Remaining Analyses Part 15

The growth data wasn’t very interesting

Well, I guess it’s interesting. But just not for my proteomics story. In this R script, I looked at differences in percent growth by site.

growth-boxplot

Figure 1. Differences in percent growth by site.

Willapa Bay had the least variation in percent growth for just my samples. The ANOVA was significant, but a Tukey test revealed that the only significant pairwise differences were FB-CI (p = 0.0313908) and SK-FB (p = 0.0102010).

To see if the growth data could explain any differences in peptide abundance, I regressed peptide abundance against percent growth. I generated 37 scatterplots, which can be found in this folder. None of the R-squared values were any different from zero, so I didn’t put them into a new table.

My conclusion is that the growth data isn’t going to be a part of my paper’s story. One less thing to worry about!

Written on December 19, 2017