Signal Peptide Efficiency: From High-Throughput Data to Prediction and Explanation.

Area of Science:

• Biochemistry
• Molecular Biology
• Biotechnology

Background:

• Protein transport across membranes via the general secretory (Sec) pathway is vital for cell function and industry.
• Signal peptides at the N-terminus direct proteins into the Sec pathway.
• The sequence variability of signal peptides hinders understanding their impact on secretion efficiency.

Purpose of the Study:

• To identify critical physicochemical features of signal peptides that influence protein secretion efficiency.
• To develop a predictive model for signal peptide function.
• To create a tool for designing and evaluating signal peptides in silico.

Main Methods:

• Evaluation of approximately 12,000 designed signal peptides using a miniaturized high-throughput assay.
• Training a machine learning model with 156 physicochemical features per signal peptide.
• Post-hoc explanation of the machine learning model to determine feature importance.

Main Results:

• Quantification of the importance of specific physicochemical features in signal peptides.
• Accurate prediction of protein secretion levels based on signal peptide features.
• Identification of key signal peptide characteristics that enhance secretion efficiency.

Conclusions:

• Physicochemical features of signal peptides significantly impact protein secretion efficiency.
• A machine learning model can predict and explain signal peptide function.
• This work provides a valuable tool for the rational design and in silico assessment of signal peptides for biotechnological purposes.