SFile2_HRMSResults

This is the results file of the untargeted analysis using an in-house R package (https://github.com/kuppal2/xmsPANDA) on raw HRMS data after removing features that did not meet data quality thresholds and dropping poor quality and redundant samples. QC materials were not analyzed in the untargeted test. Columns A-B pertain to observed m/z and retention time. Columns C-E pertain to accurate mass matches to the features (Annotation, Adduct, and ppm mass accuracy error; see Supplementary Methods for details). After accurate mass matching, spectral convolutions (i.e., isotopes and neutral losses) were designated by manual inspection of co-elution, intensity correlation and m/z shift. Retention times were taken into consideration for assignments (e.g., longer retention times generally correspond to hydrophilicity and spectral convolutions of adducts and isotopes should co-elute). Note, the [M+H] ion of lidocaine (235.1804 m/z by 127.6 sec; used as local anesthetic during bronchoscopy) held the maximum intensity value in each sample and was normalized to the same maximal value in each distribution, resulting in its removal during testing. Ppm error was calculated by METLIN except for spectral convolutions, which were calculated manually. Accurate mass annotations are only provided for those results that met significance criteria (p < 0.05, q < 0.25). Columns F-H indicate the raw p-value (“P.value”), q-value (“adjusted.p.value”) and Pearson’s r statistic (“Estimate_var1”), respectively. Remaining columns show log2-transformed and quantile normalized HRMS intensities for each of the study samples. Missing values were imputed as “NA” to avoid skewing the correlation. The sample columns were automatically ordered from low-to-high PRAGMA-%Dis. [This file description may also be found in the Supplementary Information of the corresponding publication.]