initial classifier predicts individual interactions, and a second classifier adjusts the topology of thepredicted graph. We achieve good results, and we demonstrate that further improving the initial classifier will yield even greater improvements overall. |
| Long Abstract: Click Here |
| Poster L23 |
| Indirect Learning Generative Models of Microtubule Distribution from fluorescence microscopy images |
| Aabid Shariff- Carnegie Mellon University |
| Gustavo Rohde (Carnegie Mellon University, Biomedical Engineering); Robert Murphy (Carnegie Mellon University, Biological Sciences, Biomedical Engineering, Machine Learning); |
| Short Abstract: An automated method to extract information on microtubules from fluorescence microscopy images has been a challenge due to microtubule structure complexity and technology limitations. Our method attempts the above without the use of segmentation/tracing. We develop a generative model that can recover parameters (e.g. number of microtubules) from images. |
| Long Abstract: Click Here |
| Poster L24 |
| Discovery of HuR binding sites using a machine learning approach |
| Shweta Bhandare- University of Colorado at Boulder |
| Jon Miller (University of Colorado at Boulder, Computer Science; Chemistry and Biochemistry); Paul Johnson (University of Colorado at Boulder, Computer Science); Nick Farina (University of Colorado at Boulder, Molecular, Cellular, and Developmental Biology); |
| Short Abstract: HuR is an mRNA binding protein that leads to mRNA stabilization and translation. Although a probabilistic binding motif has been identified, it is insufficient to accurately predict HuR binding targets. We developed a model combining motif searching and machine learning that greatly improves the accuracy of HuR binding prediction. |
| Long Abstract: Click Here |
| Poster L25 |
| A Bayesian combinatorial partitioning method for analyzing gene-gene interactions |
| An-Kwok Wong- University of Pittsburgh |
| Shyam Visweswaran (University of Pittsburgh, Department of Biomedical Informatics); |
| Short Abstract: We have developed a Bayesian combinatorial partitioning (BCP) method for analyzing genetic effects on a dichotomous outcome variable that exhaustively evaluates combinations to detect epistatic interactions among genetic markers. In preliminary evaluations, BCP was faster and had greater power than multifactor dimensionality reduction (MDR), a widely used method. |
| Long Abstract: Click Here |
| Poster L26 |
| Comparing Sequence and Structure-based Classifiers for Predicting RNA Binding Sites in Specific Families of RNA Binding Proteins |
| Deepak Reyon- Iowa State University |
| Michael Terribilini (Iowa State University, Bioinformatics & Computational Biology); Cornelia Caragea (Iowa State University, Computer Science); Jeffry Sander (Iowa State University, Bioinformatics & Computational Biology); Jae-Hyung Lee (Iowa State University, Bioinformatics & Computational Biology); Robert L. Jernigan (Iowa State University, Biochemistry, Biophysics & Molecular Biology); Vasant Honavar (Iowa State University, Computer Science); Drena Dobbs (Iowa State University, Genetics, Development, & Cell Biology); |
| Short Abstract: We evaluate machine learning classifiers for predicting RNA-binding residues in proteins, using either sequence-based information only, or a combination of sequence and structure-derived information and quantitate relative contributions of these different input types to overall prediction performance. We also present novel classifiers optimized for specific families of RNA binding proteins. |
| Long Abstract: Click Here |
| Accepted Posters |
| View Posters By Category |
- A) Arrays
- B) Bioinformatics of Health and Disease
- C) Chemical and Pharmaceutical Informatics
- D) Comparative Genomics
- E) Databases
- F) Evolution
- G) Functional Genomics
- H) Gene Prediction
- I) Genome Annotation
- J) Genomics
- K) Interactions
- L) Machine Learning
- M) Population Genetics and Variation
- N) Proteomics
- O) Regulation
- P) Sequence Analysis
- Q) Structure and Function Prediction
- R) Systems Biology and Networks
- S) text mining
- T) Other (includes posters with fewer than 10 submissions)
| Search Posters: |