Help interpreting residual vs fitted plot and normality (ANOVA on R) I'm carrying out a statistical analysis on R using ANOVA and am not sure if the data meets the assumptions of normality of residuals or homogeneity of variance. 
My data :

And my plots: 


Any help is really appreciated as I'm not sure how to interrupt these (as the res v fit does have a straight line, though diagonal, and it is fairly normal until the last few points).
 A: I both agree and disagree with @Bernhard, and with respect one doesn't have to be "very picky" to disagree partially. 
Agree: Sure, there is, given these data, a significant difference between control and mutant groups in general level. The fact that a Wilcoxon test supports analysis of variance on the original data underlines that this conclusion is robust to what summary statistic is the focus and quite what ideal conditions lie behind each test. (Many small discussions would be clarified at least a little if "ideal conditions" were the terminology, not "assumptions".) 
Disagree: The original analysis of variance is far from the best way to work with these data. On elementary biological as well as statistical grounds, a positive response (presumably thickness can never be negative or even zero) with a skewed distribution and mild heteroscedasticity and possible high outliers is better analysed on logarithmic scale. The implication is that geometric means are better ways to summarize the data than means.  One kind of machinery is a generalized linear model with logarithmic link, which is easily done in R. 
Here is one corresponding view of the data. Quantile plots are combined with median-quartile boxes. The other horizontal lines show geometric means. The data are not tamed by being viewed on logarithmic scale, but they are better behaved. I would name the units of measurement on my plot if I knew what they were. (I used Stata for the plot.) 

A: To put your data in a more R friendly format
control <- c(252.56, 283.36, 264.88, 523.4, 264.88, 247.19, 277.2, 237.16)
mutant <- c(150.92, 135.52, 215.6, 150.92, 147.84, 172.48, 138.6, 147.84)

A simple plot of the data makes clear, that there is clearly a distinction between the groups:
boxplot(control, mutant)

From looking at the plot everyone will believe you, that there is a statistical difference and nobody should get picky about a significant result:

Nothing in the world is ever perfectly normally distributed, few things are truly homoscedastic. Judging from your above plots I personally would accept your ANOVA results but someone very picky about it might start arguing. However, as I stated in my comment above, there is a simple means of clearing that once and for all:
> wilcox.test(control, mutant)

    Wilcoxon rank sum test with continuity correction

data:  control and mutant
W = 64, p-value = 0.0009148
alternative hypothesis: true location shift is not equal to 0

(BTW: Even if 523.4 in case 4 was a typo and was truly 253.4, that would not change the result of the rank sum statistic.)
