Systematic pre-annotation explains the “dark matter” in LC-MS metabolomics

The majority of features in global metabolomics from high-resolution mass spectrometry are typically not identified, referred as the “dark matter”. Are these features real compounds or junk? Understanding this problem is critical to the annotation and interpretation of metabolomics data and future development of the field. Recent debates also brought attention to in-source fragments, which appear to be prevalent in spectral databases. We report here a systematic analysis of 61 representative public datasets from LC-MS metabolomics, the most common data type in biomedical studies. The results indicate that in-source fragments contribute to less than 10% of features in LC-MS metabolomics. Khipu-based pre-annotation shows that majority of abundant features have identifiable ion patterns. This suggests that the “dark matter” in LC-MS metabolomics is explainable in an abundance dependent manner; most features are from real compounds; the number of compounds is much smaller than that of features; most compounds are yet to be identified.