| How can genome-wide expression data gathered from gene chips and related techniques contribute to the description and definition of the functional role of uncharacterized proteins that have been discovered through the genome projects? This work systematically explores the relationship between expression data and various protein attributes, such as protein families, protein abundance levels, subcellular localization, biological function, and protein-protein interactions in the model organism yeast. Each of these attributes provides a framework for interpreting expression data. Conversely, expression data can aid in the prediction of some of these properties. One of the main conclusions of this work is that although much of the genomic information available at this point is noisy and error-prone, there is a substantial benefit in integrating multiple and even heterogeneous genomic datasets. This is illustrated in a case example of integrating various types of genomic data to validate and predict protein complexes. |
