Carmen Villagrasa

THE Geant4-DNA project

The Geant4-DNA project proposes to develop an open-source simulation software based and fully included in the general-purpose Geant4 Monte-Carlo simulation toolkit. The main objective of this software is to simulate biological damages induced by ionizing radiations at the cellular and sub-cellular scale. This project was originally initiated by the European Space Agency for the prediction of the deleterious effects of radiations that may affect astronauts during future long duration space exploration missions. In this paper, the Geant4-DNA collaboration presents an overview of the whole on-going project, including its most recent developments that are available in the Geant4 toolkit since December 2009 (release 9.3), as well as an illustration example simulating the direct irradiation of a biological chromatin fiber. Expected extensions involving several research domains, such as particle physics, chemistry and cellular and molecular biology, within a fully interdisciplinary activity of the Geant4 collaboration are also discussed.

Track structure modeling in liquid water: A review of the Geant4-DNA very low energy extension of the Geant4 Monte Carlo simulation toolkit

Understanding the fundamental mechanisms involved in the induction of biological damage by ionizing radiation remains a major challenge of todays radiobiology research. The Monte Carlo simulation of physical, physicochemical and chemical processes involved may provide a powerful tool for the simulation of early damage induction. The Geant4-DNA extension of the general purpose Monte Carlo Geant4 simulation toolkit aims to provide the scientific community with an open source access platform for the mechanistic simulation of such early damage. This paper presents the most recent review of the Geant4-DNA extension, as available to Geant4 users since June 2015 (release 10.2 Beta). In particular, the review includes the description of new physical models for the description of electron elastic and inelastic interactions in liquid water, as well as new examples dedicated to the simulation of physicochemical and chemical stages of water radiolysis. Several implementations of geometrical models of biological targets are presented as well, and the list of Geant4-DNA examples is described.

Molecular scale track structure simulations in liquid water using the Geant4-DNA Monte-Carlo processes

This paper presents a study of energy deposits induced by ionising particles in liquid water at the molecular scale. Particles track structures were generated using the Geant4-DNA processes of the Geant4 Monte-Carlo toolkit. These processes cover electrons (0.025 eV-1 MeV), protons (1 keV-100 MeV), hydrogen atoms (1 keV-100 MeV) and alpha particles (10 keV-40 MeV) including their different charge states. Electron ranges and lineal energies for protons were calculated in nanometric and micrometric volumes.

Physical models implemented in the GEANT4-DNA extension of the GEANT-4 toolkit for calculating initial radiation damage at the molecular level.

The ROSIRIS project aims to study the radiobiology of integrated systems for medical treatment optimisation using ionising radiations and evaluate the associated risk. In the framework of this project, one research focus is the interpretation of the initial radio-induced damage in DNA created by ionising radiation (and detected by γ-H2AX foci analysis) from the track structure of the incident particles. In order to calculate the track structure of ionising particles at a nanometric level, the Geant4 Monte Carlo toolkit was used. Geant4 (Object Oriented Programming Architecture in C++) offers a common platform, available free to all users and relatively easy to use. Nevertheless, the current low-energy threshold for electromagnetic processes in GEANT4 is set to 1 keV (250 eV using the Livermore processes), which is an unsuitable value for nanometric applications. To lower this energy threshold, the necessary interaction processes and models were identified, and the corresponding available cross sections collected from the literature. They are mostly based on the plane-wave Born approximation (first Born approximation, or FBA) for inelastic interactions and on semi-empirical models for energies where the FBA fails (at low energies). In this paper, the extensions that have been introduced into the 9.3 release of the Geant4 toolkit are described, the so-called Geant4-DNA extension, including a set of processes and models adapted in this study and permitting the simulation of electron (8 eV-1 MeV), proton (100 eV-100 MeV) and alpha particle (1 keV-10 MeV) interactions in liquid water.

Future development of biologically relevant dosimetry

Proton and ion beams are radiotherapy modalities of increasing importance and interest. Because of the different biological dose response of these radiations as compared with high-energy photon beams, the current approach of treatment prescription is based on the product of the absorbed dose to water and a biological weighting factor, but this is found to be insufficient for providing a generic method to quantify the biological outcome of radiation. It is therefore suggested to define new dosimetric quantities that allow a transparent separation of the physical processes from the biological ones. Given the complexity of the initiation and occurrence of biological processes on various time and length scales, and given that neither microdosimetry nor nanodosimetry on their own can fully describe the biological effects as a function of the distribution of energy deposition or ionization, a multiscale approach is needed to lay the foundation for the aforementioned new physical quantities relating track structure to relative biological effectiveness in proton and ion beam therapy. This article reviews the state-of-the-art microdosimetry, nanodosimetry, track structure simulations, quantification of reactive species, reference radiobiological data, cross-section data and multiscale models of biological response in the context of realizing the new quantities. It also introduces the European metrology project, Biologically Weighted Quantities in Radiotherapy, which aims to investigate the feasibility of establishing a multiscale model as the basis of the new quantities. A tentative generic expression of how the weighting of physical quantities at different length scales could be carried out is presented.

Simulation of early DNA damage after the irradiation of a fibroblast cell nucleus using Geant4-DNA

In order to improve the understanding of the mechanisms involved in the generation of early DNA damage, a new calculation chain based on the Geant4-DNA toolkit was developed. This work presents for the first time the simulation of the physical, physicochemical and chemical stages of early radiation damage at the scale of an entire human genome (fibroblast, male) and using Geant4-DNA models. The DnaFabric software was extended to generate and export this nucleus model to a text file with a specific format that can be read by Geant4 user applications. This calculation chain was used to simulate the irradiation of the nucleus by primary protons of different energies (0,5; 0,7; 0,8; 1; 1,5; 2; 3; 4; 5; 10; 20 MeV) and the results, in terms of DNA double strand breaks, agree with experimental data found in the literature (pulsed field electrophoresis technique). These results show that the simulation is consistent and that its parameters are well balanced. Among the different parameters that can be adjusted, our results demonstrate that the criterion used to select direct strand break appears to have a very significant role on the final number of simulated double strand breaks.

Geant4-DNA simulations using complex DNA geometries generated by the DnaFabric tool

Abstract Several DNA representations are used to study radio-induced complex DNA damages depending on the approach and the required level of granularity. Among all approaches, the mechanistic one requires the most resolved DNA models that can go down to atomistic DNA descriptions. The complexity of such DNA models make them hard to modify and adapt in order to take into account different biological conditions. The DnaFabric project was started to provide a tool to generate, visualise and modify such complex DNA models. In the current version of DnaFabric, the models can be exported to the Geant4 code to be used as targets in the Monte Carlo simulation. In this work, the project was used to generate two DNA fibre models corresponding to two DNA compaction levels representing the hetero and the euchromatin. The fibres were imported in a Geant4 application where computations were performed to estimate the influence of the DNA compaction on the amount of calculated DNA damage. The relative difference of the DNA damage computed in the two fibres for the same number of projectiles was found to be constant and equal to 1.3 for the considered primary particles (protons from 300 keV to 50 MeV). However, if only the tracks hitting the DNA target are taken into account, then the relative difference is more important for low energies and decreases to reach zero around 10 MeV. The computations were performed with models that contain up to 18,000 DNA nucleotide pairs. Nevertheless, DnaFabric will be extended to manipulate multi-scale models that go from the molecular to the cellular levels. Program summary Program title: DnaFabric Catalogue identifier: AEZV_v1_0 Program summary URL: http://cpc.cs.qub.ac.uk/summaries/AEZV_v1_0.html Program obtainable from: CPC Program Library, Queen’s University, Belfast, N. Ireland Licensing provisions: Apache License, 2.0 No. of lines in distributed program, including test data, etc.: 13514 No. of bytes in distributed program, including test data, etc.: 186753 Distribution format: tar.gz Programming language: C++. Computer: Computer with a GPU and OpenGL3.3 compatible drivers. Operating system: Linux (Ubuntu). RAM: 4 gigabytes Classification: 3, 14, 20. External routines: Qt5 and OpenGL3.3 Nature of problem: Simulations implying DNA geometrical models often show limitations to support the huge number of DNA constituents. In order to allow users to build, visualise and perform calculations on detailed DNA models including hundreds of thousands of DNA elements, a dedicated framework is needed. Solution method: The DnaFabric library is a framework that allows users to easily build their own DNA models, display them and perform calculations. The DnaFabric includes: hierarchically organised DNA models (binary-executable example named “Fibre”), a dedicated 3D render engine, an optimised OpenGL interface and some multi-threading facilities. Unusual features: The DnaFabric uses 3D technologies from the computer graphics world allowing the rendering of huge DNA models in real-time. Additional comments: Three examples are provided in the Examples folder. The “Basic” example describes how to set-up a simple DnaFabric user-application. The “Fibre” example shows the two DNA fibre models used for the calculations in this paper. The “MovingSpheres” example, demonstrates how to implement a simulation interacting with the DNA geometrical model. Running time: Once a user application is started, an auto-generated window will show the 3D model. The efficiency of the rendering depends highly on the user hardware. However, the user can customise each of the rendered elements contained in its application to adjust the required computer power.

Cell to Cell Variability of Radiation-Induced Foci: Relation between Observed Damage and Energy Deposition

Most studies that aim to understand the interactions between different types of photon radiation and cellular DNA assume homogeneous cell irradiation, with all cells receiving the same amount of energy. The level of DNA damage is therefore generally determined by averaging it over the entire population of exposed cells. However, evaluating the molecular consequences of a stochastic phenomenon such as energy deposition of ionizing radiation by measuring only an average effect may not be sufficient for understanding some aspects of the cellular response to this radiation. The variance among the cells associated with this average effect may also be important for the behaviour of irradiated tissue. In this study, we accurately estimated the distribution of the number of radiation-induced γH2AX foci (RIF) per cell nucleus in a large population of endothelial cells exposed to 3 macroscopic doses of gamma rays from 60Co. The number of RIF varied significantly and reproducibly from cell to cell, with its relative standard deviation ranging from 36% to 18% depending on the macroscopic dose delivered. Interestingly, this relative cell-to-cell variability increased as the dose decreased, contrary to the mean RIF count per cell. This result shows that the dose effect, in terms of the number of DNA lesions indicated by RIF is not as simple as a purely proportional relation in which relative SD is constant with dose. To analyse the origins of this observed variability, we calculated the spread of the specific energy distribution for the different target volumes and subvolumes in which RIF can be generated. Variances, standard deviations and relative standard deviations all changed similarly from dose to dose for biological and calculated microdosimetric values. This similarity is an important argument that supports the hypothesis of the conservation of the association between the number of RIF per nucleus and the specific energy per DNA molecule. This comparison allowed us to calculate a volume of 1.6 μm3 for which the spread of the specific energy distribution could explain the entire variability of RIF counts per cell in an exposed cell population. The definition of this volume may allow to use a microdosimetric quantity to predict heterogeneity in DNA damage. Moreover, this value is consistent with the order of magnitude of the volume occupied by the hydrated sugar-phosphate backbone of the DNA molecule, which is the part of the DNA molecule responsible for strand breaks.

Influence of chromatin condensation on the number of direct DSB damages induced by ions studied using a Monte Carlo code

The purpose of this work is to evaluate the influence of the chromatin condensation on the number of direct double-strand break (DSB) damages induced by ions. Two geometries of chromosome territories containing either condensed or decondensed chromatin were implemented as biological targets in the Geant4 Monte Carlo simulation code and proton and alpha irradiation was simulated using the Geant4-DNA processes. A DBSCAN algorithm was used in order to detect energy deposition clusters that could give rise to single-strand breaks or DSBs on the DNA molecule. The results of this study show an increase in the number and complexity of DNA DSBs in condensed chromatin when compared with decondensed chromatin.

Influence of the geometrical detail in the description of DNA and the scoring method of ionization clustering on nanodosimetric parameters of track structure: a Monte Carlo study using Geant4-DNA.

The aim of this study was to evaluate the influence of the geometrical detail of the DNA on nanodosimetric parameters of track structure induced by protons and alpha particles of different energies (LET values ranging from 1 to 162.5 keV µm-1) as calculated by Geant4-DNA Monte Carlo simulations.The first geometry considered consisted of a well-structured placement of a realistic description of the DNA double helix wrapped around cylindrical histones (GeomHist) forming a 18 kbp-long chromatin fiber. In the second geometry considered, the DNA was modeled as a total of 1800 ten bp-long homogeneous cylinders (2.3 nm diameter and 3.4 nm height) placed in random positions and orientations (GeomCyl). As for GeomHist, GeomCyl contained a DNA material equivalent to 18 kbp. Geant4-DNA track structure simulations were performed and ionizations were counted in the scoring volumes. For GeomCyl, clusters were defined as the number of ionizations (ν) scored in each 10 bp-long cylinder. For GeomHist, clusters of ionizations scored in the sugar-phosphate groups of the double-helix were revealed by the DBSCAN clustering algorithm according to a proximity criteria among ionizations separated by less than 10 bp. The topology of the ionization clusters formed using GeomHist and GeomCyl geometries were compared in terms of biologically relevant nanodosimetric quantities.The discontinuous modeling of the DNA for GeomCyl led to smaller cluster sizes than for GeomHist. The continuous modeling of the DNA molecule for GeomHist allowed the merging of ionization points by the DBSCAN algorithm giving rise to larger clusters, which were not detectable within the GeomCyl geometry. Mean cluster size (m1) was found to be of the order of 10% higher for GeomHist compared to GeomCyl for LET < 15 keV µm-1. For higher LETs, the difference increased with LET similarly for protons and alpha particles. Both geometries showed the same relationship between m1 and the cumulative relative frequency of clusters with v≥3 (f3) within statistical variations, independently of particle type. In order to obtain ionization cluster size distributions relevant for biological DNA lesions, the complex DNA geometry and a scoring method without fixed boundaries should be preferred to the simple cylindrical geometry with a fixed scoring volume.