LiCABEDS II. Modeling of Ligand Selectivity for G-Protein-Coupled Cannabinoid Receptors

Abstract

The cannabinoid receptor subtype 2 (CB2) is a promising therapeutic target for blood cancer, pain relief, osteoporosis, and immune system disease. The recent withdrawal of Rimonabant, which targets at another closely related cannabinoid receptor (CB1), accentuates the importance of selectivity for the development of CB2 ligands in order to minimize their effects on the CB1 receptor. In our previous study, LiCABEDS (Ligand Classifier of Adaptively Boosting Ensemble Decision Stumps) was reported as a generic ligand classification algorithm for the prediction of categorical molecular properties. Here, we report extension of the application of LiCABEDS to the modeling of cannabinoid ligand selectivity with molecular fingerprints as descriptors. The performance of LiCABEDS was systematically compared with another popular classification algorithm, support vector machine (SVM), according to prediction precision and recall rate. In addition, the examination of LiCABEDS models revealed the difference in structure diversity of CB1 and CB2 selective ligands. The structure determination from data mining could be useful for the design of novel cannabinoid lead compounds. More importantly, the potential of LiCABEDS was demonstrated through successful identification of newly synthesized CB2 selective compounds.

Structural insight into the serotonin (5-HT) receptor family by molecular docking, molecular dynamics simulation and systems pharmacology analysis

“…First, we carried out the BBB prediction for Captagon. The BBB predictor was built by applying the SVM (Support Vector Machine) and LiCABEDS (Ligand Classifier of Adaptively Boosting Ensemble Decision Stumps) [69,70] algorithms to four types of fingerprints (MACCS, PubChem, Molprint 2D, and FP2) of 1593 reported compounds [71]. Unity was used in our original publication and was replaced with PubChem Structure Fingerprint in a subsequent upgrade.…”

An Algorithm to Identify Target-Selective Ligands – A Case Study of 5-HT7/5-HT1A Receptor Selectivity

“…For example, Ma et al used a selectivity index ≥ 10, which was selected based on the findings for the selective CB 1 /CB 2 cannabinoid ligand by J.W. Huffman [ 10 ]. However, Wang et al used a threshold for the selectivity index ≥ 3 for the kNN QSAR Classification model for 5-HT 1E /5-HT 1F receptor selectivity [ 5 ].…”

“…Other studies have described QSAR modeling to predict the ligand selectivity for serotonin 5-HT 1E /5-HT 1F [ 5 ] or dopamine D 2 /D 3 receptors [ 6 ] and for a panel of 45 different kinases [ 7 ]. Yet other investigations used machine learning (ML) methods to construct selectivity prediction models, e.g., ML based on neural networks to generate structure-selectivity relationship models [ 8 ], the binary classification SVM (Support Vector Machines) algorithm to solve multiclass predictions and compound ranking to distinguish between selective, active but non-selective, and inactive compounds [ 9 ], and the LiCABEDS (Ligand Classifier of Adaptively Boosting Ensemble Decision Stumps) algorithm to model cannabinoid CB 1 /CB 2 selectivity [ 10 ].…”

Virus-CKB: an integrated bioinformatics platform and analysis resource for COVID-19 research

“…Our established [ 39 , 40 ] blood−brain barrier (BBB) predictor [ 20–22 ] was integrated into Virus-CKB, in which BBB predictor will predict whether a query compound can move across the BBB to the central nervous system (CNS).…”