Preventing Proteomics Data Tombs Through Collective Responsibility and Community Engagement

Area of Science:

• Proteomics
• Bioinformatics
• Data Science

Background:

• Public proteomics repositories contain vast mass spectrometry data.
• Much of this data is difficult to reuse, creating
• data tombs
• that hinder re-analysis.

Purpose of the Study:

• To assess the re-analyzability of publicly available mass spectrometry proteomics data.
• To identify systemic barriers hindering data reuse and reproducibility.
• To propose solutions for improving data accessibility and re-analysis.

Main Methods:

• A graduate-level course used a common R-based workflow to reanalyze six projects from the Proteomics Identification Database.
• Student teams performed identification, quantification, normalization, imputation, and differential expression analyses.
• Outcomes were compared to the original studies to identify discrepancies and barriers.

Main Results:

• Systemic barriers included lack of metadata standards, missing decoy set details, proprietary software outputs, absent spectral libraries/FASTA files, vague parameters, inconsistent file naming, and insufficient replication.
• These shortcomings led to significant discrepancies in analysis results (e.g., protein counts, differentially expressed proteins).
• Reproducibility depends on transparent metadata, open formats, and executable analysis provenance, not just instrumentation.

Conclusions:

• A minimum re-analysis package, including raw data, open formats, community standards, QC summaries, spectral libraries, and complete parameter/code sets, should be provided by data creators.
• Repositories should incentivize or mandate the submission of such comprehensive data packages.
• This approach trains students and improves community data practices to prevent proteomics
• data tombs
• .