General
Split-Correlation Heatmap
Standard correlation heatmaps are symmetric, displaying identical
information in both triangles. When comparing two
groups, gg_splitcorr() uses this redundancy by
displaying one group’s correlations in the upper triangle and the other
group’s in the lower triangle. The function requires a data frame with
numeric variables and a binary splitting
variable. It computes pairwise correlations for each group,
adjusts p-values for multiple testing within each, and labels only
statistically significant correlations with their values.
# Compare correlations between V-shaped vs straight engines
gg_splitcorr(
data = mtcars,
split = "vs",
prefix = "Engine Type: "
)
Currently there are two styles to choose from, tile (default) and point.
# Alternative style
gg_splitcorr(
data = mtcars,
split = "vs",
prefix = "Engine Type: ",
style = "point"
)
See ?gg_splitcorr for correlation methods, p-value
adjustment options, color schemes, and style customization.
Rank Shift Plots
A frequently encountered task is assessing how the ranking of a set
of samples changes across two conditions. gg_rankshift()
creates a three-panel visualization: side panels
display ranked distributions for each condition, while
the center panel connects corresponding samples with lines colored by
rank change direction. The function requires data with
sample identifiers, a grouping variable containing exactly two
levels, and a numeric value for ranking. Ranks are calculated
using summary statistics (mean or median) controlled by
stat_summary.
# Synthetic data - bacterial strain growth rates
growth_data <- data.frame(
strain = rep(paste0("Strain", 1:13), each = 6),
condition = rep(c("Control", "Treated"), each = 3, times = 13),
growth_rate = c(
rnorm(39, mean = 0.85, sd = 0.12), # Control
rnorm(39, mean = 0.45, sd = 0.10) # Treated
)
)
gg_rankshift(
data = growth_data,
id = "strain",
group = "condition",
value = "growth_rate"
)
# Alternative style & minor customizations
gg_rankshift(
data = growth_data,
id = "strain",
group = "condition",
value = "growth_rate",
style = "bar",
fill = c("#e41a1c", "#377eb8"),
rank_change_colors = c(
increase = "#1b9e77",
decrease = "#d95f02",
no_change = "#7570b3"
),
panel_ratio = 0.65,
point_size = 2.5,
line_width = 1,
decreasing = TRUE
)
See ?gg_rankshift for options to customize colors,
adjust panel widths, and control point display.
Confusion/Contingency Tables
gg_conf() creates bubble plots where bubble size
represents frequency counts for each unique combination
of two categorical variables. The function
automatically computes these counts from the raw data using
table(). Results can also be facetted when additional
categorical variables are supplied.
data(mtcars)
mtcars$horsepower <-
cut(mtcars$hp, breaks = 5,
labels = c("Very Low", "Low", "Medium", "High", "Very High"))
mtcars$`miles per gallon` <-
cut(mtcars$mpg, breaks = 5,
labels = c("Very Low", "Low", "Medium", "High", "Very High"))
gg_conf(data = mtcars, x = "horsepower", y = "miles per gallon")
# Custom styling
gg_conf(data = mtcars, x = "horsepower", y = "miles per gallon",
fill = "lightcoral", point_size_range = c(5, 20),
show_grid = FALSE)
# With faceting by "vs" column
gg_conf(data = mtcars, x = "horsepower", y = "miles per gallon",
fill = "lightcoral", point_size_range = c(5, 20),
facet_x = "vs")
See ?gg_conf for customization of colors, sizes, grid
display, and faceting options.
Criteria Heatmaps
gg_criteria() displays samples against multiple
criteria as a heatmap. Optional barplots can
be added to show continuous metrics. As the vertical alignment between
heatmap and barplot panels depends on the exported figure dimensions,
the function provides recommended sizes based on the
number of candidates and criteria.
# Create example data
# Example: Gene prioritization criteria
gene_data <- data.frame(
gene = c("BRCA1", "TP53", "EGFR", "KRAS", "MYC",
"PTEN", "APC", "CDKN2A", "RB1", "VHL"),
`Missense Variant_crit` = c("Yes", "Yes", "Yes", NA, "Yes",
"Yes", NA, "Yes", NA, "Yes"),
`eQTL_crit` = c("Yes", "Yes", NA, "Yes", "Yes",
"Yes", "Yes", "Yes", "Yes", NA),
`pQTL_crit` = c("Yes", NA, "Yes", "Yes", NA,
"Yes", "Yes", NA, "Yes", "Yes"),
`GWAS Hit_crit` = c("Yes", "Yes", "Yes", "Yes", NA,
"Yes", "Yes", "Yes", NA, NA),
`Loss of Function_crit` = c(NA, "Yes", NA, NA, "Yes",
"Yes", "Yes", NA, "Yes", NA),
`High Conservation_crit` = c("Yes", "Yes", "Yes", "Yes", "Yes",
"Yes", "Yes", "Yes", "Yes", "Yes"),
`mRNA DE_crit` = c("Yes", NA, "Yes", NA, NA,
"Yes", "Yes", "Yes", NA, "Yes"),
`Prot DE_crit` = c(NA, "Yes", NA, NA, NA,
"Yes", NA, NA, NA, "Yes"),
check.names = FALSE
)
# Calculate total criteria met
crit_cols <- grep("_crit$", names(gene_data), value = TRUE)
gene_data$`Total` <- rowSums(gene_data[crit_cols] == "Yes", na.rm = TRUE)
head(gene_data)
#> gene Missense Variant_crit eQTL_crit pQTL_crit GWAS Hit_crit
#> 1 BRCA1 Yes Yes Yes Yes
#> 2 TP53 Yes Yes <NA> Yes
#> 3 EGFR Yes <NA> Yes Yes
#> 4 KRAS <NA> Yes Yes Yes
#> 5 MYC Yes Yes <NA> <NA>
#> 6 PTEN Yes Yes Yes Yes
#> Loss of Function_crit High Conservation_crit mRNA DE_crit Prot DE_crit Total
#> 1 <NA> Yes Yes <NA> 6
#> 2 Yes Yes <NA> Yes 6
#> 3 <NA> Yes Yes <NA> 5
#> 4 <NA> Yes <NA> <NA> 4
#> 5 Yes Yes <NA> <NA> 4
#> 6 Yes Yes Yes Yes 8# Base criteria plot
gg_criteria(
data = gene_data,
id = "gene",
criteria = "_crit$",
show_text = FALSE
)
#> Recommended dimensions: 5.2 x 5.5 inches
# With added barplot
gg_criteria(
data = gene_data,
id = "gene",
criteria = "_crit$",
bar_column = "Total",
show_text = FALSE,
tile_fill = c(Yes = "#A6CEE3", No = "white"),
bar_fill = "#A6CEE3",
panel_ratio = 0.7
)
#> Recommended dimensions: 7.2 x 5.5 inches
Criterion values can be displayed as text labels on tiles.
# Example: VHH Variant Analysis
# Define amino acid chemistry colors
aa_colors <- c(
"D" = "#E60A0A", "E" = "#E60A0A", # Acidic (red)
"K" = "#145AFF", "R" = "#145AFF", # Basic (blue)
"H" = "#8282D2", # Histidine (purple)
"S" = "#FA9600", "T" = "#FA9600", # Polar uncharged (orange)
"N" = "#00DCDC", "Q" = "#00DCDC", # Polar amides (cyan)
"C" = "#E6E600", # Cysteine (yellow)
"G" = "#EBEBEB", # Glycine (light gray)
"P" = "#DC9682", # Proline (tan)
"A" = "#C8C8C8", # Alanine (gray)
"V" = "#0F820F", "I" = "#0F820F", # Hydrophobic (green)
"L" = "#0F820F", "M" = "#0F820F",
"F" = "#3232AA", "W" = "#B45AB4", # Aromatic (dark blue/purple)
"Y" = "#3232AA"
)
vhh_variants <- data.frame(
variant = c("WT", "Mut1", "Mut2", "Mut3", "Mut4", "Mut7", "Mut5",
"Mut6", "Mut8", "Mut9", "Mut10", "Mut11"),
Q5_mut = c(NA, "H", NA, NA, NA, NA, "H", "H", "H", "D", NA, "H"),
S55_mut = c(NA, NA, "P", NA, NA, "P", "P", NA, "P", "P", NA, NA),
N73_mut = c(NA, NA, NA, "E", NA, NA, NA, "E", "E", NA, "E", NA),
K80_mut = c(NA, "L", NA, NA, "S", "V", NA, NA, NA, "L", "S", NA),
F99_mut = c(NA, NA, "L", NA, NA, NA, NA, NA, NA, "W", NA, "W"),
KD_nM = c(45, 18, 5.2, 38, 42, 20, 3.8, 15, 3.2, 4.5, 40, 22),
yield_mg_L = c(12, 11.8, 10, 13, 11, 10, 10, 12, 10, 7.8, 12.5, 8.5),
Tm_C = c(68.5, 67.8, 68, 72.3, 35, 66, 67.5, 70, 70.5, 72, 38, 74)
)# Create plot
gg_criteria(
data = vhh_variants,
id = "variant",
criteria = "_mut$",
tile_fill = aa_colors,
bar_column = c("KD_nM", "yield_mg_L", "Tm_C"),
panel_ratio = 2,
tile_width = 0.70,
tile_height = 0.70,
show_text = TRUE,
border_color = "grey40",
border_width = 0.4,
text_size = 10,
show_legend = FALSE
)
#> Recommended dimensions: 10.0 x 5.9 inches
See ?gg_criteria for tile customization, border styling,
and multi-barplot options.
