A regression system for TOPAS-nBio

Jose Ramos-Mendez, Naoki D. Kondo, and Thongchai A. M. Masilela

Department of Radiation Oncology

University of California San Francisco.

[email protected]

May 2024.

###############################################################

Aim

Perform a regression test for TOPAS-nBio based on examples with existing published reference data.

Test

1. DBSCAN. Quantifies the ratio between SSB to DSB for protons using the clustering algorithm DBSCAN. Reference data from previous Geant4-DNA version (Francis et al., 2017) and digitized data from PARTRAC and measured is provided.

2. GvalueStepByStep. Quantifies the G-value as a function of the time for fast electrons of 1 MeV. Reference data from Wang et al., 2018 at the shortest times (7 ps) is provided.

3. LET. Quantifies restricted LET. Reference data from PARTRAC is provided.

4. NanodosimetryI. Quantifies the conditional ionization cluster size distribution (nu > 0) from carbon ions of 88 MeV incident in a single cylinder. Measured data obtained in gas from Hilgers et al., 2017 is provided.

5. NanodosimetryII. Quantifies F2 vs M1, where F2 is the cumulative probability of having ionization clusters of size 2 or bigger, and M1 is the fist moment of the unconditional ionization cluster size distribution. Measured data obtained from Conte et al. 2017 is provided.

6. NanodosimetryIII. Quantifies ionization cluster size distribution produced in randomly oriented cylinders for ions from proton to oxygen ions. Calculated reference data from Ramos-Mendez et al., 2017 is provided.

7. GValueIRT. Quantifies the G-value as a function of the time for fast electrons of 1 MeV.

8. FrickeIRT. Quantifies the G-value of Fe^3+ which comes from the oxidation of Fe^2+. This example must give a value of around 15.5 +- 0.1 reported by the ICRU.

9. GValue_LET-IRT. LET-dependent G values for e-, p and alpha at selected energies.

10. GValue_LET-SBS. LET-dependent G values for e-, p and alpha at selected energies.

11. GvalueIRT-Temperature. Temperature-dependent G values for fast electrons at T < 200C.

12. GvalueIRT_H2O2. OH-scavenger-dependent G value for H2O2 within microsecond time range.

13. GvalueIRT_H. H-scavenger-dependent G value for H within microsecond time range.

Use

Each directory has TOPAS parameters with the follow pattern:

• mainABCD.txt, where ABCD is Topas, Opt2, Opt4 or Opt6 that refer to used physics list.
• depFileN.txt, with N=1,... dependence file used by the main file.
• inputfiles.txt. Lists the main file (s) to be submitted as a job.
• tcsh script files. Each directory has three tcsh files to submit the example to e.g. a cluster system or locally. For instance, to submit 5 simulation jobs in the local system (different random seeds) use:

tcsh submitLocally.sh 5

A directory named run is created which contains a sub-directory, its name contains the current date. Later, compare between simulation results with the python script in analysis/analysis.py

python analysis/analysis.py run/2020July/mainTopas run/2020July/mainOpt2 --sut_label Topas --ref_label Opt2

Look for the images in the directory results/. A table which contains averaged execution time per CPU is also available.

Optional

If you have modified the names of the image files resulting from analysis.py, open the python script copy_and_paste.py in the directory Summary/tex_openTOPAS and change the names of the appropriate files such that they match. Then run the script. This will copy and paste all the regression test images into a new directory called Summary/openTOPAS. Run the .tex file in order to generate a PDF summarising all the results.