TMB-Hunt

TMB-Hunt classifies protein sequences as transmembrane beta-barrel (TMB) or non-TMB by analyzing whole-sequence amino acid composition with a modified k-nearest neighbour (k-NN) algorithm.

Key Features:

• Classification algorithm: Uses a modified k-nearest neighbour (k-NN) algorithm to distinguish TMB and non-TMB sequences.
• Sequence representation: Analyzes whole-sequence amino acid composition as the input feature set.
• Weighted residues: Applies differentially weighted amino acids to emphasize informative residues.
• Evolutionary information: Incorporates evolutionary information into the classification process.
• Performance validation: Performance assessed by cross-validation, reporting 92.5% discrimination accuracy.
• Throughput: Capable of screening up to 10,000 protein sequences per query.

Scientific Applications:

• TMB identification: Identification and classification of transmembrane beta-barrel proteins in sequence datasets and proteomes.
• Membrane protein studies: Distinguishing TMBs from non-TMB proteins to support studies of bacterial outer membrane proteins and other beta-barrel-containing systems.
• High-throughput screening: Large-scale screening of protein sequence collections for candidate TMBs.

Methodology:

Analysis of whole-sequence amino acid composition using a modified k-nearest neighbour (k-NN) algorithm with differentially weighted amino acids and incorporation of evolutionary information, with performance evaluated by cross-validation (92.5% accuracy).