iAMY-SCM

iAMY-SCM predicts amyloidogenic proteins and estimates amino acid and dipeptide propensity scores using a scoring card method for computational analysis of amyloid proteins.

Key Features:

• Scoring Card Methodology: Implements a scoring card method that applies a weighted-sum function combined with dipeptide propensity scores to identify amyloid proteins.
• Performance Metrics: Reports cross-validation accuracy of 0.895 and independent test accuracy of 0.827, with cross-validation performance ~10–22% higher than several machine learning models and independent-test performance comparable to the random forest-based RFAmy and ~9–13% better than other prevalent models.
• Simplicity and Interpretability: Uses a simple, interpretable scoring scheme based on explicit propensity values rather than complex black-box models.
• Biophysical and Biochemical Insights: Provides estimated propensity scores for amino acids and dipeptides to inform biophysical and biochemical interpretation of amyloidogenic regions.

Scientific Applications:

• Amyloidogenic Region Prediction: Predicts amyloidogenic regions within protein sequences using dipeptide and amino acid propensity scores.
• Therapeutic Target Identification: Assists in identifying candidate amyloid-related therapeutic targets by highlighting proteins with high amyloid propensity.
• Disease Mechanism Studies: Supports studies of molecular underpinnings of amyloid-associated diseases such as Alzheimer's disease and Parkinson's disease.

Methodology:

Applies a scoring card method with a weighted-sum function using dipeptide and amino acid propensity scores, with performance assessed by cross-validation and independent testing and compared to random forest-based RFAmy and other machine learning models.