Predicting function from sequence in a large multifunctional toxin family.

Area of Science:

• Biochemistry
• Toxicology
• Bioinformatics

Background:

• Venoms contain bioactive substances with therapeutic potential.
• Experimental characterization of toxin function is slow, necessitating in-silico prediction tools.
• Existing predictive tools for toxin activity have low accuracy due to protein family similarities and minor sequence variations.

Purpose of the Study:

• To develop and evaluate a novel classification model for predicting venom toxin activity.
• To improve the accuracy of in-silico prediction of toxin function from amino acid sequences.
• To provide a framework for comparing toxin active sites and identifying potential drug leads.

Main Methods:

• Developed a classification model using physico-chemical attributes from amino acid sequences.
• Utilized a dataset of over 250 viperid phospholipase A₂ toxins.
• Clustered proteins by sequence profiles and performed in-vitro activity tests on novel toxins and crude venoms.

Main Results:

• The new model demonstrated higher prediction accuracy compared to existing methods.
• The majority of in-vitro detected toxin activities aligned with the model's predictions.
• Identified a framework for comparing functional toxin sub-groups and active sites.

Conclusions:

• The developed physico-chemical attribute-based model offers a more accurate approach to predicting venom toxin activity.
• This method facilitates a targeted strategy for discovering novel drug leads from venom components.
• The findings support the use of in-silico tools for accelerating venom toxin research and development.