Logistic Biclustering Models for Protein Network Inference

Protein interaction information is fundamental to understand the cellular processes. Due to that, much is being done to automatically infer protein networks from all types of biological data. In this work we propose a model based on logistic biclustering, to analyze biological data in binary format, and combine that with a plaid model analysis of microarray expression data to predict protein interactions. We compare the results with a supervised model applied to the same data sets, and the experiments show that the proposed method achieves a performance very close to the one of a supervised approach trained with up to 25% of knowledge of the target network.