Module networks revisited: computational assessment and prioritization of model predictions

The solution of high-dimensional inference and prediction problems in computational biology is almostalways a compromise between mathematical theory and practical constraints such as limitedcomputational resources. As time progresses, computational power increases but well-established inferencemethods often remain locked in their initial suboptimal solution. We revisit the approach ofSegal et al. (2003) to infer regulatory modules and their condition-specific regulators from gene expressiondata. In contrast to their direct optimization-based solution we use a more representativecentroid-like solution extracted from an ensemble of possible statistical models to explain the data. Theensemble method automatically selects a subset of most informative genes and builds a quantitativelybetter model for them. Genes which cluster together in the majority of models produce functionallymore coherent modules. Regulators which are consistently assigned to a module are more often supportedby literature but a single model always contains many regulator assignments not supported bythe ensemble. Reliably detecting condition-specific or combinatorial regulation is particularly hard ina single optimum but can be achieved using ensemble averaging.

Joshi, A., De Smet, R., Marchal, K., Van de Peer, Y., Michoel, T. (2009) Module networks revisited: computational assessment and prioritization of model predictions. Bioinformatics 25(4):490-6.

VIB / UGent
Bioinformatics & Evolutionary Genomics
Technologiepark 927
B-9052 Gent
+32 (0) 9 33 13807 (phone)
+32 (0) 9 33 13809 (fax)

Don't hesitate to contact the in case of problems with the website!