Nucleos

Nucleos predicts nucleotide-binding sites in protein structures using sequence and structural methodologies to identify positions of phosphate, nucleobase, and carbohydrate moieties.

Key Features:

• Modular Structure Approach: Exploits the modular nature of nucleotide-binding sites by treating nucleotides as phosphate, nucleobase, and carbohydrate fragments bound by recurring structural motifs across protein folds.
• Template-Based Prediction: Compares a query protein against a database of templates derived from proteins with known structures to predict fragment binding-site positions and transfer fragment positions to the query.
• Surface Filtering and Scoring: Discards predictions located on the protein surface and scores remaining predictions using clustering and conservation metrics.

Scientific Applications:

• High Prediction Accuracy: Ranks a correct prediction as first in 48% of cases for the nucleobase, carbohydrate, and phosphate fragments, and for the top five predictions achieves 71%, 65%, and 86% respectively.
• Binding Site Reconstruction: Reconstructs full or partial binding-site structures from predicted fragment positions and ranks a reconstructed binding site as first in 59% of analyzed proteins.
• Unbound Protein Prediction: Demonstrates reliable prediction of nucleotide-binding sites in unbound proteins despite structural variations caused by presence or absence of the nucleotide.

Methodology:

Compares query proteins to a template database of known structures, identifies matching structural motifs for phosphate, nucleobase, and carbohydrate fragments, transfers fragment positions to the query, discards surface-localized predictions, and scores remaining predictions by clustering and conservation metrics.