SPOT-Struct-DNA

SPOT-Struct-DNA predicts DNA-binding proteins from three-dimensional protein structures to identify proteins that interact with DNA.

Key Features:

• Volume-Fraction Correction: Introduces a volume-fraction correction to a statistical energy function derived from a distance-scaled, finite, ideal-gas reference (DFIRE) state.
• Performance Metrics: Reports a Matthews correlation coefficient (MCC) of 0.76 with accuracy 98%, precision 93%, and sensitivity 64%, improving over a previous best MCC of 0.69.
• Integration with TM-align: Utilizes the structural alignment program TM-align to align target protein structures with known DNA-binding templates.
• Template-Based Criteria: Requires both structural similarity and predicted binding affinity between the target protein and DNA for positive predictions.

Scientific Applications:

• Structural Genomics Screening: Applied to large structural genomics datasets to identify potential DNA-binding candidates.
• Benchmark Evaluation: Validated on a benchmark containing 179 known DNA-binding proteins and 3,797 non-binding proteins to assess discrimination performance.
• Large-Scale Target Analysis: Analyzed 2,235 structural genomics targets, identifying 37 as DNA-binding proteins, of which 27 (73%) were putatively confirmed, one was annotated without DNA-binding functions, and the remainder had unknown functions.

Methodology:

Employs a template-based approach using TM-align for structural alignment and incorporates a volume-fraction correction into the DFIRE statistical energy function, requiring structural similarity and predicted binding affinity between target proteins and DNA.