Computational Regression and Prediction Analysis on a Dataset of 52 Pyrrolo[2,3-b]Pyridines and Pyrimidines Rheumatoid Arthritis Inhibitors

Abstract

Objectives: To evaluate the dependency of physico-chemical properties of a set of BTK inhibitors on activity by linear regression analysis. Methods/Statistical Analysis: A multivariate regression analysis was implemented on 52 Pyrrolo[2,3-b]pyridines and pyrimidines which are reported as Bruton's Tyrosine Kinase (BTK) inhibitors to construct a regression model using complete data set in the relationship between dependent variable and independent variable was estimated using python based regression analysis. Leverages were calculated in order to exclude outlying data from analysis Findings: A linear regression analysis on a complete dataset of BTK inhibitors as dependent variables and few independent variables resulted in F-test: 4.87, r value: 0.737 and r2 value of 0.543, respectively. The dataset investigated for the existence of outliers resulted in 50 BTK inhibitors after excluding 4q and 4zd from the dataset which resulted in improved r2 of 0.701 with better statistics. Further 44 compound training set resulted in improved r and r2 coefficients such as r: 0.816 and r2: 0.666, respectively and applying on a 6 compound validation set which determines regression model reliability and significance. Application/Improvements: Application of regression model to screen novel compounds with decreased H-bond acceptors, logP and KAlpha2 followed by an increase in KAlpha3 and randic index would enhance inhibitory activity against BTK. 1. Introduction Rheumatoid Arthritis (RA) is an autoimmune disease categorized by auto antibody circulation in plasma, inflammation in synovial, destruction of cartilage and bone etc.1,2 Bruton's Tyrosine Kinase (BTK) belongs to Tec kinase member family of enzymes which are necessary for activation of B-cells.3 A designated function for BTK is found in activation of cells mediated by BCR and cell survival.4-6 Compounds which inhibit BTK are known to hinder B cell receptor signaling7, several such inhibitors are currently in clinical trials and the first BTK inhibitor Ibrutinib was approved for treating mantle cell lymphoma. 8 Efforts from medicinal chemistry, synthesis, crystallography has led to the discovery of some non-covalent BTK inhibitory compounds. Several BTK inhibitors such as Pyrrolo[2,3-b]pyridines9, Pyrrolo[2,3-d]pyrimidines10, thieno[3,2-c]pyridin-4-amines11, phenylpyridin-2(1H)- ones12, Imidazo[1,5-a]quinoxalines13, Purine derivatives14, CarbazoleCarboxamides15, PyridazinoneAnalogs16, Diaminopyrimidines17, imidazo[1,5‑a]pyrazines18 are reported. In this paper, linear regression analysis was implemented to study the dependency of physico-chemical properties on