Optimization of pre-processing and data fusion strategies for multi-block spectroscopic characterization of cellular growth phases in the chlorophyte, Tetraselmis suecica

In an era where phytoplankton-based technologies are expanding, optimization for rapid cellular growth detection and detailed biochemical characterization is key to enhancing productivity. Hence, the complex analysis and development of highly sensitive and rapid in-vivo methods for precise real-time molecular monitoring of phytoplankton cells is imperative. Herein, a multi-block spectroscopic based methodology for the identification and characterization of molecular changes occurring across different growth phases in the chlorophyte species, Tetraselmis suecica, is presented. Confocal Raman microscopy with near-infrared (NIR) excitation spectra and concurrent excitation-emission fluorescence matrix (EEM) measurements were taken at different intervals across a twenty-day cell growth cycle period encompassing three distinct growth phases (exponential, stationary and decline). Three different data fusion strategies were explored: low-level, midlevel and a mixed level with subsequent multi-block model building using the Common Components and Specific Weights Analysis (CCSWA) algorithm. Various pre-processing sequences in regard to the raw data and single-block exploratory methods were evaluated for all three strategies and selected based on the optimum computed salience contribution towards the multi-block global components within each model. A detailed characterization of the biochemical changes happening within cellular growth phases of the chlorophyte species was constructed. Additionally, the study establishes a novel paradigm for data manipulation within a multi-block framework for complex spectroscopic data of biological cells, such as phytoplankton.