MemLoci

MemLoci predicts the subcellular localization of proteins associated with or inserted into eukaryotic membranes to support functional annotation.

Key Features:

• Support Vector Machine-based classification: MemLoci employs a support vector machine (SVM) algorithm to perform localization classification of membrane proteins.
• Predicted localization classes: The tool discriminates among three membrane protein localizations: plasma membrane, internal membranes, and organelle membranes.
• Performance metrics: Independent testing reported an overall accuracy of 70% and a generalized correlation coefficient as high as 0.50.
• Specialization for membrane proteins: The method is specifically trained for membrane proteins rather than globular proteins to address localization challenges unique to membrane-associated sequences.

Scientific Applications:

• Proteomics: Provides localization predictions that assist in interpreting protein function and interactions within cellular membranes in proteomic studies.
• Functional genomics and hypothesis generation: Supports functional annotation and generation of hypotheses about protein roles when experimental localization data are unavailable or incomplete.

Methodology:

MemLoci uses a support vector machine classifier trained on membrane proteins to assign sequences to plasma membrane, internal membranes, or organelle membranes and was evaluated by independent testing reporting 70% overall accuracy and a generalized correlation coefficient up to 0.50.