ИСТИНА

Signal drift correction with batch effect removal in LC-MS based metabolomics is aimed to remove unwanted variation components, such as within- and between-batch variations, both sample and instrument sensitivity changes over time in a long injection sequence and dependency of measured values by injection order and/or batch number. “QC-LOESS” [1] regression for signal correction has become the “gold” standard over the last decade. Some improvements include the implementation of other nonlinear algorithms, i.e. random forest (RF) [2]. A complex examination of other regression models was performed [3] by various numeric criteria [4]. As a result, the new signal correction method, which outperformed other QC regression algorithms, based on gradient boosting machine (XGboost, QC-XGB) was proposed. A more precise comparison based on graphical estimation was carried out on Amide [5] metabolomics dataset for QC-LOESS, QC-RF and QCXGB methods. References 1. Dunn W.B., Broadhurst D., Begley P., Zelena E., Francis-McIntyre S., Anderson N., Brown M., Knowles J.D., Halsall A., Haselden J.N., Nicholls A.W., Wilson I.D., Kell D.B., Goodacre R. Nat. Protoc. 6 (2011), 1060–1083. 2. Luan H., Ji F., Chen Y., Cai Z. Anal. Chim. Acta. 1036 (2018), 66–72. 3. Plyushchenko I. V., Fedorova E. S., Potoldykova N. V., Polyakovskiy K. A., Glukhov A. I., Rodin I. A. J. Proteome Res. (2021), 1c00392. 4. Sánchez-Illana Á., Piñeiro-Ramos J.D., Sanjuan-Herráez J.D., Vento M., Quintás G., Kuligowski J. Anal. Chim. Acta. 1019 (2018), 38–48. 5. Deng K., Zhang F., Tan Q., Huang Y., Song W., Rong Z., Zhu Z.-J., Li K., Li Z. Anal. Chim. Acta. 1061 (2019), 60–69.