Statistical modelling in metabolomic research for the identification and isolation of bio active compounds

The need for novel anti-cancer compounds has led to vast screenings of metabolites from marine and terrestrial organisms for their cytotoxic activity. Coupling of bio-active screenings with non-targeted metabolomics approaches using gas and liquid mass spectrometry might assist in identifying compounds of interest that are responsible for the observed biological effect. Yet, such methodology usually results in large data sets, unsuitable for manual exploring. Furthermore, the identification and isolation of an unknown active compound is a costly and time-consuming process, usually involving untargeted separation guided by bioassay models, which might result in the isolation of an already known compound. the use of statistical modeling might help to pinpoint a compound of interest and to avoid such unfruitful efforts.

Under this work frame, methanolic extracts of 38 genera of East Mediterranean seaweeds were screened annually for their anti-cancer activity on lung (A549), prostate (PC3), and colon (HT29) carcinoma. We analyzed each extract by means of liquid mass spectrometry and analyzed the metabolomic profile through the Global Natural Product Social platform (GNPS) workflow. We then combined both PLS-DA and Pearson correlation statistical models to highlight compounds of interest and narrow down the search for a compound with an anti-cancer effect. These compounds are now the main interest for further investigation.

By Integrating statistical models with bio guided assay, we aim to provide tools for more efficient, money, and time saving method for isolation of bioactive compound of interest, which can be appley for any bio guided assay process in drug research.