Skip to content

Ph.D: Weizmann Institute of Science, Israel
Position: Lecturer
Department of Microbiology and Immunology
Faculty of Health Sciences
E-mail: erubin@bgu.ac.il 

New methods for the identification of novel multidimensional predictors of treatment outcome

  • Background

Phenotypes, or the observable traits of an individual, arise from a complex interaction of genes, the environment and stochastic processes. One of the major challenges of modern biology is to understand the role genetics plays in the development of diseases or phenotypes with a deleterious affect on life quality and span. Despite the completion of the human genome sequencing project, we are only beginning to appreciate the complexity of the translation of genes to phenotypes, or of genetic variation to phenotypic variation. It is currently impossible to computationally predict which genes are involved in disease processes, to suggest how a new drug will affect health, or understand why some patients will benefit from using some drug, while others with the same disease will develop severe side effects.

To study the processes that govern the translation of genetic variation into phenotypic variation in medicine, there is a need for sufficiently rich phenotypic databases, and a simplified yet relevant system and computational and mathematical tools for their analysis. In our lab, we are developing computational and mathematical tools for the study of phenotypes and their formation. As a model system, we are using Mendelian diseases, the relatively rare cases where a severe phenotype – a disease – arises from a single genetic variation. We are also developing tools for the use of clinical repositories and knowledge bases as phenotypic data sources.

  • Current research
  1. Novel methods for disease stratification – We are currently developing new methods to stratify patients using medical records. We identify novel multi-dimensional predictors of treatment outcome. This research is expected to provide superior designs for clinical trials. In addition, we are using genetic analysis of clinically normal values as predictors of treatment outcomes.

  2. Improved methods for computational target prioritization in genetic association studies – We are extending existing methods for the prediction of genes likely to be associated with Mendelian diseases using insight and data obtained from systems biology studies, testing the relative strengths of different methods and using more detailed descriptions of diseases. For example, we are developing a system that seeks a connection between a gene’s ability to be associated with a Mendelian disease and its place in cellular networks, i.e. genetic control, protein interaction, and metabolic networks.

  3. An end-user programming language for biologists – We have developed (in collaboration with Harvard University and MIT) a programming language dedicated to biology. In this language, users who are familiar with Excel can automate many tasks that are required for bioinformatics with only a few hours of training. We are working to extend this language and its interface, and seek to enlarge our user community as a way to ensure continuing support.

  • Selected publications

Ner-Gaon H., Halachmi R., Savaldi-Goldstein S., Rubin E., Ophir R. and Fluhr R. (2004) Intron retention is a major phenomenon in alternative splicing in Arabidopsis. Plant J., 39:877-885.

Ben-Dor S., Esterman N., Rubin E. and Sharon N. (2004) Biases and complex patterns in the residues flanking protein N-glycosylation sites. Glycobiology, 14:95-101.

Shachak A., Ophir R. and Rubin E. (2005) Applying instructional-design theories to bioinformatics education: Development of conditions of learning-based microarray analysis and primer design workshops. Cell Biology Education, 4:199-206.