ggplot x variable according to vector

Question

I have a large dataset (14295,58). Each column is a different element from the periodic table (e.g. Fe, Ca, Zr) and the rows are arranged according to depth (in mm); the last column is the depth value. I am trying to make a code that can be customized to a given group of elements over a given depth interval but I don't want to have to go through and change a bunch of lines of code everytime I look at a different subset. So far I have created a dataframe called Section:

Section <- df[50:100,]

and a vector called Elements:

Elements <- c("Fe", "Ca", "Zr")

I can subsample the Section data frame by:

Section %>%
select(., Elements, depth)

but now I want to plot this with ggplot and I can't figure out how to call the Elements vector to the x-variable. I tried:

Section %>%
select(., Elements, depth) %>%
ggplot() +
geom_path (aes(Elements, depth))

but the arguments don't have the same length. How can I plot the selected elements from the Elements vector?

Answer 1

I think your problem is actually that your data is not formatted in the most useful way (wide vs. long), so you aren't actually giving ggplot what you think you are. If you give it a vector as an aesthetic (Elements here), it will try its best to plot it. In this case, it will do it if the length matches by just matching up values in depth to things in Elements. So this works:

# Toy Data
df <- data.frame(O = 1:3,
                 Fe = 2:4,
                 Ca = 3:5,
                 Zr = 4:6,
                 depth = 5:7)
Elements <- c('Fe', 'Ca', 'Zr')
ggplot(df) +
    geom_point(aes(x=Elements, y=depth))

But it just matches the first depth to 'Fe', the second depth to 'Ca', etc. I don't think that's what you are hoping to have happen.

Long vs Wide Data

You have separate columns for every all these elements, but do they actually represent different things? You are probably better off re-formatting your data so that all these "element" columns get collapsed into key-value pairs using tidyr:

# Wide:
df
  O Fe Ca Zr depth
1 1  2  3  4     5
2 2  3  4  5     6
3 3  4  5  6     7

# Long
library(tidyr)
longDf <- tidyr::gather(df, element, amount, -depth)
longDf
       depth element amount
1      5       O      1
2      6       O      2
3      7       O      3
4      5      Fe      2
5      6      Fe      3
6      7      Fe      4
7      5      Ca      3
8      6      Ca      4
9      7      Ca      5
10     5      Zr      4
11     6      Zr      5
12     7      Zr      6

Now you can get the elements you want using dplyr's filter (which is also probably a better option for subsetting by depth) and use the new element column as the x coordinate for plotting:

longDf %>% 
    filter(element %in% Elements) %>%
    ggplot() +
        geom_path(aes(x=element, y=depth))

I'm not sure what you're expecting the graph to look like, but that should get you started.