Search

Gene expression is modulated by transcription factors (TFs), which are proteins that generally bind to DNA adjacent to coding regions and initiate transcription. Each target gene can be regulated by more than one TF, and each TF can regulate many targets. For a complete molecular understanding of transcriptional regulation, researchers must first associate each TF with the set of genes that it regulates. Here we present a summary of completed work on the ability to associate 104 TFs with their binding sites using support vector machines (SVMs), which are classification algorithms based in statistical learning theory. We use several types of genomic datasets to train classifiers in order to predict TF binding in the yeast genome. We consider motif matches, subsequence counts, motif conservation, functional annotation, and expression profiles. A simple weighting scheme varies the contribution of each type of genomic data when building a final SVM classifier, which we evaluate using known binding sites published in the literature and in online databases. The SVM algorithm works best when all datasets are combined, producing 73% coverage of known interactions, with a prediction accuracy of almost 0.9. We discuss new ideas and preliminary work for improving SVM classification of biological data.

VisANT is a web-based software framework for visualizing and analyzing many types of networks of biological interactions and associations. Networks are a useful computational tool for representing many types of biological data, such as biomolecular interactions, cellular pathways and functional modules. Given user-defined sets of interactions or groupings between genes or proteins, VisANT provides: (i) a visual interface for combining and annotating network data, (ii) supporting function and annotation data for different genomes from the Gene Ontology and KEGG databases and (iii) the statistical and analytical tools needed for extracting topological properties of the user-defined networks. Users can customize, modify, save and share network views with other users, and import basic network data representations from their own data sources, and from standard exchange formats such as PSI-MI and BioPAX. The software framework we employ also supports the development of more sophisticated visualization and analysis functions through its open API for Java-based plug-ins. VisANT is distributed freely via the web at http://visant.bu.edu and can also be downloaded for individual use.

Background Information obtained from diverse data sources can be combined in a principled manner using various machine learning methods to increase the reliability and range of knowledge about protein function. The result is a weighted functional linkage network (FLN) in which linked neighbors share at least one function with high probability. Precision is, however, low. Aiming to provide precise functional annotation for as many proteins as possible, we explore and propose a two-step framework for functional annotation (1) construction of a high-coverage and reliable FLN via machine learning techniques (2) development of a decision rule for the constructed FLN to optimize functional annotation. Results We first apply this framework to Saccharomyces cerevisiae. In the first step, we demonstrate that four commonly used machine learning methods, Linear SVM, Linear Discriminant Analysis, Naïve Bayes, and Neural Network, all combine heterogeneous data to produce reliable and high-coverage FLNs, in which the linkage weight more accurately estimates functional coupling of linked proteins than use individual data sources alone. In the second step, empirical tuning of an adjustable decision rule on the constructed FLN reveals that basing annotation on maximum edge weight results in the most precise annotation at high coverages. In particular at low coverage all rules evaluated perform comparably. At coverage above approximately 50%, however, they diverge rapidly. At full coverage, the maximum weight decision rule still has a precision of approximately 70%, whereas for other methods, precision ranges from a high of slightly more than 30%, down to 3%. In addition, a scoring scheme to estimate the precisions of individual predictions is also provided. Finally, tests of the robustness of the framework indicate that our framework can be successfully applied to less studied organisms. Conclusion We provide a general two-step function-annotation framework, and show that high coverage, high precision annotations can be achieved by constructing a high-coverage and reliable FLN via data integration followed by applying a maximum weight decision rule.