Supplemental Table 1. List of blood brain barrier permeation datasets, modeling approaches used and their literature references.

Number of molecules in training set / Number of molecules in external test set / Modeling approach / Notes / Reference
60 / AM1 –derived parameters / Dipole, logP, molecular weight and HOMO important / (1)
55 / 6 test set 1
10 test set 2 / Solvation free energy computed with AMSOL 5 / (2)
18 / Used 5 solute descriptors and a linear free energy relationship / Data from rat, / (3)
63 / MolSurf to generate 14 molecular descriptors used including logP and hydrogen bonding and PLS / Data from (2) / (4)
55 / 7 test set 1
6 test set 2 / Rapid polar molecular surface area and ClogP or MlogP / Data from (2) / (5)
12 HIV antiretrovirals / No modeling performed / Summary of in vitro BMEC, human CSF/Plasma and rat brain/plasma data / (6)
61 / 15 test set 1
25 test set 2 / 18 Topological and constitutional Descriptors and PLS / Included a measure of distance from the training set DModX / (7)
~9000 / 13000 BBB+, 57000 BBB- test set 1
275 test set 2 / 7 1D and 166 2D descriptors with Bayesian Neural networks / 93% accuracy for actives and 72 % accuracy for inactives in test set 2 / (8)
20 histamine antagonists and 6 B-blockers / GRID calculations and regression / (9)
45 / Dynamic and static polar molecular surface Correlated with log brain/blood / Rat brain/blood data – also looked at frequency distribution of 776 oral CNS molecules and 1590 non-CNS molecules Optimal polar surface area is 60-70A2
110 (includes stereoisomers) / 120 diverse (49 BBB+, 71 BBB-) / Categorical data with VolSurf and GRID. / Predicted 90% of BBB+ and 65% for BBB- / (10)
7000 (3500 BBB+ and 3500 BBB-) / Test set 1. 7000 diverse (3500 BBB+ and 3500 BBB-), Test set 2. 704 molecules from the MedChem database / Neural network using Unity fingerprint descriptors / Classification test set 1. 84.2% active 87,2 % inactive, test set 2. 89.9 predictive / (11)
61 / 14 test set 1
25 test set 2 / logP, polar surface area, solvated hydrogen bond acceptor number with Principal component regression / (12)
76 / Monte Carlo simulations to calculate the solvent-accessible surface area and other descriptors followed by linear regression / (13)
55 / 11 diverse / Electrotopological state indices with multiple linear regression or artificial neural network / Data from (2) test set correlation r2= 0.84 / (14)
148 / Linear free energy relationship descriptors and multivariate linear regression analysis / (15)
12 / No modeling performed / Comparison of in vitro and in vivo BBB permeability data / (16)
106 / 28 / Electrotopological descriptors and linear regression / Correctly predicted 27 of 28 in the test set / (17)
56 / 7 / Intermolecular interaction descriptors with Genetic function approximation / Suggested importance of solute flexibility, solute-membrane flexibility etc. / (18)
83 / VolSurf PLS models / (19)
48 / 17 diverse Test set 1
150 diverse test set 2. / Stepwise linear regression with 34 descriptors in cerius 2 / 88% overall correct predictions on testing – however authors recommend the Cerius2 ADME log BB 2D model / (20)
58 / 34 test set 1
181 test set 2
2000 test set 3 / Linear regression and Genetic PLS with electrotopological descriptors / Predictions better for BBB+
logBB values >-1 used as cutoff / (21)
1696 / 82 (10) / Electronic and molecular surface area descriptors with PLS- discriminant analysis / Test set prediction classification > 90%
P-gp substrates also assessed / (22)
150 / Intermolecular interaction descriptors 4D molecular similarity with Genetic function approximation / Dataset divided into 3 subsets / (23)
35 / 3 / CODES topological descriptors with neural network algorithm / Log BB > -0.3 used as cutoff / (24)
119 / 28 (10) / TOPS-MODE descriptors with linear regression / 85.7% correct classification for test set / (25)
190 / 350 / Charged polar surface area, rule of 5 and MOE descriptors with Recursive partitioning / Test set poor predictions suggested due to P-gp efflux and distance from training set. Local models also built for histamine, NK-1 and kinase data sets / (26)
88 / 13 test set 1
15 test set 2
92 test set 3 (10) / 324 descriptors with a variable selection methodwith multiple linear regression / ~80 % accuracy for BBB+ and BBB- with test set 3 / (27)
415 / 199 diverse descriptors calculated and used with a feature reduction algorithm and Linear regression, linear discriminant analysis, decision tree, k-nearest neighbors, probabilistic neural network and Support vector machines / SVM had the highest prediction accuracy upon cross validation / (28)
302 / Linear free energy relationship / Suggested experimental error in logBB data is 0.3log units / (29)
113 / 19 test set 1
40 test set 2 / CODESSA_Pro and multilinear regression, and ISIDA fragment descriptors / Both approaches predict the test sets well, test set 1 r2 = 0.77 CODESSA, r2 = 0.27 ISIDA and test set 2, 34 / 40 and ISIDA 30 / 40 / (30)
291 (29) / Linear free energy relationship descriptors with k-nearest neighbors, multiple linear regression / Used leave group out cross validation / (31)
103 / COSMO-RS -moments with linear regression / (32)
1093 (22) / 500 / 19 descriptors with recursive partitioning and binomial PLS / 95% accuracy on test set / (33)

References

1.M.E. Brewster, E. Pop, M.-J. Huang, and N. Bodor. AM1-based model system for estimation of brain / blood concentration ratios. Int J Quant Chem: Quant Biol Symp. 23:1775-1787 (1996).

2.F. Lombardo, J.F. Blake, and W.J. Curatolo. Computation of brain-blood partitioning of organic solutes via free energy calculations. Journal of medicinal chemistry. 39:4750-4755 (1996).

3.J.A. Gratton, M.H. Abraham, M.W. Bradbury, and H.S. Chadha. Molecular factors influencing drug transfer across the blood-brain barrier. J Pharm Pharmacol. 49:1211-1216 (1997).

4.U. Norinder, P. Sjoberg, and T. Osterberg. Theoretical calculation and prediction of brain-blood partitioning of organic solutes using molsurf parameterization and PLS statistics. Journal of pharmaceutical sciences. 87:952-959 (1998).

5.D.E. Clark. Rapid calculation of polar molecular surface area and its application to the prediction of transport phenomena. 2. Prediction of blood-brain barrier penetration. Journal of pharmaceutical sciences. 88:815-821 (1999).

6.M. Yazdanian. Blood-brain barrier properties of human immunodeficiency virus antiretrovirals. Journal of pharmaceutical sciences. 88:950-954 (1999).

7.J.M. Luco. Prediction of Brain-Blood distribution of a large set of drugs from structurally derived descriptors using partial least squares (PLS) modeling. Journal of chemical information and computer sciences. 39:396-404 (1999).

8.Ajay, G.W. Bemis, and M.A. Murcko. Designing libraries with CNS activity. Journal of medicinal chemistry. 42:4942-4951 (1999).

9.V. Segarra, M. Lopez, H. Ryder, and J.M. Palacios. Prediction of drug permeability based on Grid calculations. QSAR. 18:474-481 (1999).

10.P. Crivori, G. Cruciani, P.A. Carrupt, and B. Testa. Predicting blood-brain barrier permeation from three-dimensional molecular structure. Journal of medicinal chemistry. 43:2204-2216 (2000).

11.G.M. Keseru, L. Molnar, and I. Greiner. A neural network based virtual high throughput screening test for the prediction of CNS activity. Combinatorial chemistry & high throughput screening. 3:535-540 (2000).

12.M. Feher, E. Sourial, and J.M. Schmidt. A simple model for the prediction of blood-brain partitioning. International journal of pharmaceutics. 201:239-247 (2000).

13.Y.N. Kaznessis, M.E. Snow, and C.J. Blankley. Prediction of blood-brain partitioning using Monte Carlo simulations of molecules in water. Journal of computer-aided molecular design. 15:697-708 (2001).

14.R. Liu, H. Sun, and S.S. So. Development of quantitative structure-property relationship models for early ADME evaluation in drug discovery. 2. Blood-brain barrier penetration. Journal of chemical information and computer sciences. 41:1623-1632 (2001).

15.J.A. Platts, M.H. Abraham, Y.H. Zhao, A. Hersey, L. Ijaz, and D. Butina. Correlation and prediction of a large blood-brain distribution data set--an LFER study. European journal of medicinal chemistry. 36:719-730 (2001).

16.S. Lundquist, M. Renftel, J. Brillault, L. Fenart, R. Cecchelli, and M.P. Dehouck. Prediction of drug transport through the blood-brain barrier in vivo: a comparison between two in vitro cell models. Pharmaceutical research. 19:976-981 (2002).

17.K. Rose, L.H. Hall, and L.B. Kier. Modeling blood-brain barrier partitioning using the electrotopological state. Journal of chemical information and computer sciences. 42:651-666 (2002).

18.M. Iyer, R. Mishru, Y. Han, and A.J. Hopfinger. Predicting blood-brain barrier partitioning of organic molecules using membrane-interaction QSAR analysis. Pharmaceutical research. 19:1611-1621 (2002).

19.F. Ooms, P. Weber, P.A. Carrupt, and B. Testa. A simple model to predict blood-brain barrier permeation from 3D molecular fields. Biochimica et biophysica acta. 1587:118-125 (2002).

20.M. Lobell, L. Molnar, and G.M. Keseru. Recent advances in the prediction of blood-brain partitioning from molecular structure. Journal of pharmaceutical sciences. 92:360-370 (2003).

21.G. Subramanian and D.B. Kitchen. Computational models to predict blood-brain barrier permeation and CNS activity. Journal of computer-aided molecular design. 17:643-664 (2003).

22.M. Adenot and R. Lahana. Blood-brain barrier permeation models: discriminating between potential CNS and non-CNS drugs including P-glycoprotein substrates. Journal of chemical information and computer sciences. 44:239-248 (2004).

23.D. Pan, M. Iyer, J. Liu, Y. Li, and A.J. Hopfinger. Constructing optimum blood brain barrier QSAR models using a combination of 4D-molecular similarity measures and cluster analysis. Journal of chemical information and computer sciences. 44:2083-2098 (2004).

24.I. Dorronsoro, A. Chana, M.I. Abasolo, A. Castro, C. Gil, M. Stud, and A. Martinez. CODES/Neural network model: a useful toolfor in silico prediction of oral absorption and blood brain barrier permeability of structurally diverse drugs. QSAR Comb Sci. 23:89-98 (2004).

25.M.A. Cabrera, M. Bermejo, M. Perez, and R. Ramos. TOPS-MODE approach for the prediction of blood-brain barrier permeation. Journal of pharmaceutical sciences. 93:1701-1717 (2004).

26.S.R. Mente and F. Lombardo. A recursive-partitioning model for blood-brain barrier permeation. Journal of computer-aided molecular design. 19:465-481 (2005).

27.R. Narayanan and S.B. Gunturi. In silico ADME modelling: prediction models for blood-brain barrier permeation using a systematic variable selection method. Bioorganic & medicinal chemistry. 13:3017-3028 (2005).

28.H. Li, C.W. Yap, C.Y. Ung, Y. Xue, Z.W. Cao, and Y.Z. Chen. Effect of selection of molecular descriptors on the prediction of blood-brain barrier penetrating and nonpenetrating agents by statistical learning methods. Journal of chemical information and modeling. 45:1376-1384 (2005).

29.M.H. Abraham, A. Ibrahim, Y. Zhao, and W.E. Acree, Jr. A data base for partition of volatile organic compounds and drugs from blood/plasma/serum to brain, and an LFER analysis of the data. Journal of pharmaceutical sciences. 95:2091-2100 (2006).

30.A.R. Katritzky, M. Kuanar, S. Slavov, D.A. Dobchev, D.C. Fara, M. Karelson, W.E. Acree, Jr., V.P. Solov'ev, and A. Varnek. Correlation of blood-brain penetration using structural descriptors. Bioorganic & medicinal chemistry. 14:4888-4917 (2006).

31.D.A. Konovalov, D. Coomans, E. Deconinck, and Y.V. Heyden. Benchmarking of QSAR models for blood-brain barrier permeation. Journal of chemical information and modeling. 47:1648-1656 (2007).

32.K. Wichmann, M. Diedenhofen, and A. Klamt. Prediction of blood-brain partitioning and human serum albumin binding based on COSMO-RS sigma-moments. Journal of chemical information and modeling. 47:228-233 (2007).

33.Y.H. Zhao, M.H. Abraham, A. Ibrahim, P.V. Fish, S. Cole, M.L. Lewis, M.J. de Groot, and D.P. Reynolds. Predicting penetration across the blood-brain barrier from simple descriptors and fragmentation schemes. Journal of chemical information and modeling. 47:170-175 (2007).