Modeling of Inhibition of Tetrahymena pyriformis growth by Aliphatic Alcohols and Amines pollution of l’ environmental

Abstract

To assess the relative toxicity of a mixed series of 21(linear and branched-chain) alcohols and 9 normal aliphatic amines in terms of the 50% inhibitory growth concentration (IGC50) of Tetrahymena Pyriformis, a Quantitative Modeling study know as a Structure-Activity/property/Toxicity Relationship (QSAR/QSPR/QSTR) was conducted (20 training,10 tests). The used least squares LS method has been using MINITAB 16 Software and nom-parametric estimation (least absolute deviation LAD) (robust regression method) has been using Calculation Programs by MATLAB Software. The applied simple linear regression approach is based on theoretical H4p (GETAWAY descriptor) molecular descriptor from DRAGON software The performance of regression is better if the distribution of errors has normal, in this case we use the least squares LS method for statistical analysis. When the data does not have a natural assumption, we move to another method of analysis that is more robust and more frequent for the presence of the points of articulation, which is the least absolut deviation method (LAD). The findings of statistical analysis for the chosen model (QSAR) using simple linear Regression using the least Squares Method were R^2=97.39% ,Q^2=96.69% ,Q_bOOT^2=96.24%,Q_EXT^2=93.91% ,R_adj^2=97.24%, S=0.248 Anderson Darling (AD) test =1.57 >0.752 , symmetry coefficient (ou skeweness) (sk= 2.14>0 ) , flatness coefficient (Kurtosis) (ku=5.75>3) and Jarque and Bera Test (JB= 42.84>5.9942. the results did not follow the normal law (unnormal). The coefficient of determination and the value of standard deviation are both highly sensitive to the presence of aberrant compounds(abnormales), as the R^2value moved from 87, 96 % to 94.18 %, which increased by a value of 6.22% and the value of standard deviation (S) moved from 0.399 to 0.303, it increased by a value of 25 % after removing aberrant compound (abnormalie) are interpreted as better adjustment and they are positively. After removing the aberrant compound, we did not see any change in the lines coefficients, indicatting that the function’s graph is stable, demonstrating the LAD method and increased power, which are unaffected by the presence of aberrant compounds Consequently, which means that the model of one descriptor selected is good and statistically strong, Three influential compounds detected ((one compound of training, two compounds of Test) and important the model and absence of studied sample aberrants compounds.