TI2BioP

TI2BioP predicts proteinaceous bacteriocin-like sequences using topological indices computed from inferred and artificial two-dimensional protein structures in an alignment-free framework for quantitative sequence–function relationship (QSFR) modeling.

Key Features:

• Alignment-Free Prediction: Employs a TOPS-MODE and MARCH-INSIDE–inspired strategy to compute topological indices from inferred and artificial 2D protein structures without sequence alignment.
• Calculation of Spectral Moments: Computes spectral moments as topological indices to develop quantitative sequence–function relationship (QSFR) models.
• Hydrophobicity and Basicity Considerations: Derives spectral moments ((HP)μ(k)) from artificial secondary structures by clustering amino acids into a Cartesian system based on hydrophobicity and polarity to capture hydrophobicity and basicity effects.
• Application to Bacteriocins: Applied to 196 bacteriocin-like sequences and 200 representative CATH domains to develop an alignment-free QSFR model with 76.92% discrimination accuracy for identifying bacteriocin proteins.
• Performance and Validation: Reports overall prediction performance of 72.16% with a bacteriocin detection rate of 66.7%, outperforming InterProScan (60.2%), and validated by predicting the cryptic bactericide function of the Cry 1Ab C-terminal domain from Bacillus thuringiensis endotoxin.

Scientific Applications:

• Bacteriocin Identification: Identification and characterization of proteinaceous bacteriocins using alignment-free topological indices.
• Structure–Function Modeling: Development of QSFR models linking computed spectral moments with bacteriocin function.
• Cryptic Function Discovery: Detection of cryptic functional domains, exemplified by prediction of bactericide activity in the Cry 1Ab C-terminal domain from Bacillus thuringiensis.

Methodology:

Topological indices and spectral moments are computed using methods inspired by TOPS-MODE and MARCH-INSIDE on inferred/artificial 2D protein structures; amino acids are clustered into a Cartesian hydrophobicity–polarity system to calculate (HP)μ(k) spectral moments, which are used to build QSFR discrimination models validated against CATH domains and compared to InterProScan.