RocSampler

RocSampler (Regularizing Overlapping Complexes) predicts protein complexes from protein–protein interaction (PPI) networks using a sampling-based framework with overlap-aware regularization.

Key Features:

• Sampling-Based Complex Detection: Applies a sampling strategy to identify candidate protein complexes within PPI networks, enabling detection of small complexes with two or three components.
• Regularized Scoring Function: Incorporates regularization terms modeling overlap between predicted complexes and the size distribution of complexes to improve prediction accuracy for overlapping and variably sized complexes.

Scientific Applications:

• Protein Complex Identification in PPI Networks: Detects small and overlapping protein complexes in datasets such as yeast PPI networks, supporting analysis of cellular protein organization.

Methodology:

ROCsampler uses a sampling-based optimization approach on PPI networks, evaluates candidate complexes with a scoring function incorporating overlap and size-distribution regularization terms, and selects complexes that maximize the regularized objective.