TMHMM

TMHMM predicts transmembrane helices in protein sequences using a hidden Markov model (HMM) to determine membrane topology and distinguish membrane versus soluble proteins.

Key Features:

• Hidden Markov model (HMM): Uses a hidden Markov model to model transmembrane helices and membrane topology.
• Transmembrane helix prediction: Predicts the locations of transmembrane helices in protein sequences.
• Topology prediction: Predicts membrane topology including orientation designations such as N(in)-C(in) and N(out)-C(in).
• Membrane vs soluble classification: Classifies proteins as membrane or soluble with reported specificity and sensitivity exceeding 99%.
• Helix prediction accuracy: Reports predictive accuracy of approximately 97–98% for transmembrane helices.
• Signal peptide caveat: Prediction accuracy can be reduced in the presence of signal peptides.
• Genome-scale prevalence estimates: Based on predictions, estimates that roughly 20–30% of genes in most genomes encode membrane proteins.
• Topology preference detection: Identifies a general preference for N(in)-C(in) topologies across organisms and notes an exception in Caenorhabditis elegans due to many 7TM receptors producing more N(out)-C(in) topologies.
• Validation: The method has been rigorously validated and benchmarked.

Scientific Applications:

• Genome-wide membrane protein identification: Enables reliable identification of integral membrane proteins across genomes.
• Estimating membrane protein abundance: Supports estimation that ~20–30% of genes encode membrane proteins in most genomes.
• Topology and assembly studies: Provides data on topology preferences useful for studying membrane protein assembly mechanisms.
• Structural biology support: Informs structural biology analyses of membrane protein architecture and orientation.

Methodology:

Applies a hidden Markov model (HMM) to predict transmembrane helices and membrane topology.