Prediction of Blood-Brain Barrier Permeability of Compounds Based on Machine Learning
FU Ming-yu1a, DONG Yi-xiao2, WU Chun-yong1b, HOU Feng-zhen1a*
1a. Key Laboratory of Biomedical Functional Materials, School of Science, 1b. Department of Pharmaceutical Analysis, China Pharmaceutical University, Nanjing 210009, China; 2. UNC Gillings School of Global Public Health, UNC, NC 27599, USA
Abstract：OBJECTIVE To build the blood-brain barrier permeability(logBB) prediction model based on machine learning, and evaluate its prediction effect. METHODS Molecular structure information and logBB of 360 compounds were collected. Genetic algorithm(GA) combined with extreme gradient boosting(XGBoost) was used to build the compound logBB prediction model. RESULTS After 10-fold cross validation, it was shown that the model had a squared correlation coefficient of 0.63 and a mean squared error of 0.23, indicating good prediction performance. Furthermore, the importance of features of the model was analyzed. The top five molecular descriptors with the highest influence on logBB were summarized. The top five properties were topological polar surface area, hydrogen bond, hydrophobic/hydrophilic, oil-water partition coefficient, and octanol-water partition coefficient. CONCLUSION The performance of the established model is superior to that of the model built using only XGBoost or using genetic algorithms combined with support vector machine, and is far superior to the prediction algorithm that comes with Discovery Studio (DS, 2016 version). This study can provide guidance for drug research in the field of treatment of brain related diseases.
BRIAN T H, TOMAS P D. The blood-brain barrier/neurovascular unit in health and disease [J]. Pharmacol Rev, 2005, 57(2): 173-185.
[2]
RAJNISH K, ANJU S, RAJESH K T. Can we predict blood barrier permeability of ligands using computational approaches [J]. Interdiscip Sci, 2013, 5(2): 95-101.
[3]
JAKUB C, JAN K, RAFAEL D, et al. In vitro and in silico acetylcholinesterase inhibitory activity of thalictricavine and canadine and their predicted penetration across the blood-brain barrier [J]. Molecules, 2019, 24(7): 1340-1350.
[4]
ZHANG J Y, DIAO J F, LV Q L, et al. Research progress in in vitro models for assessing drug transport across blood-brain barrier [J]. Chin J New Drug(中国新药杂志), 2015, 24(21): 2453-2458.
[5]
YANG Y F, WU N, YANG X W. Establishment of MDCK-p Ha MDR cell model and standard operation procedure for assessing blood-brain barrier permeability of chemical components of traditional Chinese medicine [J]. China J Chin Mater Med(中国中药杂志), 2016, 41(14): 2706-2711.
[6]
KIRIL L, PRANAS J, REMIGIJUS D. Improving the prediction of drug disposition in the brain [J]. Expert Opin Drug Met, 2013, 9(4): 473-486.
[7]
NAGPAL K, SINGH S K, MISHRA D N. Drug targeting to brain: a systematic approach to study the factors, parameters and approaches for prediction of permeability of drugs across BBB [J]. Expert Opin Drug Del, 2013, 10(7): 927-955.
[8]
MARTINS I F, TEIXEIRA A L, PINHEIRO L, et al. A bayesian approach to in silico blood-brain barrier penetration modeling [J]. J Chem Inf Model, 2012, 52(6): 1686-1697.
[9]
ADENOT M, LAHANA R. Blood-brain barrier permeation models: discriminating between potential CNS and non-CNS drugs including P-glycoprotein substrates [J]. J Chem Inf Comp Sci, 2008, 44(1): 239-248.
[10]
WU Z Y, PAN J, YUAN Y, et al. Comparison of prediction models for blood brain barrier permeability and analysis of the molecular descriptors [J]. Pharmazie, 2012, 67(7): 628-634.
[11]
KARELSON M, DOBCHEV D, TAMM T, et al. Correlation of blood-brain penetration and human serum albumin binding with theoretical descriptors [J]. Arkivoc, 2008, 28(16): 38-60.
[12]
NIE Z H, LI J F, ZHAO L. Research progress in cell-culture models of blood-brain barrier [J]. Chin Pharm J(中国药学杂志), 2018, 53(3), 165-168.
[13]
ZHANG M Y, LIU M, HONG Z Y, et al. Research progress in the traditional Chinese medicine treatment of Alzheimer's disease and the blood-brain barrier transporters [J]. Chin Pharm J(中国药学杂志), 2018, 53(7), 490-495.
[14]
ZHANG L Y, ZHU H, TUDOR I T, et al. QSAR modeling of the blood-brain barrier permeability for diverse organic compounds [J]. Pharm Res, 2008, 25(8): 1902-1914.
[15]
SANTIAGO V, MAYUKH C, STEFANO C. Prediction of passive blood-brain partitioning: Straightforward and effective classification models based on in silico derived physicochemical descriptors [J]. Mol Graph Model, 2010, 28(8): 899-903.
[16]
JELICA V, KATARINA N, VLADIMIR D, et al. Prediction of blood-brain barrier permeation of α-adrenergic and imidazoline receptor ligands using PAMPA technique and quantitative structure permeability relationship analysis [J]. Eur J Pharm Sci, 2015, 68(1): 94-105.
[17]
ZHANG D Q, XIAO J F, ZHOU N N, et al. A genetic algorithm based support vector machine model for blood-brain barrier penetraion prediction [J]. Biomed Res Int, 2015, 34(1): 60-67.
[18]
CHEN T Q, CARLOS G. XGBoost: A scalable tree boosting system[C]. San Fransisco: ACM Digital Library, 2016:785-794.
[19]
VILAR S, CHAKRABARTI M, COSTANZI S. Prediction of passive blood-brain partitioning: Straightforward and effective classification models based on in silico derived physicochemical descriptors [J]. Mol Graph Model, 2010, 28(8): 899-903.
[20]
WINKLER D A, Burden F R. Modelling blood-brain barrier partitioning using Bayesian neural nets [J]. J Mol Graph Model, 2004, 22(6): 499-505.
[21]
HUTTER M C, Prediction of blood-brain barrier permeation using quantum chemically derived information [J]. J Comput Aid Mol Des, 2003, 17(7): 415-433.
[22]
RAMAMURTHI N, SITARAMA B G. In silico ADME modelling: prediction models for blood-brain barrier permeation using a systematic variable selection method [J]. Bioorg Med Chem, 2005, 13(8): 3017-3028.
[23]
ABRAHAM M H, IBRAHIM A, ZHAO Y, et al. A data base for partition of volatile organic compounds and drugs from blood/plasma/serum to brain, and an LFER analysis of the data [J]. Eur J Pharm Sci, 2006, 95(10): 2091-2100.
2021, Vol. 56 
