Proteochemometric Method for pIC50 Prediction of Flaviviridae

Area of Science:

• Virology
• Computational Chemistry
• Drug Discovery

Background:

• Flaviviridae viruses pose significant health risks, necessitating novel therapeutic strategies.
• Antiviral peptides (AVPs) show promise, but experimental efficacy assessment is resource-intensive.
• Proteochemometric modeling (PCM) offers a computational alternative for predicting peptide bioactivity.

Purpose of the Study:

• To develop a computational model for predicting the bioactivity (pIC50) of AVPs against the Flaviviridae family.
• To facilitate the selection of effective AVPs by predicting their efficacy.
• To reduce the time and cost associated with experimental screening of AVPs.

Main Methods:

• Collected AVP data from public databases and curated Flaviviridae target sequences from literature.
• Utilized PCM by calculating sequence-derived descriptors for peptides and targets.
• Incorporated individual and cross-term features between peptide and target sequences for model development.

Main Results:

• Achieved a high prediction accuracy for pIC50 values of AVPs.
• Reported a coefficient of determination (R²) of 0.85.
• Obtained a Mean Absolute Percentage Error (MAPE) of 8.44%.

Conclusions:

• The developed PCM model effectively predicts AVP bioactivity against Flaviviridae.
• This computational approach can guide the selection of potent AVPs, accelerating antiviral drug development.
• PCM provides a cost-effective and efficient method for assessing AVP efficacy.