Figure 8

Random forest analysis accurately discriminates heat exposure from unheated controls at 48 hours. The top 30 metabolites were identified in OOB selection to have 100% predictive power by random forest analysis at 48 hours. The inset represents the top-scoring molecule (5-methyl-2’deoxycytidne). Colored circles indicate biochemicals that were also significantly up (red) or down (green) regulated, p < 0.05, 2-way ANOVA with contrasts. One biochemical was not significantly altered by ANOVA analysis (yellow). Biochemicals marked by an asterisk indicate likely identifications based on MS/MS fragmentation and other chemical properties.