ИСТИНА

In plant analysis omics approaches help to solve a wide range of different problems – from botanic classification to environment pollution studies. Genomic and metabolomic approaches were main instruments in plant studies for many years. In spite of the fact that DNA based methods seem to be most suitable for plant classification tasks, the resulting phenotype of a species could vary because of different factors, such as different gene expression influence on phenotype and specifics of plant genome. In this situation, an untargeted metabolomics approach seems convenient for confirmation of other types of classification and cross-omics studies, since it does not require difficult conditions of sample preparation and interpretation of raw data. In our study, we provide workflow of brewing Humulus lupulus cultivars classification according to previously obtained EST-SSR based results [1] by LC-HRMS profiling, including feature selection and identification of significant metabolites. 18 cultivars of hops (in pellets) were analyzed, each in 3 biological repeats (54 samples in total). Randomization of the experimental sequence and quality control (QC) samples were used to reduce influence of batch effect and systematic error. Sample preparation consisted of ultrasonic extraction of homogenised pellets using methanol. Mixture of all samples were used for QC samples preparation. LC-HRMS profiling was provided via Shimadzu LCMS-IT-TOF system, equipped with electrospray ionization ion source (ESI) and reversed-phase column Thermo Acclaim RSLC C18. Peak picking and chromatogram alignment were performed using xcms R package with parameters were optimized by IPO package. Random forest based missing value imputation and XGboost based signal correction were performed for the peaks table. abels, obtained in reference genomic study, were fine-tuned by hierarchical cluster analysis based on principal components coordinates. Marker compounds were selected using rational cut off according to three tests: moderated t-test, fold change and PLS-DA algorithm. Collision induced dissociation fragmentation mass-spectra were than obtained for selected biomarkers and processed by various annotation tools, including MetFrag, CFM-ID and McSearch. References 1. Josef Patzak, Alena Henychová. Czech Journal of Genetics and Plant Breeding 54.2 (2018): 86-91