Fitting a generalised linear model with linear and quadratic term

Fitting a generalised linear model with linear and quadratic terms for dose, and removing the PLX-4720 mouse highest dose until the quadratic term was not significant, also identified the linear part of the dose response, and the residuals were consistent with the method’s assumptions. The linear portion of the curve was used to compare the slopes of dose responses. A test for difference in slopes was investigated using an analysis of covariance model containing terms for dose, PM and a PM-by-dose interaction term. Where PM-by-dose was significant (p < 0.05), the difference in slopes was statistically significant. Occasionally, linear dose responses were parallel (PM-by-dose p ⩾ 0.05).

The PM samples CHIR-99021 cost were then compared for differences in overall magnitudes (mean responses). This was done by subjecting data pooled across doses to ANCOVA, with dose as a covariate and a term for PM as a fixed effect. Where the PM term was significant (p < 0.05), the difference in magnitudes was statistically significant. There were also some data-sets where a linear part of the dose response could not be established for one or both of the PM samples. In this case,

different PMs were compared at each common dose level using t-tests, two-sided at the 5% level of significance. For the MLA, Levene’s test (Levene, 1960) for equality of variances between the two PM samples was performed prior to the t-test and where this showed evidence of heterogeneity (p < 0.01) the data was rank transformed prior to analysis ( Conover and Iman, 1981). Levene’s test is used to test if samples have equal variances. Equal variances across samples is called homogeneity of variance. Some statistical tests, for example the t-test, assume that variances are equal across groups or samples. Levene’s test can be used to verify that assumption. For the Ames test and IVMNT, the data was Poisson

and binomially distributed respectively, thus standard parametric tests based on the assumption of normally distributed Dichloromethane dehalogenase data are not appropriate and the data were rank transformed prior to the t-test. Rank transformation procedures are ones in which the usual parametric approach is applied to the ranks of the data instead of the data themselves. In situations where the number of observations is low, non-parametric methods can be insensitive and in some cases it is not possible to obtain statistically significant differences at all. Therefore for these assays the analysis of rank transformed data is considered to be more appropriate. The combined statistical methods are summarised in Fig. 1. Historical data was reviewed to identify the most responsive PM treatment conditions for each assay. The most sensitive responses in the Ames test were obtained with TA98, TA100 and TA1537, and S9 metabolic activation.

This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>