Computational protocol: Lipid Profiling following Intake of the Omega 3 Fatty Acid DHA Identifies the Peroxidized Metabolites F4-Neuroprostanes as the Best Predictors of Atherosclerosis Prevention

Similar protocols

Protocol publication

[…] Atherosclerotic lesions were assessed by measuring lipid deposit in the aortic sinus as previously described . Briefly, four sections of 10 µm thickness were harvested/slide and 28 slides/mouse were prepared and stained with Oil red O (Merck, Darmstadt, Germany) and counterstained with hematoxylin (Diapath, Martinengo, Italy). Each section was evaluated for Oil red O staining area under microscope. Image analysis was carried out using the ImageJ free software ( to quantify the cross-section surface areas of the lesions and the cross-section surface area of the aorta. The lesion area was calculated by dividing the surface of the lesion by the surface of the vessel and expressed as a percentage. This approach allows correcting for errors caused by oblique sections that may lead to overestimation of the surface area occupied by a lesion. [...] All data are presented as means ± SEM. Data following a Gaussian distribution (test of Kolmogorov & Smirnov) and having no significant difference between SD (Bartlett’s test) were analyzed by one-way ANOVA followed by a Tukey-Kramer post-hoc analysis using GraphPad InStat version 3.06 (GraphPad Software, San Diego California USA, The other data were analysed using a non-parametric test (Kruskal Wallis’s test). Differences were considered significant at p<0.05.The cumulative results were also subjected to a suite of multivariate analyses including hierarchical cluster analysis and PLS-DA. Data treatment prior to multivariate analysis included missing value imputation, and normality transformation, and autoscaled i.e. each variable was mean centred and scaled to unit variance . Where variable data sets were greater than 90% complete, missing values were computationally imputed using a probabilistic principal components analysis ,. Data was transformed to normality using the procedures of Box and Cox . Differences in means were determined using 2-tailed t-tests or Mann-Whitney U-tests if normality was not achieved. Data transformations, multivariate analyses and non-parametric statistics were performed using the Excel Add-In imDEV , which provides a graphical user interface to the R Statistical Computing Environment. Hierarchical clusters were based on a Minkowski distance matrix agglomerated using Ward’s minimum variance method. PLS-DA was performed using the SIMPLS algorithm with leave one out cross-validation, and predictive models were iteratively compared (n = 100) to those built with data randomly assigned to discriminate class data. Predictive models were constructed to either classify animals based on relative dietary n-3 content [discriminate classes: 0 = Control; 1 = DHA1; 10 = DHA2; 20 = DHA3] or atherosclerotic plaque area modeled as a continuous variable. Model variable selection with correlation, ANOVA, entropy minimization, and chi square filtering were compared between 10 independent training/test set selections. Correlation based filtering provided the model set with the lowest variance in the root mean squared error of prediction. […]

Pipeline specifications

Software tools ImageJ, imDEV
Applications Miscellaneous, Microscopic phenotype analysis
Organisms Mus musculus
Diseases Amino Acid Metabolism, Inborn Errors, Atherosclerosis
Chemicals Fatty Acids, Unsaturated, Fatty Acids, Omega-3