DDAP

DDAP predicts the order of proteins and substrates and docking domain affinities in biosynthetic pathways of products synthesized by type I modular polyketide synthase (PKS) systems to support elucidation of PKS-mediated polyketide biosynthesis.

Key Features:

• Module Docking Domain Affinity Prediction: Predicts affinity between docking domains within PKS modules, reporting an Area Under Curve (AUC) of 0.88 on hold-out test datasets.
• Pathway Prediction Performance: Ranks biosynthetic pathway predictions with a reported Mean Reciprocal Rank (MRR) of 0.67.
• Probability Representation: Represents predicted docking domain affinities as probabilities between 0 and 1.
• Independence from Large Databases: Operates without dependence on large external sequence databases.
• Competitive Performance: Performs comparably to current rule-based algorithms for type I PKS pathway prediction.
• Machine Learning-Based Algorithm: Implements a machine learning–based algorithm specifically developed for type I PKS pathway prediction.

Scientific Applications:

• Natural Product Biosynthesis: Supports elucidation of polyketide biosynthetic mechanisms in studies of natural product biosynthesis.
• PKS Pathway Analysis and Engineering: Aids analysis of PKS pathway order and interactions to inform understanding and engineering of PKS-mediated polyketide synthesis.

Methodology:

Uses a machine learning approach to predict docking domain affinities and biosynthetic pathway orders and outputs affinity predictions as probabilities between 0 and 1.