Finding Potentially Erroneous Entries in METLIN SMRT

Background: METLIN SMRT is a widely-used dataset of retention times for high-performance liquid chromatography (HPLC). Besides direct application it is used for training models aimed at predicting retention times in HPLC. Although there are quite a number of articles featuring METLIN SMRT, the pipelines used for filtering from errors are either simplistic or nonexistent. Two more datasets of HPLC retention times - RepoRT and MCMRT - have emerged recently. RepoRT maintainers used a 10-fold cross-validation strategy with gradient boosting models to validate retention times in the database. MCMRT maintainers suggested a projection method for transferring retention times from one chromatographic method to another, but there is no information about applying the method for data validation. Therefore, a reliable method for filtering potentially erroneous entries is still required. Results: An approach to filter potentially erroneous entries, as suggested in our earlier work for a database of gas chromatography retention indexes, was repurposed for METLIN SMRT using five predictive models (GNN, CNN, FCFP, FCD, and CatBoost). The retention times were predicted for the whole dataset using a 5-fold cross-validation strategy. Entries with retention times differing significantly from the predictions (bottom 5%) were flagged with a “yellow card”. This procedure was repeated for each model, leading to obtaining a group containing 1544 entries (about 2% of the dataset) with 5 “yellow cards”. These entries were considered potentially erroneous, as anomalous behavior was observed in the analyzed trends (with the increasing number of “yellow cards”) for both the size of each group and the standard deviation of the predictions. Significance: The previously proposed filtering approach was expanded to a retention time database, enabling finding potentially erroneous entries in METLIN SMRT. This work demonstrates the viability of the approach and its potential to improve the quality of other large-scale chromatography-related databases both for machine learning and experimental use.