Identification and classification of conopeptides using profile Hidden Markov Models.

Area of Science:

• Marine biology
• Biochemistry
• Bioinformatics

Background:

• Conopeptides are toxins from marine snails (genus Conus) with significant potential in neuroscience and pharmacology.
• The vast majority of estimated conopeptides remain undiscovered, with only ~1000 described out of an estimated 1 million.
• Advancements in high-throughput sequencing are rapidly increasing the number of known conopeptides, necessitating efficient identification methods.

Purpose of the Study:

• To develop a fast and accurate computational method for identifying and classifying novel conopeptides from large datasets.
• To create predictive models for conopeptide superfamilies and families.

Main Methods:

• Construction of 62 profile Hidden Markov Models (pHMMs) based on protein sequences of known conopeptides.
• Testing the specificity and accuracy of pHMMs against the UniProtKB/Swiss-Prot database.
• Evaluating classification accuracy for mature, pro-, and signal peptides.

Main Results:

• The developed pHMMs demonstrated high specificity, with 56 out of 62 models showing no false positives in the UniProtKB/Swiss-Prot database.
• Accurate classification rates were achieved: 96% for mature peptide models and 100% for pro- and signal peptide models.
• The study represents the first application of this computational approach to predict all known conopeptide superfamilies and some families.

Conclusions:

• Profile Hidden Markov Models provide a robust and accurate computational tool for the identification and classification of conopeptides.
• These models are effective for annotating new conopeptides discovered through transcriptome and genome sequencing.
• The methodology facilitates the exploration of the vast, yet largely uncharacterized, conopeptide repertoire for potential therapeutic applications.