Elastic net regularized regression for time-series analysis of plasma metabolome stability under sub-optimal freezing condition

Abstract

In this paper, the stability of the plasma metabolome at −20 °C for up to 30 days was evaluated using liquid chromatography-high resolution mass spectrometric metabolomics analysis. To follow the time-series deterioration of the plasma metabolome, the use of an elastic net regularized regression model for the prediction of storage time at −20 °C based on the plasma metabolomic profile, and the selection and ranking of metabolites with high temporal changes was demonstrated using the glmnet package in R. Out of 1229 (positive mode) and 1483 (negative mode) metabolite features, the elastic net model extracted 32 metabolites of interest in both positive and negative modes. L-gamma-glutamyl-L-(iso)leucine (tentative identification) was found to have the highest time-dependent change and significantly increased proportionally to the storage time of plasma at −20 °C (R2 = 0.6378 [positive mode], R2 = 0.7893 [negative mode], p-value < 0.00001). Based on the temporal profiles of the extracted metabolites by the model, results show only minimal deterioration of the plasma metabolome at −20 °C up to 1 month. However, majority of the changes appeared at around 12–15 days of storage. This allows scientists to better plan logistics and storage strategies for samples obtained from low-resource settings, where −80 °C storage is not guaranteed.

Canopy Effects on Snow Accumulation: Observations from Lidar, Canonical-View Photos, and Continuous Ground Measurements from Sensor Networks

“…where λ 1 and λ 2 are the weightings for the L1 and L2 regularizations, respectively. The Elastic Net was chosen over other regularized regression approaches for its ability to address correlated covariates and high numbers of covariates [40,41]. We have also considered nonlinear and ensemble models, such as the Random Forest regression, but the performance was not as good as the Elastic Net regression (results not shown).…”

Personalized Nutrition—Genes, Diet, and Related Interactive Parameters as Predictors of Cancer in Multiethnic Colorectal Cancer Families

“…In short, it is more likely to balance variance and bias than other methods. Unlike linear least squares, when estimating the unknown parameters in a linear regression model, GR can simply zero out certain unused predictors [ 92 , 93 , 94 , 95 ]. In this case, the p -values in the linear regression model at most could only be 0.9999, but not exactly 1.…”

In Silico Model for Chemical-Induced Chromosomal Damages Elucidates Mode of Action and Irrelevant Positives

“…Regarding positive predictions, 47 OFGs with positive correlations and 13 OFGs with negative correlations were identified. As OR is “a measure of association between exposure and outcome” [ 24 ], and although no standard errors for parameters can be calculated directly in glmnet [ 28 ], it can be expected that a higher OR is correlative to a higher contribution to results, as suggested in previous studies [ 33 , 34 ]. The OFGs identified were sorted in descending order of OR.…”