ProIn-Fuse

ProIn-Fuse predicts proinflammatory peptides (PIPs) from peptide sequences to identify peptides involved in immune signaling for applications in vaccine development and immunotherapeutic drug design.

Key Features:

• Multiple Feature Representation Fusion: Integrates multiple sequence feature representations to create a fused representation that leverages diverse encoding schemes.
• Probabilistic Scoring with Random Forest Models: Trains random forest classifiers on feature representations to generate probabilistic scores for each peptide and linearly combines those scores to produce the final prediction.
• Eight Sequence Encoding Schemes: Employs eight distinct sequence encoding schemes to capture diverse aspects of peptide sequences in the feature representation learning.
• Improved Accuracy: Reports an accuracy of 0.746, representing approximately a 10% improvement over prior state-of-the-art predictors.
• FASTA Input Support: Operates on peptide sequences provided in FASTA format.

Scientific Applications:

• Immunology research: Enables computational identification of PIPs to support studies of immune signaling and inflammatory responses.
• Vaccine candidate exploration: Aids prioritization of peptide candidates relevant to vaccine development by predicting proinflammatory potential.
• Immunotherapeutic and pharmacology applications: Supports discovery and prioritization of immunotherapeutic agents and accelerates peptide-based drug discovery for inflammatory diseases.

Methodology:

Uses feature representation learning with random forest classifiers applied to eight sequence encoding schemes to generate probabilistic scores per peptide, which are linearly combined to yield final PIP predictions.