Challenges and open problems in computational prediction of protein complexes: the case of membrane complexes

TL;DR

This paper discusses the unique challenges in computationally predicting membrane protein complexes, emphasizing recent experimental advances and open problems in the field.

Contribution

It highlights the specific difficulties in membrane complex prediction and discusses new experimental techniques that could address these challenges.

Findings

Membrane complexes are often overlooked due to detection difficulties.

Recent techniques like MY2H enable detection of membrane protein interactions.

Open problems remain in effectively predicting membrane complexes.

Abstract

Identifying the entire set of complexes is essential not only to understand complex formations, but also to map the high level organisation of the cell. Computational prediction of protein complexes faces several challenges including the lack of sufficient protein interactions, presence of noise in protein interaction datasets and difficulty in predicting small and sparse complexes. These challenges are covered in most reviews of complex prediction methods. However, an important challenge that needs to be addressed is the prediction of membrane complexes. These are often ignored because existing protein interaction detection techniques do not detect interactions between membrane proteins. But, recently there have been several new experimental techniques including MY2H that are capable of detecting membrane protein interactions. In the light of this new data, we discuss here new challenges and the kind of open problems that need to be solved to effectively detect membrane complexes.