Universal Biosensor for High-Accuracy Cancer Cell Classification and Clustering

Fluorescence biosensors have widespread application in high-throughput and high-content screening technologies which are widely used in cancer research for identifying potential drug candidates and disease detection. However, most biosensors are specifically designed to recognize a single target, so selection of biomarkers requires prior-knowledge about the target and identification results are affected by prior knowledge. As a trade-off between accuracy and computational complexity, for most current biosensors, only one variable such as the intensity of a single wavelength or the average of certain wavelengths from the emission spectrum is collected and analyzed, and the information contained in the emission shape cannot be properly utilized. University of Texas at San Antonio in collaboration with University of Texas Health San Antonio and University of Florida researchers have conjugated P-C-3, a broad-spectrum polymeric biosensor which is able to provide high content multi-channel data without adding multiple biosensors, and a methodology for high-throughput fluorescence spectral shape analysis. P-C-3 is able to classify nucleoside phosphates in one minute and achieves 100% classification accuracy. By mixing P-C-3 solution with different nucleoside phosphates on 96-well plates, the fluorescence spectrum of each solution is collected by the plate reader. By analyzing the fluorescence spectral shapes, we demonstrated that fluorescence spectral shapes of P-C-3 are sensitive to both the charge and the structure of analytes. An algorithm was also developed to select useful features from the fluorescence spectrum to reduce computational complexity and prevent overfitting. This is the first classification method to utilize the information contained in fluorescence spectrum shapes of a polymeric biomarker. With the feature selection algorithm, measuring the normalized intensity of a few selected wavelengths instead of the whole fluorescence spectrum drastically reduces the measurement time, which reveals the potential of fluorescence spectrum shape analysis in high-throughput screening. By analyzing high-dimensional spectral data, developing a universal biomarker for high-accuracy classification and clustering of cell type and phenotype becomes possible. University of Texas at San Antonio andUniversity of Texas Health San Antonio researchers have conjugated P-C-3, abroad-spectrum polymeric biosensor which is able to provide high contentmulti-channel data without adding multiple biosensors, and a methodology forhigh-throughput fluorescence spectral shape analysis. Robert Graham robert.graham@utsa.edu (210) 458-8139