Molecular Descriptors Family on Structure Activity Relationships 6. Octanol-Water Partition Coefficient of Polychlorinated Biphenyls
Lorentz Jäntschi and Sorana Bolboacă
Technical University of Cluj-Napoca, Romania, http://lori.academicdirect.org
“Iuliu Haţieganu” University of Medicine and Pharmacy, http://sorana.academicdirect.ro
Abstract
Octanol-water partition coefficient of two hundred and six polychlorinated biphenyls was model by the use of an original method based on complex information obtained from compounds structure. The regression analysis shows that best results are obtained in four-varied model (r2 = 0.9168). The prediction ability of the model was studied through leave-one-out analysis (r2cv(loo) = 0.9093) and in training and test sets analysis. Modeling the octanol-water partition coefficient of polychlorinated biphenyls by integration of complex structural information provide a stable and performing four-varied model, allowing us to make remarks about relationship between structure of polychlorinated biphenyls and associated octanol-water partition coefficients.
Keywords
PolyChlorinated Biphenyls (PCBs), Molecular Descriptors Family (MDF), Structure-Property Relationships (SAR), Octanol-water partition coefficient
Background
Polychlorinated biphenyls (PCBs), stable organic industrials chemicals widely used as insulating fluids, hydraulic and lubricating fluids, heat exchanger fluids and as additives in adhesive inks and paints [1] are persistent in the environment [2] as well as in the living tissue [3].
Quantitative structure-property relationships of PCBs were previous studied taking into consideration octanol-water partition coefficients and soil-water partition coefficients [4] and/or other physicochemical properties [5].
Based on the complex information offered by the structure of polychlorinated biphenyls congeners, octanol-water partition coefficients express as log Kow was modeled by applying of an original methodology. Thus, the aim of the paper is to present the performances of the original methodology in estimation and prediction of octanol-water partition coefficients of polychlorinated biphenyls.
Materials and Methods
A set of two-hundred and six polychlorinated biphenyls congeners with measured octanol-water partition coefficients were included into analysis. The values for the octanol-water partition coefficients were take from a previous reported study [6]. There were included ten PCBs congener group: mono-, di-, tri-, terta-, penta-, hexa-, hepta-, octa-, nona-, decachlorobiphenyl. Table 1 contains the PBCs number, the structure (chlorine-filled) and associated octanol-water partition coefficients (express as logKow).
The original methodology is based on molecular descriptors family computed based on the structure of the PCBs. The steps used to model the activity of interest were presented in details on [7] and were:
· Step 1: Sketch of the three-dimensional structure of polychlorinated biphenyls congeners;
· Step 2: Create the file with the measured octanol-water partition coefficients of the polychlorinated biphenyls congeners;
· Step 3: Generating, computing and filtering the members of molecular descriptors family for polychlorinated biphenyls congeners;
· Step 4: Finding and identifying the SAR models for polychlorinated biphenyls congeners;
· Step 5: Validate the SAR model by a cross-validation analysis [8];
· Step 6: Analyze the selected SAR model.
Results and Discussions
Modeling of the octanol-water partition coefficients of the polychlorinated biphenyls congeners was run on mono-, bi-, and tetra-varied SARs. The model which obtained best performance was the four-varied model and is presented here. The equation of the four varied model is:
ŶlogKow = 3.039 - 0.421·IIDDKGg + 0.044·IHDRKEg + 0.070·aHMmjQti - 37.502·aSMMjQg
The abbreviation associated with the studied PCBs congener (PBC no.), the measured octanol-water partition coefficients (express as logKow), the values of the descriptors used and estimated octanol-water partition coefficients by the model (ŶlogKow) and the absolute differences between estimated by the model and measured octanol-water partition coefficients (|Ŷ-logKow|) are in table 1.
Table 1. Polychlorinated biphenyls abbreviation, logKow, values for descriptors used by model,
ŶlogKow, and |ŶlogKow - logKow|
PCB no. |
Structure |
logKow |
IIDDKGg |
IHDRKEg |
aHMmjQt |
aSMMjQg |
ŶlogKow |
|Ŷ-logKow| |
1 |
2 |
4.6010 |
5.7503 |
91.1540 |
0.0244 |
3.67·10-5 |
4.6477 |
0.0467 |
2 |
3 |
4.4210 |
6.8329 |
100.870 |
0.0286 |
5.60·10-4 |
4.6022 |
0.1812 |
3 |
4 |
4.4010 |
7.1099 |
105.020 |
0.0303 |
1.50·10-4 |
4.6845 |
0.2835 |
4 |
2,2' |
5.0230 |
5.6688 |
98.0270 |
0.0454 |
2.17·10-4 |
4.9804 |
0.0426 |
5 |
2,3' |
5.0210 |
6.0092 |
104.370 |
0.1765 |
4.61·10-5 |
5.1330 |
0.1120 |
6 |
2,4 |
5.1500 |
7.0663 |
113.130 |
0.0205 |
4.01·10-5 |
5.0646 |
0.0854 |
7 |
2,4' |
5.3010 |
7.2970 |
115.000 |
0.0079 |
8.57·10-5 |
5.0476 |
0.2534 |
8 |
2,5 |
5.1800 |
5.8788 |
102.720 |
0.1013 |
4.82·10-5 |
5.1096 |
0.0704 |
9 |
2,6 |
5.3110 |
5.3684 |
95.9580 |
0.1265 |
4.87·10-5 |
5.0274 |
0.2836 |
10 |
3,3' |
5.3430 |
6.8183 |
115.510 |
0.0464 |
2.30·10-4 |
5.2688 |
0.0742 |
11 |
3,4 |
5.2950 |
7.2304 |
118.150 |
0.0067 |
4.30·10-5 |
5.2163 |
0.0787 |
12 |
3,5 |
5.4040 |
6.7261 |
115.520 |
0.0357 |
5.34·10-4 |
5.2959 |
0.1081 |
13 |
4,4' |
5.3350 |
7.3646 |
124.560 |
0.0065 |
1.00·10-4 |
5.4409 |
0.1059 |
14 |
2,2',3 |
5.3110 |
6.5110 |
114.030 |
0.0668 |
2.44·10-4 |
5.3336 |
0.0226 |
15 |
2,2',4 |
5.7610 |
6.9032 |
119.960 |
0.0867 |
2.52·10-4 |
5.4317 |
0.3293 |
16 |
2,2',5 |
5.5510 |
6.8266 |
120.050 |
0.0579 |
2.66·10-4 |
5.4654 |
0.0856 |
17 |
2,2',6 |
5.4810 |
5.8973 |
106.490 |
0.0865 |
3.80·10-4 |
5.2550 |
0.2260 |
18 |
2,3,3' |
5.5770 |
7.8504 |
128.320 |
0.7667 |
8.00·10-5 |
5.4564 |
0.1206 |
19 |
2,3,4 |
5.5170 |
7.4010 |
125.250 |
0.0235 |
5.28·10-5 |
5.4591 |
0.0579 |
20 |
2,3,4' |
5.4210 |
8.1564 |
132.790 |
0.0508 |
7.75·10-5 |
5.4753 |
0.0543 |
21 |
2,3,5 |
5.5770 |
6.5446 |
122.510 |
0.7310 |
1.40·10-4 |
5.7444 |
0.1674 |
22 |
2,3,6 |
5.6710 |
6.2218 |
110.890 |
0.0325 |
1.01·10-4 |
5.3195 |
0.3515 |
23 |
2,3',4 |
5.6770 |
8.1210 |
134.330 |
0.0325 |
5.58·10-5 |
5.5578 |
0.1192 |
24 |
2,3',5 |
5.6670 |
7.3770 |
129.310 |
0.1674 |
8.11·10-5 |
5.6575 |
0.0095 |
25 |
2,3',6 |
5.4470 |
6.2046 |
113.400 |
0.0385 |
1.01·10-4 |
5.4381 |
0.0089 |
26 |
2,4,4' |
5.6910 |
8.4470 |
139.210 |
0.0266 |
6.66·10-5 |
5.6355 |
0.0555 |
27 |
2,4,5 |
5.7430 |
7.1079 |
126.640 |
0.0245 |
5.28·10-5 |
5.6439 |
0.0991 |
28 |
2,4,6 |
5.5040 |
6.5149 |
114.430 |
0.0420 |
1.50·10-4 |
5.3515 |
0.1525 |
29 |
2,4',5 |
5.6770 |
7.5935 |
133.180 |
0.0270 |
7.78·10-5 |
5.7278 |
0.0508 |
30 |
2,4',6 |
5.7510 |
6.4853 |
116.740 |
0.0429 |
1.60·10-4 |
5.4657 |
0.2853 |
31 |
2',3,4 |
5.5720 |
7.7422 |
129.640 |
0.0889 |
7.82·10-5 |
5.5131 |
0.0589 |
32 |
2',3,5 |
5.6670 |
7.2479 |
126.270 |
0.0128 |
7.99·10-5 |
5.5668 |
0.1002 |
33 |
3,3',4 |
5.8270 |
8.6854 |
141.730 |
0.1163 |
3.69·10-4 |
5.6414 |
0.1856 |
34 |
3,3',5 |
4.1510 |
8.1095 |
137.500 |
0.3422 |
3.53·10-2 |
4.4021 |
0.2511 |
35 |
3,4,4' |
4.9410 |
8.9668 |
146.520 |
0.3565 |
1.86·10-4 |
5.7583 |
0.8173 |
36 |
3,4,5 |
5.7670 |
7.0857 |
129.590 |
0.0297 |
1.89·10-3 |
5.7151 |
0.0519 |
37 |
3,4',5 |
5.8970 |
8.3823 |
142.120 |
0.0550 |
3.49·10-4 |
5.7827 |
0.1143 |
38 |
2,2',3,3' |
5.5610 |
7.7406 |
134.940 |
0.1340 |
2.94·10-4 |
5.7430 |
0.1820 |
39 |
2,2',3,4 |
6.1110 |
7.8567 |
137.950 |
0.0258 |
2.88·10-4 |
5.8198 |
0.2912 |
40 |
2,2',3,4' |
5.7670 |
8.0115 |
139.920 |
0.1247 |
2.83·10-4 |
5.8488 |
0.0818 |
41 |
2,2',3,5 |
5.7570 |
7.2531 |
133.770 |
0.0875 |
2.68·10-4 |
5.8942 |
0.1372 |
42 |
2,2',3,5' |
5.8110 |
7.5381 |
135.540 |
0.0446 |
3.09·10-4 |
5.8479 |
0.0369 |
43 |
2,2',3,6 |
5.5370 |
6.5064 |
121.120 |
0.1066 |
3.47·10-4 |
5.6478 |
0.1108 |
44 |
2,2',3,6' |
5.5370 |
6.6121 |
121.520 |
0.1112 |
4.71·10-4 |
5.6166 |
0.0796 |
45 |
2,2',4,4' |
6.2910 |
8.1688 |
145.660 |
0.2853 |
3.00·10-4 |
6.0468 |
0.2442 |
46 |
2,2'4,5 |
5.7870 |
7.2785 |
136.050 |
0.0489 |
2.52·10-4 |
5.9821 |
0.1951 |
47 |
2,2',4,5' |
6.2210 |
7.6295 |
141.160 |
0.1152 |
2.59·10-4 |
6.0646 |
0.1564 |
48 |
2,2',4,6 |
5.6370 |
6.6210 |
124.040 |
1.3559 |
4.02·10-4 |
5.8136 |
0.1766 |
49 |
2,2',4,6' |
5.6370 |
6.7699 |
126.190 |
0.0853 |
3.62·10-4 |
5.7589 |
0.1219 |
50 |
2,2',5,5' |
6.0910 |
6.6586 |
135.000 |
0.0994 |
2.62·10-4 |
6.1997 |
0.1087 |
51 |
2,2',5,6' |
5.6270 |
6.1810 |
121.830 |
0.0035 |
9.77·10-6 |
5.8215 |
0.1945 |
52 |
2,2',6,6' |
5.9040 |
5.3274 |
109.030 |
0.1105 |
4.85·10-4 |
5.6047 |
0.2993 |
53 |
2,3,3',4 |
6.1170 |
8.3206 |
146.050 |
0.0345 |
4.93·10-5 |
5.9921 |
0.1249 |
54 |
2,3,3',4' |
6.1170 |
8.3589 |
146.460 |
0.0962 |
5.52·10-5 |
5.9982 |
0.1188 |
55 |
2,3,3',5 |
6.1770 |
7.8097 |
142.860 |
0.8606 |
7.89·10-5 |
6.1226 |
0.0544 |
56 |
2,3,3',5' |
6.1770 |
7.9437 |
142.850 |
0.0479 |
5.54·10-5 |
6.0100 |
0.1670 |
57 |
2,3,3',6 |
5.9570 |
6.9758 |
129.260 |
0.0476 |
9.87·10-5 |
5.8151 |
0.1419 |
58 |
2,3,4,4' |
5.4520 |
8.2526 |
145.460 |
0.0345 |
4.82·10-5 |
5.9947 |
0.5427 |
59 |
2,3,4,5 |
5.9430 |
6.7297 |
133.750 |
0.0286 |
3.72·10-5 |
6.1181 |
0.1751 |
60 |
2,3,4,6 |
5.8970 |
6.2938 |
123.830 |
0.0630 |
1.40·10-4 |
5.8617 |
0.0353 |
61 |
2,3,4',5 |
6.1770 |
8.0492 |
147.320 |
0.0969 |
6.32·10-5 |
6.1662 |
0.0108 |
62 |
2,3,4',6 |
5.9570 |
7.2049 |
133.110 |
0.0597 |
1.63·10-4 |
5.8873 |
0.0697 |
63 |
2,3,5,6 |
5.8670 |
5.6388 |
122.010 |
0.1451 |
9.11·10-5 |
6.0644 |
0.1974 |
64 |
2,3',4,4' |
5.4520 |
8.6051 |
153.380 |
0.0353 |
5.40·10-5 |
6.1961 |
0.7441 |
65 |
2,3',4,5 |
6.2070 |
8.1728 |
148.280 |
0.0359 |
5.11·10-5 |
6.1529 |
0.0541 |
66 |
2,3',4,5' |
6.2670 |
8.1003 |
148.960 |
0.0334 |
5.87·10-5 |
6.2129 |
0.0541 |
67 |
2,3',4,6 |
6.0470 |
7.2353 |
133.270 |
0.0631 |
1.66·10-4 |
5.8817 |
0.1653 |
68 |
2,3',4',5 |
6.2310 |
7.5977 |
146.340 |
0.1246 |
5.56·10-5 |
6.3151 |
0.0841 |
69 |
2,3',4',6 |
5.9870 |
6.4091 |
128.760 |
0.0684 |
1.26·10-4 |
6.0319 |
0.0449 |
70 |
2,3',5,5' |
6.2670 |
7.2087 |
143.450 |
0.0517 |
5.87·10-5 |
6.3459 |
0.0789 |
71 |
2,3',5',6 |
6.0470 |
6.0292 |
126.110 |
0.0503 |
1.18·10-4 |
6.0737 |
0.0267 |
72 |
2,4,4',5 |
6.6710 |
8.2735 |
152.290 |
0.0305 |
5.04·10-5 |
6.2873 |
0.3837 |
73 |
2,4,4',6 |
6.0570 |
7.4294 |
137.080 |
0.1203 |
4.40·10-4 |
5.9621 |
0.0949 |
74 |
2',3,4,5 |
6.1370 |
7.3073 |
139.090 |
0.0521 |
5.68·10-5 |
6.1119 |
0.0251 |
75 |
3,3',4,4' |
6.5230 |
9.1490 |
161.430 |
0.2535 |
9.76·10-5 |
6.3366 |
0.1864 |
76 |
3,3',4,5 |
6.3570 |
8.1041 |
151.240 |
1.4919 |
1.77·10-4 |
6.4093 |
0.0523 |
77 |
3,3',4,5' |
6.4270 |
8.5725 |
156.830 |
0.5807 |
1.28·10-4 |
6.3975 |
0.0295 |
78 |
3,3',5,5' |
6.5830 |
7.9892 |
152.370 |
0.3196 |
2.56·10-4 |
6.4229 |
0.1601 |
79 |
3,4,4',5 |
6.3670 |
8.5262 |
157.620 |
0.1010 |
1.19·10-4 |
6.4188 |
0.0518 |
80 |
2,2',3,3',4 |
6.1420 |
8.2976 |
153.620 |
0.4084 |
3.31·10-4 |
6.3517 |
0.2097 |
81 |
2,2',3,3',5 |
6.2670 |
7.6890 |
148.730 |
0.7941 |
3.69·10-4 |
6.4172 |
0.1502 |
82 |
2,2',3,3',6 |
6.0410 |
6.9185 |
136.000 |
0.0442 |
3.88·10-4 |
6.1260 |
0.0850 |
83 |
2,2',3,4,4' |
6.6110 |
8.4996 |
159.160 |
0.2075 |
3.32·10-4 |
6.4975 |
0.1135 |
84 |
2,2',3,4,5 |
6.2040 |
7.2394 |
145.350 |
0.1531 |
2.73·10-4 |
6.4160 |
0.2120 |
85 |
2,2',3,4,5' |
6.3710 |
7.9248 |
154.060 |
0.0658 |
3.42·10-4 |
6.5038 |
0.1328 |
86 |
2,2',3,4,6 |
7.5160 |
6.7474 |
135.070 |
19.1930 |
3.56·10-4 |
7.4926 |
0.0234 |
87 |
2,2',3,4,6' |
6.0770 |
6.8822 |
137.480 |
0.1178 |
8.98·10-4 |
6.1927 |
0.1157 |
88 |
2,2',3,4',5 |
6.3670 |
7.9310 |
154.600 |
0.1001 |
2.65·10-4 |
6.5303 |
0.1633 |
89 |
2,2',3,4',6 |
6.1370 |
7.1280 |
140.750 |
0.1762 |
3.36·10-4 |
6.2589 |
0.1219 |
90 |
2,2',3,5,5' |
6.3570 |
7.3803 |
149.770 |
0.4452 |
3.11·10-4 |
6.5710 |
0.2140 |
91 |
2,2',3,5,6 |
6.0470 |
6.3200 |
132.670 |
0.5266 |
3.58·10-4 |
6.2654 |
0.2184 |
92 |
2,2',3,5,6' |
6.1370 |
6.4532 |
134.510 |
0.0738 |
1.69·10-4 |
6.2662 |
0.1292 |
93 |
2,2',3,5',6 |
6.1370 |
6.6477 |
136.300 |
0.4964 |
7.70·10-4 |
6.2704 |
0.1334 |
94 |
2,2',3,6,6 |
5.7170 |
5.9325 |
123.370 |
0.0920 |
3.78·10-4 |
5.9865 |
0.2695 |
95 |
2,2',3',4,5 |
6.6710 |
7.8560 |
152.750 |
0.1031 |
3.34·10-4 |
6.4778 |
0.1932 |
96 |
2,2',3',4,6 |
6.1370 |
7.0705 |
139.710 |
0.1352 |
7.49·10-4 |
6.2187 |
0.0817 |
97 |
2,2',4,4',5 |
7.2110 |
8.1412 |
159.410 |
0.0577 |
2.85·10-4 |
6.6507 |
0.5603 |
98 |
2,2',4,4',6 |
6.2370 |
7.3053 |
144.740 |
0.2006 |
5.63·10-4 |
6.3537 |
0.1167 |
99 |
2,2',4,5,5' |
7.0710 |
7.6259 |
154.880 |
0.1225 |
3.04·10-4 |
6.6712 |
0.3998 |
100 |
2,2',4,5,6' |
6.1670 |
6.7261 |
138.380 |
0.0262 |
2.22·10-5 |
6.3246 |
0.1576 |
101 |
2,2',4,5',6 |
6.2270 |
6.8046 |
140.120 |
0.0279 |
5.27·10-5 |
6.3674 |
0.1404 |
102 |
2,2',4,6,6 |
5.8170 |
6.0829 |
126.540 |
0.0922 |
3.76·10-4 |
6.0634 |
0.2464 |
103 |
2,3,3',4,4' |
6.6570 |
9.1546 |
168.730 |
0.0365 |
4.78·10-5 |
6.6435 |
0.0135 |
104 |
2,3,3',4,5 |
6.6470 |
8.2227 |
158.650 |
0.0384 |
3.80·10-5 |
6.5907 |
0.0563 |
105 |
2,3,3',4',5 |
6.7170 |
8.5882 |
164.350 |
0.0579 |
5.64·10-5 |
6.6895 |
0.0275 |
106 |
2,3,3',4,5' |
6.7170 |
8.6053 |
163.610 |
0.0364 |
5.31·10-5 |
6.6482 |
0.0688 |
107 |
2,3,3',4,6 |
6.4870 |
7.5286 |
145.840 |
0.0862 |
1.67·10-4 |
6.3153 |
0.1717 |
108 |
2,3,3',4',6 |
6.5320 |
7.5906 |
148.130 |
0.3660 |
1.60·10-4 |
6.4101 |
0.1219 |
109 |
2,3,3',5,5' |
6.7670 |
8.0899 |
159.630 |
0.0377 |
5.85·10-5 |
6.6891 |
0.0779 |
110 |
2,3,3',5,6 |
6.4570 |
7.1659 |
142.920 |
0.1454 |
9.27·10-5 |
6.3458 |
0.1112 |
111 |
2,3,3',5',6 |
6.5470 |
7.2338 |
144.320 |
0.0712 |
1.41·10-4 |
6.3721 |
0.1749 |
112 |
2,3,4,4',5 |
6.6570 |
8.2871 |
163.400 |
0.0322 |
3.75·10-5 |
6.7731 |
0.1161 |
113 |
2,3,4,4',6 |
6.4970 |
7.7007 |
150.180 |
0.1935 |
6.47·10-4 |
6.4241 |
0.0729 |
114 |
2,3,4,5,6 |
6.3040 |
6.1425 |
134.100 |
0.0371 |
1.31·10-4 |
6.3777 |
0.0737 |
115 |
2,3,4',5,6 |
6.4670 |
7.3123 |
146.830 |
0.3320 |
1.63·10-4 |
6.4673 |
0.0003 |
116 |
2,3',4,4',5 |
7.1210 |
8.7170 |
168.800 |
0.0387 |
4.74·10-5 |
6.8309 |
0.2901 |
117 |
2,3',4,4',6 |
6.5870 |
7.7307 |
152.260 |
0.0954 |
5.32·10-4 |
6.5009 |
0.0861 |
118 |
2,3',4,5,5' |
6.7970 |
8.1850 |
163.840 |
0.0386 |
5.17·10-5 |
6.8354 |
0.0384 |
119 |
2,3',4,5',6 |
6.6470 |
7.2928 |
148.060 |
0.1874 |
2.93·10-4 |
6.5149 |
0.1321 |
120 |
2',3,3',4,5 |
6.6470 |
8.3558 |
159.300 |
0.0339 |
5.32·10-5 |
6.5626 |
0.0844 |
121 |
2',3,4,4',5 |
6.7470 |
8.5776 |
165.470 |
0.0506 |
1.30·10-4 |
6.7401 |
0.0069 |
122 |
2',3,4,5,5' |
6.7370 |
7.8399 |
159.710 |
0.0350 |
5.59·10-5 |
6.7977 |
0.0607 |
123 |
2',3,4,5,6' |
6.5170 |
6.9321 |
142.900 |
0.2070 |
7.11·10-4 |
6.4244 |
0.0926 |
124 |
3,3',4,4'5 |
6.8970 |
9.2142 |
175.600 |
0.2341 |
9.05·10-5 |
6.9342 |
0.0372 |
125 |
3,3',4,5,5' |
6.9570 |
8.6661 |
170.560 |
0.0774 |
1.16·10-4 |
6.9302 |
0.0268 |
126 |
2,2',3,3',4,4' |
6.9610 |
9.3307 |
179.190 |
0.7685 |
7.25·10-4 |
7.0572 |
0.0962 |
127 |
2,2',3,3',4,5 |
7.3210 |
8.5439 |
169.150 |
0.0167 |
1.44·10-3 |
6.8653 |
0.4557 |
128 |
2,2',3,3',4,5' |
7.3910 |
8.6910 |
172.480 |
0.0174 |
1.04·10-3 |
6.9659 |
0.4251 |
129 |
2,2',3,3',4,6 |
6.5870 |
7.6941 |
154.760 |
0.0299 |
8.40·10-4 |
6.6106 |
0.0236 |
130 |
2,2',3,3',4,6' |
6.5870 |
7.8668 |
157.060 |
0.0471 |
6.20·10-4 |
6.6490 |
0.0620 |
131 |
2,2',3,3',5,5' |
6.8670 |
7.8751 |
165.430 |
0.3205 |
3.97·10-4 |
7.0428 |
0.1758 |
132 |
2,2',3,3',5,6 |
7.3040 |
7.1724 |
150.190 |
0.2877 |
1.26·10-3 |
6.6304 |
0.6736 |
133 |
2,2',3,3',5,6' |
7.1510 |
7.1833 |
151.550 |
0.0310 |
1.92·10-4 |
6.7081 |
0.4429 |
134 |
2,2',3,3',6,6' |
6.5110 |
6.4063 |
138.560 |
0.0789 |
2.98·10-4 |
6.4604 |
0.0506 |
135 |
2,2',3,4,4',5' |
7.4410 |
8.6591 |
174.490 |
0.3322 |
6.17·10-4 |
7.1058 |
0.3352 |
136 |
2,2',3,4,4',6 |
6.6770 |
7.7162 |
158.610 |
0.0244 |
6.15·10-4 |
6.7795 |
0.1025 |
137 |
2,2',3,4,4',6' |
6.6770 |
7.8664 |
159.440 |
17.463 |
4.04·10-2 |
6.4781 |
0.1989 |
138 |
2,2',3,4,5,5' |
7.5920 |
7.8362 |
166.540 |
0.1417 |
4.20·10-4 |
7.0949 |
0.4971 |
139 |
2,2',3,4,5,6 |
6.5170 |
6.7464 |
146.160 |
0.0415 |
5.16·10-4 |
6.6424 |
0.1254 |
140 |
2,2',3,4,5,6' |
6.6070 |
6.9031 |
149.310 |
0.5610 |
1.94·10-4 |
6.7639 |
0.1569 |
141 |
2,2',3,4,5',6 |
6.6770 |
7.1854 |
153.300 |
0.1587 |
1.13·10-2 |
6.3768 |
0.3002 |
142 |
2,2',3,4,6,6' |
6.2570 |
6.4146 |
138.950 |
0.0774 |
2.91·10-4 |
6.4743 |
0.2173 |
143 |
2,2',3,4',5,5' |
6.8970 |
8.0823 |
169.690 |
0.0963 |
6.14·10-4 |
7.1201 |
0.2231 |
144 |
2,2',3,4',5,6 |
6.6470 |
7.3016 |
154.700 |
0.0982 |
5.27·10-4 |
6.7896 |
0.1426 |
145 |
2,2',3,4',5,6' |
6.7370 |
7.4207 |
155.500 |
0.1185 |
1.83·10-4 |
6.7892 |
0.0522 |
146 |
2,2',3,4',5',6 |
7.2810 |
7.1955 |
153.590 |
0.0208 |
1.48·10-3 |
6.7440 |
0.5370 |
147 |
2,2',3,4',6,6' |
6.3270 |
6.5980 |
141.640 |
0.0790 |
3.00·10-4 |
6.5158 |
0.1888 |
148 |
2,2',3,5,5',6 |
6.6470 |
6.7917 |
149.800 |
0.0450 |
1.87·10-4 |
6.7967 |
0.1497 |
149 |
2,2',3,5,6,6' |
6.2270 |
6.0509 |
135.760 |
0.0815 |
2.94·10-4 |
6.4866 |
0.2596 |
150 |
2,2',4,4',5,5' |
7.7510 |
8.1709 |
174.310 |
0.1986 |
4.17·10-4 |
7.3015 |
0.4495 |
151 |
2,2',4,4',5,6' |
6.7670 |
7.4824 |
159.260 |
0.0188 |
9.35·10-5 |
6.9258 |
0.1588 |
152 |
2,2',4,4',6,6' |
7.1230 |
6.7120 |
145.340 |
0.0787 |
2.99·10-4 |
6.6313 |
0.4917 |
153 |
2,3,3',4,4',5 |
7.1870 |
8.9450 |
180.720 |
0.0383 |
2.82·10-5 |
7.2624 |
0.0754 |
154 |
2,3,3',4,4',5' |
7.1870 |
8.9295 |
180.060 |
0.0500 |
3.70·10-5 |
7.2402 |
0.0532 |
155 |
2,3,3',4,4',6 |
7.0270 |
8.1071 |
166.380 |
0.3064 |
2.87·10-4 |
6.9903 |
0.0367 |
156 |
2,3,3',4,5,5' |
7.2470 |
8.4424 |
175.580 |
0.0401 |
3.17·10-5 |
7.2467 |
0.0003 |
157 |
2,3,3',4,5,6 |
6.9370 |
7.3571 |
156.700 |
0.0739 |
1.34·10-4 |
6.8677 |
0.0693 |
158 |
2,3,3',4,5',6 |
7.0870 |
7.6904 |
161.910 |
1.6216 |
2.19·10-4 |
7.0623 |
0.0247 |
159 |
2,3,3',4',5,5' |
7.2470 |
8.1738 |
174.530 |
0.0248 |
3.74·10-5 |
7.3121 |
0.0651 |
160 |
2,3,3',4',5,6 |
6.9970 |
7.5592 |
161.980 |
0.2656 |
9.21·10-5 |
7.0309 |
0.0339 |
161 |
2,3,3',4',5',6 |
7.0270 |
7.2468 |
158.640 |
0.1114 |
1.16·10-4 |
7.0031 |
0.0239 |
162 |
2,3,3',5,5',6 |
7.0570 |
7.2209 |
157.930 |
0.1685 |
9.63·10-5 |
6.9874 |
0.0696 |
163 |
2,3,4,4',5,6 |
6.9370 |
7.4614 |
161.140 |
0.5263 |
4.59·10-4 |
7.0393 |
0.1023 |
164 |
2,3',4,4',5,5' |
7.2770 |
8.4739 |
180.990 |
0.0537 |
4.30·10-5 |
7.4731 |
0.1961 |
165 |
2,3',4,4',5',6 |
7.1170 |
7.4940 |
163.550 |
0.2913 |
5.09·10-4 |
7.1138 |
0.0032 |
166 |
3,3',4,4',5,5' |
7.4270 |
9.1962 |
189.460 |
0.0633 |
8.19·10-5 |
7.5426 |
0.1156 |
167 |
2,2',3,3',4,4',5 |
7.2770 |
8.9382 |
188.200 |
0.1129 |
9.03·10-5 |
7.5986 |
0.3216 |
168 |
2,2',3,3',4,4',6 |
6.7040 |
8.1822 |
173.820 |
0.1327 |
1.78·10-3 |
7.2194 |
0.5154 |
169 |
2,2',3,3',4,5,5' |
7.3370 |
8.4591 |
183.610 |
0.0758 |
9.65·10-4 |
7.5620 |
0.2250 |
170 |
2,2',3,3',4,5,6 |
7.0270 |
7.6027 |
165.040 |
0.0482 |
9.50·10-4 |
7.1005 |
0.0735 |
171 |
2,2',3,3',4,5,6' |
7.1170 |
7.5973 |
167.060 |
0.0124 |
8.87·10-5 |
7.2218 |
0.1048 |
172 |
2,2',3,3',4,5',6 |
7.1770 |
7.7165 |
169.220 |
0.0266 |
8.26·10-5 |
7.2683 |
0.0913 |
173 |
2,2',3,3',4,6,6' |
6.7670 |
7.0135 |
155.320 |
0.0680 |
2.45·10-4 |
6.9467 |
0.1797 |
174 |
2,2',3,3',4',5,6 |
7.0870 |
7.6992 |
169.170 |
0.0578 |
2.24·10-4 |
7.2703 |
0.1833 |
175 |
2,2',3,3',5,5',6 |
7.1470 |
7.3049 |
165.140 |
0.1383 |
2.63·10-4 |
7.2622 |
0.1152 |
176 |
2,2',3,3',5,6,6' |
6.7370 |
6.6644 |
151.780 |
0.0715 |
2.45·10-4 |
6.9374 |
0.2004 |
177 |
2,2',3,4,4',5,5' |
7.3670 |
8.6195 |
188.140 |
0.0550 |
4.81·10-3 |
7.5491 |
0.1821 |
178 |
2,2',3,4,4',5,6 |
7.1170 |
7.7127 |
170.270 |
6.8760 |
6.77·10-3 |
7.5427 |
0.4257 |
179 |
2,2',3,4,4',5,6' |
7.2070 |
7.8062 |
171.680 |
0.6290 |
2.78·10-4 |
7.3739 |
0.1669 |
180 |
2,2',3,4,4',5',6 |
7.2070 |
7.8790 |
173.190 |
0.1139 |
3.01·10-4 |
7.3733 |
0.1663 |
181 |
2,2',3,4,4',6,6' |
6.8570 |
7.1523 |
158.990 |
0.0677 |
2.43·10-4 |
7.0505 |
0.1935 |
182 |
2,2',3,4,5,5',6 |
7.9330 |
6.8634 |
164.000 |
0.4577 |
1.39·10-5 |
7.4292 |
0.5038 |
183 |
2,2',3,4,5,6,6' |
6.6970 |
6.2008 |
148.390 |
0.0693 |
2.38·10-4 |
6.9828 |
0.2858 |
184 |
2,2',3,4',5,5',6 |
7.1770 |
7.3776 |
168.560 |
0.0448 |
1.14·10-4 |
7.3819 |
0.2049 |
185 |
2,2',3,4',5,6,6' |
6.8270 |
6.7527 |
155.210 |
0.0714 |
2.43·10-4 |
7.0519 |
0.2249 |
186 |
2,3,3',4,4',5,5' |
7.7170 |
9.1221 |
195.930 |
0.0488 |
3.59·10-5 |
7.8604 |
0.1434 |
187 |
2,3,3',4,4',5,6 |
7.4670 |
8.2487 |
178.880 |
0.5202 |
6.32·10-4 |
7.4850 |
0.0180 |
188 |
2,3,3',4,4',5',6 |
7.5570 |
8.2502 |
179.590 |
0.6986 |
6.92·10-4 |
7.5259 |
0.0311 |
189 |
2,3,3',4,5,5',6 |
7.5270 |
7.7471 |
173.540 |
0.0626 |
2.84·10-4 |
7.4413 |
0.0857 |
190 |
2,3,3',4',5,5',6 |
7.5270 |
7.6407 |
174.490 |
0.4337 |
1.22·10-4 |
7.5600 |
0.0330 |
191 |
2,2',3,3',4,4',5,5' |
8.6830 |
8.8559 |
201.230 |
0.0917 |
6.13·10-4 |
8.1879 |
0.4951 |
192 |
2,2',3,3',4,4',5,6 |
7.5670 |
8.1340 |
185.280 |
0.0365 |
5.66·10-5 |
7.8039 |
0.2369 |
193 |
2,2',3,3',4,4',5',6 |
7.6570 |
8.0500 |
185.640 |
0.0407 |
3.72·10-5 |
7.8562 |
0.1992 |
194 |
2,2',3,3',4,4',6,6' |
7.3070 |
7.5022 |
173.290 |
0.0595 |
1.97·10-4 |
7.5363 |
0.2293 |
195 |
2,2',3,3',4,5,5',6 |
7.6270 |
7.7337 |
180.830 |
0.0627 |
1.47·10-4 |
7.7742 |
0.1472 |
196 |
2,2',3,3',4,5,6,6' |
7.2070 |
6.9867 |
166.280 |
0.0619 |
1.96·10-4 |
7.4437 |
0.2367 |
197 |
2,2',3,3',4,5',6,6' |
7.2770 |
7.1190 |
169.460 |
0.0626 |
1.98·10-4 |
7.5285 |
0.2515 |
198 |
2,2',3,3',4',5,5',6 |
7.6270 |
7.6994 |
180.700 |
0.0627 |
1.52·10-4 |
7.7827 |
0.1557 |
199 |
2,2',3,3',5,5',6,6' |
8.4230 |
6.6004 |
165.330 |
0.0656 |
1.96·10-4 |
7.5645 |
0.8585 |
200 |
2,2',3,4,4',5,5',6 |
7.6570 |
7.7846 |
184.510 |
0.6727 |
3.23·10-4 |
7.9513 |
0.2943 |
201 |
2,2',3,4,4',5,6,6' |
7.3070 |
7.0818 |
170.100 |
0.0617 |
1.93·10-4 |
7.5726 |
0.2656 |
202 |
2,3,3',4,4',5,5',6 |
8.0070 |
8.1803 |
191.360 |
1.1642 |
5.30·10-4 |
8.1139 |
0.1069 |
203 |
2,2',3,3',4,4',5,5',6 |
9.1430 |
7.9885 |
197.410 |
0.0270 |
6.07·10-5 |
8.4003 |
0.7427 |
204 |
2,2',3,3',4,4',5,6,6' |
7.7470 |
7.4690 |
184.980 |
0.0550 |
1.57·10-4 |
8.0680 |
0.3210 |
205 |
2,2',3,3',4,5,5',6,6' |
8.1640 |
7.1318 |
180.950 |
0.0577 |
1.60·10-4 |
8.0319 |
0.1321 |
206 |
2,2',3,3',4,4',5,5',6,6' |
9.6030 |
7.4035 |
197.030 |
0.0512 |
1.33·10-4 |
8.6287 |
0.9743 |
Three molecular descriptors take into consideration the geometry of PCBs (IIDDKGg, IHDRKEg, and aSMMjQg) and one the topology of compounds (aHMmjQt). As atomic property, two descriptors consider the partial change (aHMmjQt, and aSMMjQg), one the group electronegativity (IIDDKGg) and one the atomic electronegativity (IHDRKEg). Looking at the interaction descriptor (the fifth letter in descriptors name) it can be observed that all descriptors consider the elastic force.
The results of multiple linear regressions associated to the four-varied model (see table 2, and table 3) sustain the estimation and prediction abilities of the best performing SAR model.
In the table 3 are the 95% probability of confidence intervals - lower (95%CIL) and upper (95%CIU) boundaries, coefficients, standard error (StdErr) of the coefficient, Student test parameter (t) and Student probability (pt).
Table 2. Statistics associated with the tetra-varied model
Characteristic |
Notation |
Values |
Correlation coefficient |
r |
0.9575 |
Squared correlation coefficient |
r2 |
0.9168 |
Adjusted squared correlation coefficient |
r2adj |
0.9151 |
Standard error of estimated |
sest |
0.2420 |
Fisher parameter |
Fest |
554 |
Probability of wrong model |
pest(%) |
< 1·10-15 |
Cross-validation leave-one-out (loo) score |
r2cv(loo) |
0.9093 |
Fisher parameter for loo analysis |
Fpred |
504 |
Probability of wrong model for loo analysis |
ppred(%) |
< 1·10-15 |
Standard error for loo analysis |
sloo |
0.2526 |
The difference between r2 and r2cv(loo) |
r2 - r2cv(loo) |
|
Squared correlation coefficients between each descriptor and measured octanol-water partition coefficients or between pairs of descriptors |
r2(IIDDKGg, IHDRKEg) |
0.48245 |
r2(IIDDKGg, aHMmjQt) |
0.00005 |
|
r2(IIDDKGg, aSMMjQg) |
0.00385 |
|
r2(IHDRKEg, aHMmjQt) |
0.00039 |
|
r2(IHDRKEg, aSMMjQg) |
0.00073 |
|
r2(aHMmjQt, aSMMjQg) |
0.24805 |
|
r2(IIDDKGg, log Kow) |
0.15111 |
|
r2(IHDRKEg, log Kow) |
0.78907 |
|
r2(aSMMjQg, log Kow) |
0.00932 |
|
r2(aHMmjQt, log Kow) |
0.00786 |
Table 3. Statistics associated with the four-varied model
|
95%CIL |
Coefficients |
95%CIU |
StdError |
t |
pt (%) |
Intercept |
2.735 |
3.039 |
3.343 |
0.154 |
19.716 |
7.27·10-47 |
IIDDKGg |
-0.477 |
-0.421 |
-0.365 |
0.028 |
-14.804 |
5.09·10-32 |
IHDRKEg |
0.042 |
0.044 |
0.046 |
0.001 |
41.725 |
5.97·10-99 |
aHMmjQt |
0.049 |
0.070 |
0.090 |
0.010 |
6.639 |
2.89·10-8 |
aSMMjQg |
-47.601 |
-37.499 |
-27.397 |
5.123 |
-7.319 |
5.86·10-10 |
The model which consider in estimation four molecular descriptors is significant statistically, having a probability of a wrong model less than 1·10-15 (%). The estimation ability of the SAR model is sustained by the value of the correlation coefficient (r = 0.9575), the confidence boundaries associated with the coefficients (see table 3), and probabilities associated with Student tests (for all coefficients less than 0.001 - see table 3). Almost ninety-two percent (r2 = 0.9168) from variation of octanol-water partition coefficient can be explained by its linear relationship with the variation of the four molecular descriptors used in the model. The probability of wrong model for leave-one-out analysis (ppred(%) < 1·10-15) and its associated Fisher parameter (Fpred = 504) sustains the estimation ability of the model. The four-varied SAR model is a stable one, stability sustained by the values of difference between correlation coefficient and cross validation leave-one-out correlation score (r2 - r2cv(loo) = 0.0075). The power of the four-varied model in octanol-water partition coefficient prediction of PCBs is sustained by the absence of multicolinearity of descriptors used by the model (see the squared correlation coefficients between pairs of descriptors, which always is less than 0.48 - table 2).
The plot of dependency between measured (logKow) and estimated based on the structure of polychlorinated biphenyls compounds obtained with the tetra-varied model is in figure 1.
Figure 1. Measured vs. estimated logKow by the tetra-varied model
The estimation values of octanol-water partition coefficients by the use of the four-varied model of are less or greater than measured values (see figure 2). Note that, the mean and 95% confidence intervals of the mean and standard error for measured (mMeasured = 6.4802, 95%CIMeasured = [6.3709, 6.5895], StdErrMeasured = 0.0554) and estimated (mEstimated = 6.4806, 95%CIEstimated = [6.3664, 6.5947], StdErrEstimated = 0.0579) octanol-water partition coefficients are almost equal.
Figure 2. Variation of measured (blue line) and estimated (red line) by the four-varied model of octanol-water partition coefficient for PCBs
In order to seen the estimation abilities of four-varied model, measured and estimated values were sort by the absolute differences between estimated and measured octanol-water partition coefficient of PCBs and split into two subsets (first containing one-hundred PCBs and second containing the other one-hundred and six PBCs). The graphical representations are in figure 3a (one-hundred compounds) and 3b (one-hundred and six compounds), where the PCB number was associated with corresponding estimated and measured values.
Figure 3a. Measured (blue line) and estimated by the tetra-varied model (red-line)
of octanol-water partition coefficients for one-hundred PCBs
Figure 3b. Measured (blue line) and estimated by the tetra-varied model (red line)
of octanol-water partition coefficients for one-hundred and six PCB
Analyzing the residuals of the four-varied model allowed us to assess the suitability of the model. Looking at the differences between measured and estimated octanol-water partition coefficient for PCBs (figure 4) it can be observed that the values vary around zero and most of them between -0.5 and 0.5.
Figure 4. The differences between measured and estimated by the tetra-varied model of octanol-water partition coefficients for PCBs
The prediction abilities of the four-varied SAR model were studied through training and test sets analysis, and the results are in table 4. There were analyzed twelve situations, starting with a training sample size equal with 116 and increasing the number of PCBs included into training sets through randomization with seven until one hundred ninety-three. In table 4, there were included the number of PCBs in training sets (Notr), the coefficients of the model, the squared correlation coefficient for training set (rtr2), Fisher parameter associated with training set regression (Ftr), the number of the PCBs in test sets (Nots), the squared correlation coefficient for test set (rts2), Fisher parameter associated with training set regression (Fts), the mean (Mean) and standard deviation (StDev) for squared correlation coefficients and the 95% probability CI [95%CIL and 95%CIU] for coefficients.
Table 4. Results of training vs. test sets analysis
Notr |
intercept |
IIDDKGg |
IHDRKEg |
aHMmjQt |
aSMMjQg |
|
rtr2 |
Ftr |
Nots |
rts2 |
Fts |
116 |
3.070 |
-0.408 |
0.043 |
0.064 |
-34.937 |
|
0.9141 |
295* |
90 |
0.9219 |
235* |
123 |
3.058 |
-0.390 |
0.043 |
0.064 |
-43.454 |
|
0.9229 |
353* |
83 |
0.9043 |
176* |
130 |
2.957 |
-0.413 |
0.044 |
0.067 |
-33.462 |
|
0.9232 |
376* |
76 |
0.9068 |
169* |
137 |
3.011 |
-0.438 |
0.045 |
0.064 |
-32.008 |
|
0.9004 |
298* |
69 |
0.9432 |
256* |
144 |
3.090 |
-0.450 |
0.045 |
0.062 |
-45.236 |
|
0.9143 |
371* |
62 |
0.9186 |
148* |
151 |
3.102 |
-0.432 |
0.044 |
0.062 |
-42.983 |
|
0.9173 |
405* |
55 |
0.9075 |
122* |
158 |
3.137 |
-0.460 |
0.046 |
0.073 |
-37.319 |
|
0.9200 |
440* |
48 |
0.9041 |
82* |
165 |
3.091 |
-0.428 |
0.044 |
0.070 |
-37.661 |
|
0.9143 |
427* |
41 |
0.9247 |
110* |
172 |
3.063 |
-0.426 |
0.044 |
0.069 |
-36.945 |
|
0.9161 |
456* |
34 |
0.9202 |
83* |
179 |
3.085 |
-0.429 |
0.044 |
0.069 |
-37.219 |
|
0.9098 |
439* |
27 |
0.9582 |
106* |
186 |
2.990 |
-0.420 |
0.045 |
0.070 |
-37.650 |
|
0.9090 |
452* |
20 |
0.9876 |
178* |
193 |
3.067 |
-0.430 |
0.044 |
0.074 |
-37.466 |
|
0.9160 |
513* |
13 |
0.9249 |
24* |
|
|
|
|
|
|
|
|
|
|
* p < 0.001 |
|
95%CIL |
3.028 |
-0.439 |
0.044 |
0.065 |
-40.566 |
|
0.9148 |
Mean |
0.9268 |
|
|
95%CIU |
3.092 |
-0.415 |
0.045 |
0.070 |
-35.490 |
|
0.0063 |
StDev |
0.0250 |
|
All squared correlation coefficients in training as well as in test sets are greater than 0.9, sustaining the prediction ability of the four-varied model. More, the mean of squared correlation coefficients in test sets is a little bit higher compared with the mean of squared correlation coefficient in training sets, and the dispersions of squared correlation coefficients are very small for both sets. All the regressions in training and test sets are highly significant (p < 0.001).
Analyzing the regressions coefficients it can be observed that with no exception the values of coefficients respect the 95% confidence intervals associated to the four-varied model (see table 3 and table 4). More, as it is expected, the 95% CI values (table 4) obtained in training and test sets analyses are contained by the 95% CI values of four-varied model (table 3).
The plot of measured vs. estimated octanol-water partition coefficients in training set (blue line and dots) of sample size equal one-hundred thirty-seven (corresponding with 2/3 from total sample of PCBs) and corresponding test set (red line and dots) of sample size equal with sixty-nine (1/3 from total sample of PCBs) is in figure 5.
Figure 5. Training (137 PCBs) vs test (69 PCBs) analysis with four-varied model
Starting with the above describe model, and by the use of the original software [9], the octanol-water partition coefficient of new polychlorinated biphenyls can be obtains in a short time, without any experiments, following the next steps: drawing by the use of HyperChem software the three dimensional structure of the new PCB, choosing the model of prediction from the list (in our case PCB_lkow), browsing the *.hin file, and computing the octanol-water partition coefficient based on the four-varied SAR equation.
Conclusions
Modeling the octanol-water partition coefficient of polychlorinated biphenyls by integration of complex structural information provide a stable and performing four-varied model, allowing us to make remarks about relationship between structure of PCBs and associated octanol-water partition coefficients. Thus, the octanol-water partition coefficient of studied PCBs is like to be of geometry and topology nature, depending by the partial change, group and atomic electronegativity as atomic properties, and being in relation with the elastic force.
Acknowledgement
Research was in part supported through project ET36/2005 by UEFISCSU Romania.
References
[[1]] George C. J., Bennett G. F., Simoneaux D., George W. J., Polychlorinated biphenyls. A toxicological review, Journal of Hazardous Materials, 18(2), p. 113-144, 1988.
[[2]] Borja J., Taleon D. M., Auresenia J., Gallardo S., Polychlorinated biphenyls and their biodegradation, Process Biochemistry, 40(6), p. 1999-2013, 2005.
[[3]] Hertz-Picciotto I., Charles M. J., James R. A. , Keller J. A., Willman E., Teplin S., In utero polychlorinated biphenyl exposures in relation to fetal and early childhood growth, Epidemiology, 16(5), p. 648-656, 2005.
[[4]] Hansen B. G., Paya-Perez A. B., Rahman M., Larsen B. R., QSARs for K(ow) and K(oc) of PCB congeners: A critical examination of data, assumptions and statistical approaches, Chemosphere, 39(13), p. 2209-2228, 1999.
[[5]] Zou J.-W., Jiang Y.-J., Hu G.-X., Zeng M., Zhuang S.-L., Yu Q.-S., QSPR/QSAR studies on the physicochemical properties and biological activities of polychlorinated biphenyls, Acta Physico - Chimica Sinica, 21(3), p. 267-272, 2005.
[[6]] Eisler R., Belisle A. A., Planar PCB Hazards to Fish, Wildlife, and Invertebrates: A Synoptic Review, Contaminant Hazard Reviews, p. 1-96, 1996.
[[7]] Jäntschi L., Molecular Descriptors Family on Structure Activity Relationships 1. The review of Methodology, Leonardo Electronic Journal of Practices and Technologies, AcademicDirect, 6, p. 76-98, 2005.
[[8]] ***, Leave-one-out Analysis, © 2005, Virtual Library of Free Software, available at: http://vl.academicdirect.org/molecular_topology/mdf_findings/loo.
[[9]] ***, MDF SAR Predictor, © 2005, Virtual Library of Free Software, available at: http://vl.academicdirect.org/molecular_topology/mdf_findings/sar.