Room: Karl Dean Ballroom A1
Purpose: DNA damage is the origin of radiation biological effect, such as cancer. Microscopic Monte Carlo (MC) simulation is a powerful tool to accurately and quantitatively study the radiation biological effects. Conventional CPU-based microscopic MC tools suffer from low computational efficiency due to large number of events and many-body simulation problem in chemistry stage modeling. We have developed an open-source GPU-based microscopic MC package gMicroMC. This project reports our developments of a DNA geometry module in gMicroMC and corresponding algorithms to compute DNA damages.
Methods: Our DNA geometrical model contains five levels: nucleotide pair, nucleosome, chromatin fiber, chromatin fiber units and chromatin. We consider a spherical lymphocyte cell nucleus with a diameter of 10 micrometer. About 6 Gbp DNA are packed inside the nucleus. We develop computational modules to study DNA damages caused by water radiolysis in physics and chemistry stages. Once gMicroMC finishes simulation to generate damaging events, a module is launched to identify nucleotide index of each event and damage to the nucleotide. This is followed by another module to study geometry relationships among nucleotide damages and then compute DNA damages of different types. The modules employed GPU-based parallel processing to improve computational efficiency.
Results: We have successfully constructed the DNA geometrical model. In an example of irradiation with a 5 MeV proton, the computed yield of double strand break (DSB)/Gy/Gbp is 6.82, in agreement with the results presented by Meylan et al. It takes 2.7 sec to compute DNA damage of all types using a single Nvidia GTX-1080 GPU card.
Conclusion: A DNA geometrical model of lymphocyte cell nucleus together with modules to compute DNA damages have been developed. Agreement with published data in test studies demonstrated accuracy of our model. The combined accuracy and efficiency facilitates the use of gMicroMC for investigations on radiation biological effects.
Not Applicable / None Entered.
Not Applicable / None Entered.