secondary-struct-analysis.R

library(tidyverse)

data <- read_csv('~/Downloads/data-precs.csv')

counted <- data |> 
  filter(!is.na(res_id)) |> 
  select(pdb_id, name, secondary) |> 
  group_by(pdb_id) |> 
  mutate(
    previous = lag(secondary)
  ) |> 
  filter(!is.na(previous) & !is.na(secondary)) |> 
  # filter(previous != secondary) |> 
  mutate(
    previous = case_when(
      previous == 'H' ~ 'alpha helix',
      previous == 'B' ~ 'beta bridge',
      previous == 'E' ~ 'beta ladder',
      previous == 'G' ~ '310 helix',
      previous == 'I' ~ 'pi helix',
      previous == 'T' ~ 'H-bonded turn',
      previous == 'S' ~ 'bend',
      previous == '-' ~ 'loop'
    ),
    secondary = case_when(
      secondary == 'H' ~ 'alpha helix',
      secondary == 'B' ~ 'beta bridge',
      secondary == 'E' ~ 'beta ladder',
      secondary == 'G' ~ '310 helix',
      secondary == 'I' ~ 'pi helix',
      secondary == 'T' ~ 'H-bonded turn',
      secondary == 'S' ~ 'bend',
      secondary == '-' ~ 'loop'
    )
  ) |> 
  mutate(
    transition = paste0(
      previous, ' to ', secondary 
    )
  ) |> 
  group_by(secondary, previous) |> 
  count(name) |> 
  mutate(prop = n / sum(n))

counted |> 
  rename('From' = previous, 'To' = secondary) |> 
  mutate(From = paste('From', From), To = paste('to', To)) |> 
  ggplot(aes(x = name, y = prop)) +
  theme_minimal() +
  geom_col() +
  facet_grid(cols = vars(From), rows = vars(To)) +
  scale_y_continuous(breaks = c(0, 0.5, 1)) +
  scale_x_discrete(guide = guide_axis(n.dodge = 2)) +
  labs(
    x = 'Amino acid name',
    y = 'Proportion of transition',
    caption = 'Data from Rosenberg, Marx, and Bronstein, 2022. https://doi.org/10.7910/DVN/5P81D4'
  )
ggsave('secondary-structure-transitions.png', width = 12, height = 12, bg = 'white')