ProInflam

ProInflam predicts the proinflammatory potential of peptide epitopes to support studies of antigen-induced inflammation.

Key Features:

• Data-Driven Approach: ProInflam leverages a dataset of 729 experimentally validated proinflammatory epitopes and 171 non-proinflammatory epitopes sourced from the Immune Epitope Database (IEDB).
• Amino Acid Sequence Analysis: It identifies amino acids A, F, I, L, V and dipeptides AF, FA, FF, PF, IV, IN as preferentially associated with proinflammatory epitopes.
• Machine Learning Models: The tool trains models using compositional and motif-based features extracted from epitope amino acid sequences via machine learning techniques.
• High Predictive Performance: The hybrid model combining motif and dipeptide-based features achieves a Matthews Correlation Coefficient (MCC) of 0.58 and an accuracy of 87.6%.

Scientific Applications:

• Identify potential vaccine candidates: Predicts peptides likely to elicit proinflammatory responses to inform antigen selection for vaccine design.
• Explore autoimmune disease mechanisms: Provides insights into antigenicity that may elucidate pathways of inappropriate inflammatory responses in autoimmune conditions.
• Develop therapeutic interventions: Identifies proinflammatory peptides to guide development of targeted therapies that modulate immune responses.

Methodology:

Data collection and analysis from IEDB to identify key residues and dipeptides; extraction of compositional and motif-based sequence features; machine learning model development and validation against experimental epitope data.