AI-powered biotechnology platform of single-cell Raman micro-spectroscopy enables high-resolution dynamical phenotyping study of bacterial growth and cellular heterogeneity

Progresses in single-cell Raman micro-Spectroscopy (SCRS) technologies and automated measurements rapidly yield large spectroscopic datasets that require advanced artificial intelligence (AI)-powered data analytics to discover biological mechanisms in high-resolution and non-invasive manner. We developed this AI-powered SCRS biotechnology platform RamanomeSpec with 4 modules to uncover bacterial phenotypic dynamics, cellular heterogeneity, and identification using machine and deep learning algorithms.

First module RamanomeVisual was developed to visualize and quantify the separability of hierarchical structures of the 36 taxonomies via OU process and sequential ICA-UMAP algorithm. Second module RamanomeMolecule was designed to show intracellular molecular fingerprints and the dynamical fold change of key metabolites via target analysis and feature ranking. Third module RamanomeHetero was proposed to reveal cellular heterogeneity at strain level for isogenic population using PCA and spectral clustering via hard votes of 5 scoring metrics. Fourth module RamanomeDeep was developed to simultaneously achieve the identification of bacterial growth phase, taxonomy, and life cycle with accuracy above 0.95 by establishing a shallow fully connected neural network, a convolutional neural network, and a state-of-the-art Transformer architecture. This new AI-powered analytical pipeline of SCRS is also applicable to other spectroscopy methods such as NMR, LC-MS, etc.