All Times EDT
Keywords: time series,classification,multivariate time series,cell cultures,biophysics
We explore the problem of identifying mammalian cells in culture that have been infected with mycoplasma organisms, using ECIS data. ECIS (Electric Cell-substrate Impedance Sensing) measures the electrical impedance of a cell culture when an alternating current is passed through it. By repeating this measurement over time and at different AC frequencies, we can obtain a multivariate time series reflecting the growth and confluence behavior of the cell culture, without the need for optical or DNA measurement. These time series show distinctive characteristics depending on the cell line used and the presence of contaminants. We show that simple feature-based analysis methods can distinguish between infected and uninfected cells, while more sophisticated techniques are required to make the results more robust to variation in laboratory conditions.