PredictPA

PredictPA predicts protein abundance by integrating genomic, transcriptomic, and proteomic data within a Bayesian network to infer post-transcriptional and translational determinants of protein regulation.

Key Features:

• Data integration: Integrates transcriptomic measurements (e.g., mRNA levels) with proteomic observations to predict protein abundance and account for post-transcriptional regulation.
• Bayesian network model: Uses a Bayesian network for probabilistic modeling and inference under uncertainty, incorporating determinants such as mRNA levels, mRNA–protein interaction information, mRNA folding energy, mRNA half-life, and tRNA adaptation indices.
• Predictive accuracy: Reported improved identification of cell-cycle-associated proteins, recovering approximately twice as many such proteins in analyses of Saccharomyces cerevisiae and Schizosaccharomyces pombe compared with mRNA-only approaches.
• Dynamic prediction: Produces protein abundance profiles with stronger temporal dynamics than mRNA expression alone and aligns with experimental protein measurements in human cell line data.
• Molecular response modeling: Incorporates condition-specific inputs to model time-dependent molecular responses across biological processes, experimental conditions, and organisms.
• mRNA folding energy prediction: Can predict mRNA folding energy when protein abundance data are available to explore links between mRNA secondary structure and translation efficiency.

Scientific Applications:

• Cell cycle studies: Improved detection and mechanistic interpretation of proteins associated with cell-cycle progression.
• Comparative genomics: Application across species such as Saccharomyces cerevisiae and Schizosaccharomyces pombe to compare regulatory determinants of protein abundance.
• Translational efficiency research: Investigation of how codon usage proxies (tRNA adaptation indices), mRNA folding energy, and other factors relate to translation output and post-transcriptional control.

Methodology:

Bayesian network-based probabilistic modeling and inference integrating transcriptomic and proteomic measurements and incorporating determinants including mRNA levels, mRNA–protein interaction information, mRNA folding energy, mRNA half-life, and tRNA adaptation indices, with support for condition-specific, time-dependent inputs and inference of mRNA folding energy when protein abundance data are available.