Metadata for RadPhysBio: A Radiobiological Database for the Prediction of Cell Survival upon Exposure to Ionizing Radiation

SUPPLEMENTARY DATA RadPhysBio: a radiobiological database for the prediction of cell survival upon exposure to ionizing radiations Vassiliki Zanni1, Dimitris Papakonstantinou 2, Spyridon A Kalospyros1, Dimitris Karaoulanis3, Gökay Mehmet Biz4, Lorenzo Manti5, Adam Adamopoulos6, Athanasia Pavlopoulou4,7 and Alexandros G Georgakilas 1,* 1 DNA Damage Laboratory, Physics Department, School of Applied Mathematical and Physical Sciences, National Technical University of Athens (NTUA), Zografou Campous, 15780 Athens, Greece; fte[email protected] ; [email protected] 2 Department of Life Sciences, University Paris-Saclay, 91190 Saint-Aubin, Paris, France; [email protected] 3 School of Electrical and Computer Engineering, National Technical University of Athens, 15780 Athens, Greece; [email protected] 4 Izmir Biomedicine and Genome Center (IBG), Balcova, Izmir 35340, Turkey; [email protected] 5 National Institute of Nuclear Physics (INFN), Section of Naples, Naples Italy and Radiation Biophysics Laboratory, Department of Physics “E. Pancini”, University of Naples Federico II, Naples, Italy; [email protected] 6 Department of Medicine, Medical Physics Laboratory, Democritus University of Thrace, 681 00 Alexandroupolis, Greece; [email protected] 7 Izmir International Biomedicine and Genome Institute, Dokuz Eylül University, 35340 Balcova, Izmir, Turkey; [email protected] *Correspondence: [email protected]; Tel.: +30-210-7724453 The Python code for the calculation of α and β coefficients included in the database import numpy as np from scipy.optimize import curve_fit points = np.array([(4.0114,0.2176), (6.0114,0.11083), (8.0114,0.077946)]) xdata = points[:,0] ydata = points[:,1] def f(x,a,b): return np.exp(-a*x-b*(x**2)) popt, pcov = curve_fit(f, xdata, ydata, p0=[0.1, 1e-3]) print(popt) Example of MCDS input and output files Figure S1. Example of MCDS input file for 3.82 MeV α-particles. Figure S2. Example of MCDS output file, for 3.82 MeV α-particles. Example of WebPlotDigitizer software We load the file and check 2D Plot. After that, we choose 2 points in X-axis (X1 and X2) and 2 points in Y-axis (Y1 and Y2) in order to define their values. For example, in the image below we choose X1=5, X2=10, Y1=40 and Y2=60. In that way, the values on axes are being adjusted correctly. In the case where an axis is on a logarithmic scale, we choose the Log Scale square. Figure S3. Correct axis numbering. So, the diagram is ready. After that, we choose the experimental point which is of our interest and up in the right side of the image we can see the pair of its coordinates. We repeat this process for four-five different points, and we transfer the pair of coordinates into the code, in order to calculate α and β. For example, below we choose with red, a point around five hours and 20 γH2Ax foci per cell. Then, we can see in detail its coordinates, in the right side of the image. More specifically, the x-value is 4.0733 and the y-value is 24.653. Finally we load this pair (4.0733, 24.653) Figure S4. Selection of the appropriate point and the finding of its coordinates (Takanori Katsube, Masahiko Mori, Hideo Tsuji, Tadahiro Shiomi, Naoko Shiomi, Makoto Onoda, “Differences in sensitivity to DNA-damaging Agents between XRCC4and Artemis-deficient human cells”, 2011, Journal of Radiation Research). Table S1. X-ray repair data from literature and the k data from MATLAB calculations. #E xp ID PMID # E x p CellLi ne Radiat ionTy pe Tissue C el lC la ss Ener gy (Me V) LE T (ke V/μ m) Dose Rate (Gy/ min) DSBs/( Gy*cel l)_00.15h % DS Bs_ 1h % DS Bs_ 2h % DS Bs_ 4h % DS Bs_ 12h % DS Bs_ 24h DSBs_ 24h k of NHEJ 1 17644493 1 MRC-5 X-rays fibroblast n 0.120 2 1 42.624 75 40 21 7 1 0.426 Fibroblasts 2 17644493 2 MRC-5 X-rays fibroblast n 0.025 2 0.5 33.92 75 42 26 8 0 0 k1=27.665 k2=271.883 k3=5.278 3 20197645 1 AT1O S/T-n X-rays fibroblast n 0.150 2 1.3 13.952 90 82 70 38 33 4.604 k4=4.935 k5=3.880 k6=5.106 4 20197645 2 SuSa/T -n X-rays fibroblast n 0.150 2 1.3 19.008 90 80 61 14 4 7.603 k7=3.162 k8=1.511 k9=1.682 5 20394838 1 AG152 1 X-rays fibroblast n N/A N/ A 0.9 20.544 80 61 42 28 20 4.109 k10=0.065 6 19755805 1 HFFF2 X-rays fibroblast n 0.250 2 0.7 31.232 72 59 31 20 2 0.625 7 17169382 1 VH25 X-rays fibroblast n 0.200 2 2.8 39.104 85 59 55 41 27 10.558 8 17169382 2 FN1 X-rays fibroblast n 0.200 2 2.8 39.104 85 59 55 41 27 10.558 9 18604166 1 MAD A X-rays fibroblast n 0.250 2 0.53 41.344 80 N/ A 39 N/ A 6 2.481 10 18604166 2 K X-rays fibroblast n 0.250 2 0.53 45.312 78 N/ A 30 N/ A 0 0 11 26452569+ 1 AG152 2 X-rays fibroblast n N/A N/ A 0.591 29.312 67 56 N/ A N/ A 9 2.638 12 24505255+ 1 N/TER T-1 X-rays fibroblast n N/A N/ A 0.591 17.6 50 41 24 15 1 0.176 13 8618842+ 1 GM38 X-rays fibroblast n 0.225 2 1.5 11.008 50 41 37 35 29 3.192 14 22240940 1 He17 X-rays fibroblast n N/A N/ A 0.635 23.232 88 64 46 25 11 2.556 15 14744762 1 MRC-5 X-rays fibroblast n 0.09 2.5 2 24.896 N/ A 49 24 N/ A 5 1.245 16 14744762 2 180BR X-rays fibroblast n 0.09 2.5 2 36.928 N/ A 64 62 N/ A 28 10.340 17 14744762 3 AT1BR X-rays fibroblast n 0.09 2.5 2 22.656 N/ A 43 38 N/ A 21 4.758 18 24312182 1 AG152 2 X-rays fibroblast n N/A N/ A 0.59 20.8 98 81 53 15 8 1.664 19 18246480+ 1 1BR3 X-rays fibroblast n 0.200 2 1.234 39.616 69 41 31 11 2 0.792 20 18246480+ 2 46BR X-rays fibroblast n 0.200 2 1.234 37.248 75 52 27 11 6 2.235 21 18246480+ 3 180BR X-rays fibroblast n 0.200 2 1.234 37.76 93 84 73 50 42 15.859 22 30872656+ 1 AG015 22B X-rays fibroblast n N/A N/ A N/A 24 61 41 31 15 9 2.16 23 33692980+ 1 Dental follicle stroma l X-rays epithelial n 0.120 2 0.015 32.448 92 77 49 N/ A 17 5.516 Epithelial cells 24 25040548 1 LN18 X-rays epithelial t N/A N/ A 0.65 10.88 84 78 45 28 19 2.067 k1=65.548 k2=11.873 k3=6.629 25 25040548 2 U251 X-rays epithelial t N/A N/ A 0.65 11.84 94 75 63 21 9 1.066 k4=4.466 k5=4.633 k6=3.247 26 21785230 1 HCT11 6 X-rays epithelial t N/A N/ A 0.6 40.448 56 46 25 16 4 1.618 k7=5.304 k8=13.840 k9=5.470 27 26683123 1 MCF-7 X-rays epithelial t N/A N/ A 0.59 19.904 89 88 79 66 53 10.549 k10=0.062 28 24763056+ 1 MiaPa Ca2 X-rays epithelial t 0.200 2 1.3 21.056 83 63 46 20 7 1.474 29 21785230 1 HCT11 6 X-rays epithelial t N/A N/ A 0.6 40.96 69 58 29 18 1 0.410 30 26991853 1 H460 X-rays epithelial t 0.160 2 2.5 24.64 87 75 57 31 13 3.203 31 32397297+ 1 A549 X-rays epithelial t 0.200 2 0.85 19.328 88 81 61 21 4 0.773 32 22322361 1 PBMC s X-rays blood n 0.250 2 0.26 13.888 89 73 43 39 29 4.028 Lymphocytes 33 20597840 1 Tlymph ocytes X-rays blood n N/A N/ A 0.02 12.736 96 N/ A N/ A N/ A 39 4.967 k1=43.885 k2=152.894 k3=4.839 34 20597840 2 Tlymph ocytes X-rays blood n N/A N/ A 0.02 14.72 97 N/ A N/ A N/ A 0 0 k4=4.450 k5=3.079 k6=4.066 35 22144029 1 Lymp hocyte s X-rays blood n N/A N/ A N/A 6.336 65 55 N/ A N/ A 22 1.394 k7=2.642 k8=1.896 k9=2.053 k10=0.039 This model is formulated below by a system of 9 nonlinear ODEs: (Reza Taleei, Hooshang Nikjoo, “The Non-homologous End-Joining (NHEJ) Pathway for the Repair of DNA Double-Strand Breaks: I. A Mathematical Model”, RADIATION RESEARCH 179 (2013)) 3 12 4 1122334 5678 457 56 78 89 9 910 ,,,, ,,, dy dy dy dy adD vvvvvvv dt C dt dt dt dt dy dy dy dy vvv vv vv vv dt dt dt dt dy vv dt =⋅ − =− =− =− =−− =− =− =− =− Where 11 212 2 323334 44 55 656 6677 7588 87 99 9810 109 (1 ) , (1 ) , , , (1 ) , , (1 ) , (1 ) , (1 ) , vk cyvk cyvkyvky vk cyvKyvk cyvk cy vk cyv Ky =− = − = = =−==−=− =− = The scaling factor C is the sum of repair proteins (Ei) and repair complex concentrations (Yi): 8 88 11 3000, , j ji i ii i i ii y Y CEYcte y c CC = == =+== = =   The variables yi, vi, ki are the scaled repair complex, repair rate, and repair rate constant for i = 1 to 9, respectively. In the tables and figures below, the experimental data are the calculated repair data from Table S1, while theoretical values are the fitted ones from the NHEJ model simulation. Table S2. Experimental and theoretical values for epithelial cells. Figure S5. Fitting epithelial data with the NHEJ model. Table S3. Experimental and theoretical values for lymphocytes. Figure S6. Fitting lymphocyte data with the NHEJ model. 0 5 10 15 20 25 t (h) 0 20 40 60 80 100 Deq (Gy) DSB DNA Repair-NHEJ Model exp theor simulation 0 5 10 15 20 25 t (h) 0 20 40 60 80 100 Deq (Gy) DSB DNA Repair-NHEJ Model exp theor simulation Time (h) Deq (Gy) (Exp) Deq (Gy) (Theor) 1 82 68 2 68 51 4 47 44 12 28 27 24 14 13 Time (h) Deq (Gy) (Exp) Deq (Gy) (Theor) 1 84 64 2 64 46 4 46 42 12 28 31 24 23 20 (a) (b) (c) Figure S10. Plots of the distribution of β-values versus the distribution of their predictions, including the calculation of Spearman correlation coefficient: (a) β performance, (b) β performance (0-2 Gy-2). (a) (b) Table S5. Presentation of the values of Mean absolute error, RMSE and Spearman correlation coefficient for: α, α (0-2 Gy-1), β and β (0-2 Gy-2). Figure S11. Permutation variable importance. The effect of each variable on the Out-Of-Bag (OOB) error during model training for (a) α and (b) β. (a) (b)