tabviz applies styles in layers, from general to specific. Understanding this cascade helps you achieve precise control over your visualization’s appearance.

The Cascade

flowchart TB
    A["Theme defaults"] --> B["Row styling"]
    B --> C["Cell styling"]
    C --> D["Marker styling"]
    D --> E["Final appearance"]

    style A fill:#e5e7eb
    style B fill:#93c5fd
    style C fill:#6ee7b7
    style D fill:#fbbf24
    style E fill:#4f46e5,color:#fff

Later styles override earlier ones. A cell style beats a row style, which beats a theme default.

Layer 1: Theme Defaults

Themes establish the baseline appearance for all elements:

Code 
# Theme sets default colors, fonts, spacing
tabviz(data,
  columns = list(viz_forest(point = "hr", lower = "lo", upper = "hi",
                           scale = "log", null_value = 1)),
  theme = web_theme_jama()  # Sets primary color, font, spacing, etc.
)

Theme defaults include: - Colors: Primary, accent, text, background - Typography: Font family, sizes, weights - Spacing: Row height, padding, margins - Shapes: Point size, line width

Modifying Theme Defaults

Code 
web_theme_jama() |>
  set_colors(primary = "#2563eb") |>
  set_typography(font_size_base = "0.9rem") |>
  set_spacing(row_height = 28)

Layer 2: Row Styling

Row styling applies to entire rows, overriding theme defaults:

Code 
meta_data <- data.frame(
  study = c("Primary Endpoint", "  Subgroup A", "  Subgroup B", "Summary"),
  hr = c(NA, 0.72, 0.85, 0.78),
  lo = c(NA, 0.55, 0.70, 0.65),
  hi = c(NA, 0.95, 1.03, 0.93),
  rtype = c("header", "data", "data", "summary"),
  rbold = c(TRUE, FALSE, FALSE, TRUE)
)

tabviz(meta_data,
  label = "study",
  columns = list(viz_forest(point = "hr", lower = "lo", upper = "hi",
                           scale = "log", null_value = 1)),
  row_type = "rtype",
  row_bold = "rbold"
)

Row styling options:

Argument Effect
row_type Controls row behavior (header, data, summary, spacer)
row_bold Bold text
row_italic Italic text
row_color Text color
row_bg Background color
row_indent Indentation level
row_emphasis Theme-aware emphasis
row_muted Theme-aware de-emphasis
row_accent Theme accent color

Layer 3: Cell Styling

Cell styling targets specific columns, overriding row styles:

Code 
tabviz(data,
  label = "study",
  columns = list(
    col_pvalue("pval",
      bold = ~ .x < 0.05,        # Bold if p < 0.05
      color = ~ ifelse(.x < 0.05, "#16a34a", "#64748b")
    ),
    col_numeric("hr",
      color = ~ ifelse(.x < 1, "#16a34a", "#dc2626")
    ),
    viz_forest(point = "hr", lower = "lo", upper = "hi",
               scale = "log", null_value = 1)
  )
)

Cell styling uses .x to reference the column’s own values:

# In cell context, .x = the column's value for that row
col_numeric("hr", color = ~ ifelse(.x < 1, "green", "red"))

# Equivalent to:
# for each row: if (row$hr < 1) "green" else "red"

set_column_style()

Apply styling to any column after creation:

Code 
tabviz(data, label = "study", columns = list(...)) |>
  set_column_style("hr",
    bold = ~ .x < 1,
    color = ~ case_when(
      .x < 0.8 ~ "#16a34a",
      .x < 1.0 ~ "#86efac",
      TRUE ~ "#64748b"
    )
  )

Layer 4: Marker Styling

Marker styling controls the forest plot points and lines, overriding all other layers:

Code 
library(dplyr)

marker_data <- data.frame(
  study = c("Trial A", "Trial B", "Trial C", "Trial D"),
  hr = c(0.72, 0.95, 0.68, 1.15),
  lo = c(0.55, 0.78, 0.52, 0.90),
  hi = c(0.94, 1.15, 0.89, 1.47),
  design = c("RCT", "RCT", "Observational", "Observational")
) |>
  mutate(
    marker_color = case_when(
      hi < 1 ~ "#16a34a",
      lo > 1 ~ "#dc2626",
      TRUE ~ "#64748b"
    ),
    marker_shape = ifelse(design == "RCT", "circle", "square")
  )

tabviz(marker_data,
  label = "study",
  columns = list(viz_forest(point = "hr", lower = "lo", upper = "hi",
                           scale = "log", null_value = 1)),
  marker_color = "marker_color",
  marker_shape = "marker_shape"
)

Marker styling options:

Argument Effect
marker_color Fill color
marker_shape Shape (circle, square, diamond, triangle)
marker_opacity Transparency (0-1)
marker_size Size multiplier

Cascade Examples

Example 1: Theme + Row Override

Code 
tabviz(data,
  label = "study",
  columns = list(viz_forest(point = "hr", lower = "lo", upper = "hi",
                           scale = "log", null_value = 1)),
  theme = web_theme_jama(),           # Sets default text color
  row_color = ~ ifelse(pval < 0.05, "#16a34a", NULL)  # Overrides for significant rows
)
  • Theme sets default text color (black)
  • Row styling overrides to green for significant rows
  • Non-significant rows keep theme default

Example 2: Row + Cell Override

Code 
tabviz(data,
  label = "study",
  columns = list(
    col_numeric("hr", color = ~ ifelse(.x < 1, "green", "red"))  # Cell override
  ),
  row_color = "#333333"  # Row default
)
  • Row styling sets all text to dark gray
  • Cell styling overrides HR column to green/red based on value
  • Other columns keep the row style

Example 3: Full Cascade

Code 
tabviz(data,
  label = "study",
  columns = list(
    col_pvalue("pval", bold = ~ .x < 0.05),
    viz_forest(point = "hr", lower = "lo", upper = "hi",
               scale = "log", null_value = 1)
  ),
  theme = web_theme_jama(),                    # Layer 1: Theme
  row_bold = "is_summary",                     # Layer 2: Row
  # col_pvalue bold is                         # Layer 3: Cell
  marker_color = ~ ifelse(sig, "green", "gray") # Layer 4: Marker
) |>
  set_colors(primary = "#2563eb")              # Modifies Layer 1

Debugging Styles

When styles don’t appear as expected:

  1. Check layer order: Is a later layer overriding?
  2. Check NA values: NA styling values fall through to the next layer
  3. Inspect formula results: Print data |> mutate(result = ~ your_formula) to see evaluated values
  4. Simplify: Remove styling layers one at a time to find conflicts
Code 
# Debug a formula expression
library(dplyr)
data |>
  mutate(
    debug_bold = pval < 0.05,  # See what row_bold = ~ pval < 0.05 produces
    debug_color = ifelse(upper < 1, "green", "gray")
  ) |>
  select(study, pval, upper, debug_bold, debug_color)