A Bayesian Network Model of Proteins' Association with Promyelocytic Leukemia (PML) Nuclear Bodies

Mikael Bodén, Graham Dellaire, Kevin Burrage, Timothy L. Bailey
2010 Journal of Computational Biology  
The modularity that nuclear organization brings has the potential to explain the function of aggregates of proteins and RNA. Promyelocytic leukemia nuclear bodies are implicated in important regulatory processes. To understand the complement of proteins associated with these intra-nuclear bodies, we construct a Bayesian network model that integrates sequence and protein-protein interaction data. The model predicts association with promyelocytic * To whom correspondence should be addressed 1
more » ... emia nuclear bodies accurately when interaction data is available. At a false positive rate of 10%, the true positive rate is almost 50%, indicated by an independent nuclear proteome reference set. The model provides strong support for further expanding the protein complement with several important regulators and a richer functional repertoire. Using special SVM-nodes (equipped with string kernels), the Bayesian network is also able to produce predictions on the basis of sequence only, with an accuracy superior to that of baseline models.
doi:10.1089/cmb.2009.0140 pmid:20426694 fatcat:oyvos6ovvzb4jniubxsruflpxy