Community benchmarking reveals low reproducibility and need for standardized protocols in mass spectrometry detection of unannotated human microproteins (nature.com)

• Reported unannotated microproteins vary 1000-fold across 12 studies; only 3.5% of peptides shared between studies.
• Manual curation of 406 PSMs shows immunopeptidomics data yield 65% high-confidence detections vs 7.8% for conventional proteomics.
• 12% of reported peptides map to annotated proteins, highlighting database issues.
• Study calls for standardized protocols and FDR controls to improve reproducibility.

"A community benchmarking study of 12 publications on mass spectrometry-based detection of human microproteins encoded by short open reading frames (sORFs) found high variability in reported numbers (range 6 to 4903) with only 3.5% of peptides replicated across studies. Manual evaluation of 406 peptide-spectrum matches revealed that immunopeptidomics studies produced higher-quality evidence (65% high confidence) than conventional proteomics studies (7.8% high confidence). The authors estimate that many reported detections may be false positives, especially in conventional datasets, and emphasize the need for standardized protocols and reporting standards to improve reproducibility and confidence in microprotein discovery."