|
| [an error occurred while processing this directive] |
GO for Gene DocumentsPadmini SrinivasanSchool of Library and Information ScienceDepartment of Management Sciences Computer Science (courtesy appointment) College of Nursing (courtesy appointment) The University of Iowa
Friday, December 08, 2006
AbstractThe automatic annotation of genes and their products with Gene Ontology codes defines an important area of research. Gene Ontology consists of three hierarchies of codes representing characteristics related to molecular function, biological process and cellular component. One approach for annotation is to use the information available about these genes in the biomedical literature. Our goal based on this approach, is to explore automatic methods that could supplement the expensive manual annotation processes currently in place. Our annotation methods are built using a collection of Support Vector Machines classifiers. In this talk we present results from an ongoing sequence of experiments. We study different aspects of the problem such as the relationship of performance to hierarchy level and to the number of positives in the training sets. We find that hierarchy level is important especially for the molecular function and biological process hierarchies. Whereas the cellular component hierarchy stands apart in several respects. We suggest that this may be due to fundamental differences in link semantics. We also explore more relaxed criteria for classification correctness that exploits the hierarchical structure underlying the codes. Finally we present results from experiments that better approximate reality in that they are designed to handle a temporal stream of documents. We will conclude the talk with an outline of our next steps in this line of research. |
|
| Translate this page automatically. |
| ©2005 The University of Iowa, All Rights Reserved. |