MemPype

MemPype predicts signal peptides, GPI anchors, all-alpha membrane topology (ENSEMBLE), and subcellular localization (MemLoci) of eukaryotic membrane proteins to support analysis of membrane protein topology and localization.

Key Features:

• Signal Peptide Prediction (SPEP): Identifies signal peptides associated with protein targeting and secretion.
• Glycophosphatidylinositol (GPI) Anchor Prediction (PredGPI): Predicts presence of GPI anchors with a reported false positive rate (FPR) of 0.8%, where a positive prediction indicates likely cell-surface exposure.
• All-Alpha Membrane Topology Prediction (ENSEMBLE): Predicts all-alpha membrane protein topology with a reported FPR below 1%.
• Subcellular Localization Discrimination (MemLoci): Classifies eukaryotic membrane proteins into 'cell membrane', 'internal membranes', and 'organelle membranes' with reported accuracy of 70% and a generalized correlation coefficient (GCC) of 0.50 on a homology-unbiased validation set.
• Homology-Based Inheritance: Retrieves homologs to support annotation by identifying up to 25 similar proteins per query based on sequence identity ≥50% and alignment coverage ≥50%.

Scientific Applications:

• Cell biology studies: Provides topology and localization predictions to inform cellular role assignment of membrane proteins.
• Molecular biology and functional annotation: Supplies signal peptide, GPI anchor, and topology information to assist functional annotation of eukaryotic proteins.
• Interaction network analysis: Delivers localization and membrane topology data that can be used to contextualize protein–protein interaction networks.

Methodology:

For each query, up to 25 homologs are retrieved using criteria of sequence identity ≥50% and alignment coverage ≥50%; PredGPI, SPEP, ENSEMBLE, and MemLoci predictions are computed for each protein and integrated to produce localization and topology classifications.