PRED-TMBB

PRED-TMBB predicts transmembrane β-strands and overall topology of outer membrane β-barrel (OMBB) proteins and provides an OMBB classification score to support structure–function analyses and proteome-scale screening.

Key Features:

• Predictive capability: Predicts transmembrane β-strands and overall OMBB topology using a Hidden Markov Model (HMM) trained on non-homologous outer membrane proteins with atomic-resolution structures.
• Decoding methods: Provides three decoding methods to tailor prediction behavior to specific analysis needs.
• Output information: Produces predicted protein topology, an OMBB probability score, posterior probabilities for transmembrane strand positions, and a graphical depiction of predicted strand positions relative to the lipid bilayer.
• Discriminative training: The HMM is trained discriminatively to maximize classification/prediction accuracy rather than sequence likelihood, improving discrimination between β-barrel membrane proteins and water-soluble proteins.
• Accuracy and validation: Validation reports per-residue jackknife accuracy 84.2% (correlation coefficient 0.72), self-consistency accuracy 88.1% (correlation coefficient 0.824), and correct topology prediction for 10/14 proteins in self-consistency and 9/14 in jackknife tests.
• Classification success: Reported classification rates on large datasets are 88.8% for outer membrane proteins and 89.2% for water-soluble proteins.
• Proteome screening: Applied to whole-proteome scans (e.g., E. coli) to identify novel OMBB candidates for large-scale genomic and proteomic discovery.

Scientific Applications:

• OMBB discovery and annotation: Identifies and annotates candidate β-barrel outer membrane proteins in genomic and proteomic datasets.
• Structure–function studies: Provides topology predictions to inform membrane protein structural interpretation and functional hypotheses.
• Experimental candidate selection: Ranks sequences for selection as targets for biochemical characterization and structural determination.
• Antibiotic target exploration: Supports identification of outer membrane protein candidates relevant to antibiotic and vaccine target discovery.
• Bacterial physiology and pathogenesis research: Aids studies of Gram-negative bacterial envelope proteins involved in physiology and virulence.
• Proteome-scale screening: Enables large-scale scans of proteomes to discover novel OMBBs across organisms.

Methodology:

PRED-TMBB uses a discriminatively trained Hidden Markov Model fitted to a non-redundant set of 14 outer membrane proteins with atomic-resolution structures, is evaluated by self-consistency and jackknife validation, and offers three decoding methods with posterior-probability outputs and an OMBB probability score.