Functional Analytic Methods For Cytotoxicity Data

With the continuous development of social economy and the expansion of human demand,thousands of new chemicals enter the consumer market and human living environment in variaous forms,such as drugs,chemical raw materials,food additives,and so on.However,a considerable number of new and old chemicals have not carried out the necessary,efficient,and comprehensive scientific safety assessment,which impose serious consequences on food safety and environment.The emergence of real-time cell analyzer?RTCA?based on cell electrical impedance measurement makes it possible to test and evaluate high-throughput in-vitro cytotoxicity.In this dissertation,by using the functional data analysis method combined with relevant statistical theory,the cytotoxicity data obtained from the RTCA were used to make some useful exploration.Main works of this research are summarized as follows:?1?According to the mechanism of cytotoxic absorption and death,two kinds of cytotoxic kinetic models of cell necrosis and apoptosis were constructed.Combining with brain storm optimization algorithm?BSO?,the parameter identification method of the model was given.To further improve the adaptive ability of BSO optimization algorithm,the selection and variation update mode of probability parameter was proposed.The superiority of the improved BSO algorithm and the predictive performance of the dynamic model are verified by combining the cytotoxicity data of the RTCA.On this basis,the concentration of toxic substances acting on cell lines can be estimated in real time by using the extended Kalman filter algorithm and the cytotoxicity data obtained from RTCA,which greatly advance the risk warning time.?2?Based on the characteristics of data in time series,an evaluation method for the cytotoxicity level of toxic substances was proposed.The method uses double exponential function to describe the trend of"cell growth inhibition/death"and establishes the Dose-response model between the coefficient of cell death and the concentration of chemical substances to be measured.On this basis,the death coefficientKC₅₀ was extracted as the cytotoxicity assessment index,and the RTCA cytotoxicity data of 14 chemical substances to be tested and the acute in-vivo toxicity values in the RC database of the United States verified the validity of the assessment index.?3?According to the characteristics of the dynamic response curve cluster of RTCA cytotoxicity,an algorithm for chemical screening and classification based on the cytotoxicity mode of action?MoA?was presented.The algorithm effectively utilizes the strong correlation between toxic pathway and toxicity effect of exogenous chemical disturbance and maps the cytotoxicity response curve with three-dimensional characteristics to two-dimensional space by using four-parameter model.According to the curve,the characteristic parameters reflecting the mechanism of cytotoxicity are extracted.On this basis,MoA classification of chemicals was realized by combining with the K-means clustering algorithm.The 7 experimental results of the six chemicals verified that the algorithm can achieve high-throughput screening of chemicals.?4?According to the functional characteristics of chemical cytotoxicity response,an E-Plate edge effect detection algorithm based on functional principal component analysis is proposed.Aiming at the phenomenon of E-Plate edge effect caused by temperature effect and evaporation effect,B-spline basis function is selected to fit the data of cytotoxicity response curve,and the original data is expressed as functional data for processing.Functional principal component analysis method is used to calculate the scores of principal components of the fitted data.On this basis,the rules of eliminating edge effect curves are designed.The experimental results show that this method can accurately discriminate the cytotoxic response curves with edge effect and achieve high throughput screening.