TY - JOUR
T1 - The QSAR Study of Coumarin Derivatives as Anti Alzheimer Agents
TT - مطالعه کیوسار مشتقات کومارین به عنوان عوامل ضد بیماری آلزایمر
JF - RJMS
JO - RJMS
VL - 31
IS - 1
UR - http://rjms.iums.ac.ir/article-1-8389-en.html
Y1 - 2024
SP - 1
EP - 12
KW - Coumarins
KW - QSAR
KW - Anti Alzheimer activity
KW - Density functional theory
N2 - Background & Aims: Alzheimer's disease is one of the most common causes of loss of mental function, which is generally known as dementia (1) .Alzheimer's disease affects almost two percent of people in developing countries (2). The risk of developing the disease increases with age. Alzheimer's disease and its consequences are the Greatest Health Crisis of the 21st Century. Coumarin and its derivatives are one of the most abundant natural compounds in plants. This compound is a heterocyclic oxygen that is known as the parent chemical structure for a class of phytochemicals naturally found in several plant species. These compounds have been used as anticoagulant, antibacterial, anti-osteoporosis and antiasthmatic (6). Quantitative structure- activity relationship (QSAR) is widely used in drug design processes to improve the therapeutic indicators of designed compounds. QSAR models are mathematical equations written based on the relationship between the chemical structure of compounds and their biological activity (7). These equations use different descriptors; these descriptors are divided into different groups such as structural, geometric, spatial, quantum, and the like and are obtained by experimental and computational methods. Due to the time-consuming and costly experimental methods, computational methods are preferred (8). Density functional theory has been one of the most popular methods of solid-state physics since 1970. Though it was not until 1990, when the considered approximations, in theory, were revised, and a better model for exchange interactions was proposed, it was not considered an exact method in quantum chemistry. Methods: Due to the anti-Alzheimer's property of coumarin and its derivatives, In this research, 44 coumarin derivatives have been selected that were previously synthesized and their experimental activity was measured. All of compounds optimized and their quantum descriptors have been obtained using Gaussian software and computational DFTmethod and basis set 6-31G (d), other descriptors were also determined using Dragon software. At first, using IBMSPSS20 software and stepwise multiple linear regression method, the samples were randomly divided into training series (33 molecules) and prediction series (11 molecules). 75% of the samples were considered training and 25% of them as a test, and this ratio was repeated three times to get the best correlation. The training series was used to create a suitable model, and the prediction series was used to evaluate the model. The most important part of creating an efficient model is the selection of appropriate descriptors. After calculating different descriptors, they are selected as suitable descriptors to build the model. In this study, the stepwise method (SW) was used to select the most appropriate descriptors. In a stepwise method by examining all descriptors, the selection process continues until a model with a high correlation coefficient is obtained. Using the stepwise method from among 95 descriptors, six descriptors were selected as the most suitable and were modeled by the MLR method. After selecting the most suitable descriptors, the relationship between the selected descriptors and the activity of the drug compounds was established by a stepwise incremental method. To check the accuracy and validity of the model, internal and cross-validation methods were used. Internal validation is related to the statistical parameters of the training category, one of these parameters is R or the correlation coefficient. An acceptable equation has a correlation coefficient value greater than 0.6. The reported values for R2 are between 0 and 1. R2=0 means no correlation between the activity and the selected variable, and R2=1 means a very good correlation. Increasing the number of parameters in an equation R2 cannot show the correlation value correctly. For this reason, in an equation with several parameters, R2adj is a more accurate scale for expressing correlation. The closer this value is to one, the stronger the linear relationship between the dependent and independent variable. Cross-validation was used to predict the properties of the compounds that were not included in the new model. Results: The final evaluation of the model is done by external validation, which indicates how much predictive ability the QSAR model has for the compounds outside the model. For the QSAR equation R2= 0.89665 and R= 0.94691 were obtained. Also, MSE=0.51397 and RMSEP=0.81297 and Q2=0.83216 are a confirmation of the acceptability of the obtained model. The obtained equation shows that the activity of these compounds is related to the negative coefficients of RDF065m, nCbH, Jhetp, which means that by increasing the values of these descriptors, the amount of activity decreases.. On the other hand, the activity of these compounds depends on the positive coefficient of nR=Ct and SPH that means by increasing this value, the activity of these compounds also increases. By placing the effective descriptors, the anti-Alzheimer activity values of these compounds were predicted. The comparison of these values with the obtained experimental activities shows an acceptable correlation between the values of experimental and predicted activity. To investigate how much the increase or decrease of the selected descriptors affects the desired activity, a graph of the average effect of the descriptors has been drawn.The y-axis represents the standardized coefficients, and the x-axis represents the selected descriptor using the SW-MLR method. According to this chart, the Jhetp has the most negative , and the nR=Ct descriptor has the most positive effect. In this chart, the descriptors drawn above the chart mean that the amount of activity increases with increasing the value of the descriptor, and for those drawn at the bottom of the chart, the amount of activity decreases as they increase. The evaluation of the domain of application shows that there is no systematic error in the obtained model, and it also shows the efficiency of the model for predicting behavior in experimental methods. Conclusion: According to the examination of the results presented above, it can be generally concluded that according to the predicted activity values in Table 3, compounds 36 and 37 have the most activity and compounds 21 and 15 have the least anti-Alzheimer activity. The obtained model helps to design new drugs with more activity in the future based on the factors affecting the activity obtained from the QSAR equation. Since computational methods are faster and more accessible than experimental methods, it can be less predicted the necessary structures to design effective drugs.
M3 10.47176/rjms.31.44
ER -