A comprehensive evaluation of popular proteomics software workflows for label-free proteome quantification and...

Area of Science:

• Proteomics
• Mass Spectrometry
• Bioinformatics

Background:

• Label-free mass spectrometry (MS) is crucial in life sciences for protein quantification.
• Numerous software solutions exist for processing MS data, each with unique algorithms.
• A comprehensive evaluation of overall software performance and missing value handling is lacking.

Purpose of the Study:

• To systematically evaluate five popular quantitative label-free proteomics software workflows.
• To assess the impact of missing values and imputation methods on differential expression analysis.

Main Methods:

• Performance evaluation of five label-free proteomics software using four spike-in datasets.
• Analysis of protein quantification, missing values, differential expression accuracy, and fold change.
• Comparison of various data imputation and filtering methods.

Main Results:

• Progenesis software demonstrated consistent high performance and low missing value generation.
• Missing values from other software negatively impacted performance, but could be mitigated.
• Local least squares (LLS) imputation consistently improved differential expression analysis.

Conclusions:

• Software choice significantly impacts proteomics data analysis outcomes.
• Effective data filtering and imputation strategies, particularly LLS, are vital for accurate differential expression analysis.
• Combining data filtering with LLS imputation yielded the best results across tested datasets.