Development Of Algorithm And Software For Single Molecule/Single Particle Analysis

Recently, analyzing samples at single molecule/particle level has become a hot topic in the field of analytical chemistry, which can offer ultra-sensitive information that cannot be achieved by analyzing the bulk solution. Therefore, single molecule/particle analysis is proposed. The awarding of "2014 Nobel Prize in Chemistry" to the three scientists, who developed microscope with single molecule resolution, shows the importance of single molecule/particle analysis. As rising techniques for single molecule/particle analysis, nanopore technique and localized surface plasmon resourance technique have recived considerable attentions. However, both of the two techniques have limitations in data analysis, including:how to efficiently recognize the signal of single molecules/particles; how to accurately convert the signal into the information of single molecules/particles,In this paper, we focus on developing methods for nanopore technique and localized surface plasmon resourance technique to overcome the limitation in data analysis. We introduced new algorithms for the data analysis and developed automatic software to recognize and extract useful information in the raw data of nanopore experiment and dark-field image. Furthermore, our methods can also be used in other single molecule/particle detection techniques, which has one-dimensional data like current data in nanopore or two-dimensional data like dark-field image of localized surface plasmonic resourance. The detailed profile in this dissertation is shown as follows:1. Brief introduction of single molecule/particle analysis.In this part, we introduced the development and background of single molecule/particle analysis technologys. The development, theory and application of techniqures of nanopore and localized surface plasmon resourance were reviewed. We also showed the limitations and requirement in the data analysis of the techniques.2. Accurate Data Process in Nanaopore AnalysisNanopore is a promising tool for single molecule detection by analyzing current blockades induced by the captured analytes. However, the current noise inevitably exists during the nanopore data recording. This raises the difficulties in accurately extracting the dwell time and amplitude of each blockade from experimental data, especially for rapid translocations and bumping blockades. Herein, we present a robust data process including blockades location as well as evaluations of both dwell time and current amplitude for accurate analysis of nanopore blockades. For each recognized blockade, a 2nd-order differential based calibration (DBC) method was introduced to correct the over-coverage of blockades. This method highly eliminates the effect of random noise and calibrates the dwell time. As inspired by the nature of filter, a novel integration method is used to evaluate current amplitudes of blockades, which reduces the inaccuracy caused by rise time (Tr) of low-pass filter. Compared to conventional methods, our data process reduces the relative errors for extracting the current amplitude, especially for blockades with dwell time less than 2Tr. The processing results of both generated blockades and experimental nanopore data confirm that our data process could significantly increase the accuracy of measurements compared with the conventional method.3. In situ High Throughput Scattering Light Analysis of Single Plasmonic Nanoparticles in Living CellsPlasmonic nanoparticles have been widely applied in cell imaging, disease diagnosis, and photothermal therapy owing to their unique scattering and absorption spectra based on localized surface plasmon resonance (LSPR) property. Recently, it is still a big challenge to study the detailed scattering properties of single plasmonic nanoparticles in living cells and tissues, which have dynamic and complicated environment. The conventional approach for measuring the scattering light is based on a spectrograph coupled to dark-field microscopy (DFM), which is time-consuming and limited by the small sample capacity. Alternatively, RGB-based method is promising in high-throughput analysis of single plasmonic nanoparticles in dark-field images, but the limitation in recognition of nanoparticles hinders its application for intracellular analysis, In this paper, we developed an automatic and robust method for recognizing the plasmonic nanoparticles in dark-field image for RGB-based analysis. The method involves a bias-modified fuzzy C-means algorithm, through which biased illumination in the image could be eliminated. Thus, the data process method were testified in the condition of glass slide and in living cells. As confirmed, the distribution of peak wavelength obtained by our method is well agreed to the result measured by conventional method. Furthermore, we demonstrated that our method is profound in cell imaging studies, where its advantages in fast and high-throughput analysis of the plasmonic nanoparticles could be applied to confirm the presence and location of important biological molecules and provide efficiency information for cancer drug selection.