A new open-source GPU-based microscopic Monte Carlo simulation tool for the calculations of DNA damages caused by ionizing radiation — Part II: sensitivity and uncertainty analysis
Journal
Medical Physics
Journal Volume
47
Journal Issue
4
Pages
1971-1982
Date Issued
2020
Author(s)
Abstract
Purpose: Calculations of deoxyribonucleic acid (DNA) damages involve many parameters in the computation process. As these parameters are often subject to uncertainties, it is of central importance to comprehensively quantify their impacts on DNA single-strand break (SSB) and double-strand break (DSB) yields. This has been a challenging task due to the required large number of simulations and the relatively low computational efficiency using CPU-based MC packages. In this study, we present comprehensive evaluations on sensitivities and uncertainties of DNA SSB and DSB yields on 12 parameters using our GPU-based MC tool, gMicroMC. Methods: We sampled one electron at a time in a water sphere containing a human lymphocyte nucleus and transport the electrons and generated radicals until 2?Gy dose was accumulated in the nucleus. We computed DNA damages caused by electron energy deposition events in the physical stage and the hydroxyl radicals at the end of the chemical stage. We repeated the computations by varying 12 parameters: (a) physics cross section, (b) cutoff energy for electron transport, (c)–(e) three branching ratios of hydroxyl radicals in the de-excitation of excited water molecules, (f) temporal length of the chemical stage, (g)–(h) reaction radii for direct and indirect damages, (i) threshold energy defining the threshold damage model to generate a physics damage, (j)–(k) minimum and maximum energy values defining the linear-probability damage model to generate a physics damage, and (l) probability to generate a damage by a radical. We quantified sensitivity of SSB and DSB yields with respect to these parameters for cases with 1.0 and 4.5?keV electrons. We further estimated uncertainty of SSB and DSB yields caused by uncertainties of these parameters. Results: Using a threshold of 10% uncertainty as a criterion, threshold energy in the threshold damage model, maximum energy in the linear-probability damage model, and probability for a radical to generate a damage were found to cause large uncertainties in both SSB and DSB yields. The scaling factor of the cross section, cutoff energy, physics reaction radius, and minimum energy in the linear-probability damage model were found to generate large uncertainties in DSB yields. Conclusions: We identified parameters that can generate large uncertainties in the calculations of SSB and DSB yields. Our study could serve as a guidance to reduce uncertainties of parameters and hence uncertainties of the simulation results. ? 2020 American Association of Physicists in Medicine
Subjects
Bioinformatics; Computational efficiency; DNA; Electron energy levels; Electron transport properties; Graphics processing unit; Ionizing radiation; Monte Carlo methods; Parameter estimation; Comprehensive evaluation; Computation process; Double strand breaks; Electron transport; Hydroxyl radicals; Identified parameter; Sensitivity and uncertainty analysis; Single-strand breaks; Uncertainty analysis; DNA; hydroxyl radical; proton; sugar phosphate; Article; cell nucleus; chemical reaction; DNA damage; electron transport; excitation; human; human cell; ionizing radiation; linear energy transfer; lymphocyte; Monte Carlo method; physical chemistry; radiation dose; sensitivity analysis; single stranded DNA break; uncertainty; computer graphics; cytology; radiation response; uncertainty; Computer Graphics; DNA Damage; Lymphocytes; Monte Carlo Method; Radiation, Ionizing; Uncertainty
Type
journal article