TargetP

TargetP predicts the subcellular localization of eukaryotic proteins and identifies N-terminal signal peptides and cleavage sites to assign proteins to mitochondria, chloroplasts, the secretory pathway, or other cellular localizations.

Key Features:

• Neural Network Algorithm: TargetP employs a neural network-based algorithm to analyze protein sequences for localization prediction.
• N-terminal Sequence Utilization: The method leverages only N-terminal sequence information, including signal peptides and presequences, to inform predictions.
• Subcellular Localization Classification: The tool classifies proteins into mitochondria, chloroplasts, the secretory pathway, or other localizations.
• Accuracy Rates: Reported accuracies are approximately 85% for plant proteins and 90% for non-plant proteins on redundancy-reduced test sets.
• Cleavage Site Prediction: Cleavage site prediction accuracy is reported at ~40–50% for chloroplastic and mitochondrial presequences and >70% for secretory signal peptides.
• Genomic Analysis Results: Applied to Arabidopsis thaliana chromosomes 2 and 4 and the Ensembl Homo sapiens protein set, TargetP estimated ~10% of plant proteins as mitochondrial, ~14% as chloroplastic, and ~10% secretory proteins in both Arabidopsis and humans.

Scientific Applications:

• Protein Function Annotation: Predicting localization provides cellular context to support functional annotation of proteins.
• Genomic Research: Generates compartmental distribution data to support large-scale analyses of genomes and proteomes.
• Biotechnological Applications: Prediction of targeting signals and cleavage sites aids the design of recombinant proteins with specific subcellular targeting.

Methodology:

TargetP utilizes only N-terminal sequence information and a neural network model trained on extensive datasets and evaluated on redundancy-reduced test sets.