Team:WashU StLouis/Model

The goal of our project is to explore radiation resistance genes to find ways to increase UV tolerance. With a changing climate, organisms grow under increasing stress of DNA damage. The overall goal of our model is to evaluate the resistance of organisms to UV and compare the growth of resistant and non-resistant cells under different levels of UV. This modeling aims to assess the commercial advantage of resistant cells compared to normal cells by predicting their growth under UV radiation. To obtain useful parameters, our models use the well-described growth and radiation death of E. coli.

Modeling Cell Death under UV

To model cell death caused by UV radiation, we made the simplifying assumption that cell death from radiation was entirely due to DNA damage. Using differential equation models for bacterial death (Kowalski et al. 2002) and DNA lesion formation and repair (Krishna et al. 2007) we could make a connection from lesion formation to cell death. The importance of this was that reducing the number of lesions by DNA repair could be linked directly to increased cell survival.

Because lesion formation is linearly related to UV intensity when there is no lesion repair, the number of lesions could be treated as a measure of UV intensity and the lowered rate of lesion formation could instead be considered as lowering UV intensity over time.

Using a differential equation model for UV bactericide, we predicted the survival of cells under a given level of UV given their DNA lesion repair rate.

UV Resistance Parameter Estimization

We wrote an algorithm to determine the parameter for UV resistance, lambda, from laboratory data. The algorithm begins by calculating a survival curve for an overestimate and an underestimate of lambda. The algorithm then takes all non-outlier values of the measured survival percentage from laboratory experiments after 1, 2, 3, and 4 hours of UV exposure, computes the square of the vertical difference between these values and the model, and sums those values for each of the two estimates of lambda to find the total error for each estimate. The algorithm then calculates the difference of the two error values. If the error of the lower estimate of lambda is lower, the algorithm is run recursively using the same lower estimate of lambda and a new upper estimate between the previous upper estimate and the lower estimate. If the error of the upper estimate of lambda is lower, the recursion uses the same upper estimate of lambda and a new lower estimate between the previous lower estimate and the upper estimate. The algorithm returns when the absolute value of the difference between the error values of the upper and lower estimates of lambda is below a specified precision.