Integrating shotgun proteomics and mRNA expression data to improve protein identification

Area of Science:

• Proteomics
• Bioinformatics
• Computational Biology

Background:

• Tandem mass spectrometry (MS/MS) is crucial for analyzing protein mixtures but has limited sensitivity for protein identification.
• Shotgun proteomics typically assumes equal protein detectability, overlooking valuable prior information like mRNA abundance.

Purpose of the Study:

• To develop a Bayesian scoring method integrating mRNA concentration data to enhance protein identification in MS/MS experiments.
• To improve the sensitivity and accuracy of protein identification in shotgun proteomics.

Main Methods:

• Developed a Bayesian score estimating the posterior probability of protein presence using MS/MS identification and mRNA concentration.
• Implemented the method as MSpresso software.

Main Results:

• MSpresso substantially increases the number of identified proteins at a given error rate, with up to a 40% increase in yeast.
• The method improved protein identification by 19-63% across diverse datasets, including different organisms (E. coli, human) and MS/MS instruments.
• Demonstrated that incorporating prior knowledge significantly boosts protein identification scores in shotgun proteomics.

Conclusions:

• MSpresso effectively leverages prior biological information (mRNA levels) to overcome MS/MS sensitivity limitations.
• The approach offers a significant advancement for protein identification in complex biological samples, applicable across various experimental conditions.