In Middleton-Lab/MuscleTernary: Plot Muscle Resultants in Ternary Space

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

To create a ternary plot from origin and insertion data, first we need data. Attachment centroid data are available for the AL_008 specimen of Alligator mississippiensis (although similar data can be extracted directly from stl files; see the "Working with stl files" article).

library(MuscleTernary)

AL008 <- read_csv(system.file("extdata",
                              "AL_008_data.csv",
                              package = "MuscleTernary"),
                  show_col_types = FALSE) |>
  dplyr::select(-side)
AL008

We need to convert the x, y, and z origin and insertion columns into relative ternary coordinates using the coords_to_ternary() function. Here we also use the grouping option to return the mean for each muscle, because the original data have both left and right (two rows per muscle; note that the side column was dropped above).

ternary_coords <- coords_to_ternary(coords = AL008, grouping = c("muscle"))
ternary_coords

The x, y, and z columns sum to 100 for each row: these represent the percent contributions of each orthogonal direction to the muscle's overall orientation.

The force column contains the estimated force for each muscle. Force can be estimated from PCSA, which can be derived from attachment areas (see the "Working with stl files" vignette).

The simplest ternary plot simply maps the x, y, and z columns and colors points by muscle:

ggtern(data = ternary_coords,
       aes(x = x, y = y, z = z,
           color = muscle)) +
  geom_point()

We can make this a more ready-to-use plot by recoloring using the standard palette for muscle colors. muscle_color_map() is a function that returns the mapping of muscle to its hex code.

muscle_color_map

ggtern(data = ternary_coords,
       aes(x = x, y = y, z = z,
           color = muscle)) +
  geom_point() +
  muscle_color_map()

And finally we can add labels, change the theme, fix up the legend, and update the scale, resizing the points according to force. Points don't have to be scaled to force, any proxy for body size (SVL, head width, etc. would work as well).

ggtern(data = ternary_coords,
       aes(x = x, y = y, z = z,
           color = muscle,
           size = force)) +
  geom_point() +
  muscle_color_map() +
  labs(x       = "ML",
       xarrow  = "Mediolateral",
       y       = "DV",
       yarrow  = "Dorsoventral",
       z       = "RC",
       zarrow  = "Rostrocaudal") +
  theme_bw(base_size = 16) +
  theme_showarrows() +
  scale_size_continuous(range = c(5, 15), name = "Force (N)") +
  guides(colour = guide_legend(override.aes = list(size = 6),
                               ncol = 2, byrow = TRUE))