Chapter 17 Protein.df Revisited

In the last exercises, including the previous chapter using tidy data, we worked on the protein.df data frame as a long list, not as actual tidy data. Let’s take a look at how we can maximize the use of our tidyverse functions.

17.1 Import and Clean Data

Exercise 17.1 (Import data) Fill in the blanks to:

• Import in “Protein.txt”,
• Convert it to a tibble,
• Remove contaminants, and
• Assign it to the object protein.df.

Use this as a template:

read.delim("___", stringsAsFactors = FALSE) %>% 
  as_tibble() %>% 
  filter(___ != ___) -> protein.df

17.2 Process Intensities

Exercise 17.2 Using the dplyr functions, fill in the blanks to:

• Transform all Intensity.* columns to log10. (mutate_at())
• Do additions of the H+M and M+L columns as we have done previously, save the new columns as H.M, M.L (mutate())
• Select only the Uniprot, H.M and M.L. (select())
• Make a tidy data set with three columns Uniprot, Ratio and Intensity (gather())
• Assign this to an object called onlyInt

Use this as a template:

protein.df %>% 
  mutate_at(vars(starts_with(___)), ___) %>%    # Calculate log10
  mutate(H.M = ___ + ___,                       # Add log10s and rename columns
         M.L = ___ + ___) %>% 
  select(___, ___, ___) %>%                     # Take columns of interest
  gather(___, ___, -___) -> ___                 # gather and save

17.3 Process Ratios

Exercise 17.3 Using the dplyr functions, fill in the blanks to:

• Using the dplyr functions, select Uniprot and all columns that begin with Rat, but that do not end in Sig. (select())
• Make a tidy data set with three columns Uniprot, Ratio and Expression (gather())
• Remove all observations where Ratio is Ratio.H.L. (filter())
• Rename the levels in Ratio to be M.L and H.M. (mutate() and recode_factor())
• Group according to Ratio.
• Apply a transformation to log2 transform all Expression values and then shift all values to be centered on zero. (mutate()).
• Assign this to onlyRatios.

Use this as a template:

protein.df %>% 
  select(Uniprot, starts_with(___), -ends_with(___)) %>%        # Calculate log2
  gather(___, ___, -___) %>%                                    # Gather
  filter(___ != ___) %>%                                        # Remove uninteresting H.L Ratio
  mutate(Ratio = recode_factor(___,                             # Relabel ratios to match the Int data frame
                               `___` = ___,
                               `___` = ___)) %>% 
  group_by(___) %>%                                             # group according to ratios (2 groups)
  mutate(Expression = log2(___),                                # log2 transform
         Expression = ___ - mean(___, na.rm = T)) -> ___       # Apply shift

17.4 Process Significance Values and Merge

Exercise 17.4 Using the dplyr functions, fill in the blanks to:

• Select Uniprot and all columns that end in Sig. (select())
• Make a tidy data set with three columns Uniprot, Ratio and Significance (gather())
• Remove all observations where Ratio is Ratio.H.L. (filter())
• Rename the levels in Ratio to be M.L and H.M. (mutate() and recode_factor())
• In the same function, make a new variable SigCat that cuts up the Significance variable into groups according to c(-Inf, 1e-11, 1e-4, 0.05, Inf) and labels them as c("<1e-11", "<0.0001", "<0.05", "NS").
• Merge all this with the onlyRatios data frame (full_join())
• Merge all this with the onlyInt data frame (full_join())
• Remove any incomplete observations (i.e. with an NA anywhere) and where Uniprot is empty. (filter() and complete.cases(.))
• Arrange in descending order of Significance. (arrange() and desc())
• Assin to the object allData

Use this as a template:

protein.df %>% 
  select(___, ends_with(___)) %>%                              # Take columns of interest
  gather(___, ___, -___) %>%                                   # Gather
  filter(___ != ___) %>%                                       # Remove uninteresting H.L Ratio
  mutate(Ratio = recode_factor(___,                            # Relabel ratios to match the Int data frame
                               `___` = ___,
                               `___` = ___),
         SigCat = cut(___,                                     # Make colour labels for sig values
                      c(-Inf, 1e-11, 1e-4, 0.05, Inf),
                      c("<1e-11", "<0.0001", "<0.05", "NS"))) %>%
  full_join(___) %>%                                           # Merge with the log2 ratios
  full_join(___) %>%                                           # Merge with the Intensities
  filter(complete.cases(.), ___ != ___) %>%                    # Take only observations that have complete data and non-empty Uniprot
  arrange(___(___)) -> ___                                     # Order according to sig so that low sig are plotted first

17.5 Make a plot

Fill in the details to produce the following plot:

ggplot(______, aes(______, ______, col = ______)) +
  geom_point(alpha = 0.5, shape = 16) +
  scale_colour_manual(limits = c("<1e-11", "<0.0001", "<0.05", "NS"),
                      values = c("red", "orange", "blue", "grey30")) +
  facet_grid(. ~ ______) +
  labs(x = "Log2 Ratios", y = "Intensity", col = "p-value")