paper-viz · House style

示例 / Examples

点卡片打开画廊大图（含代码与数据）；卡片下方的 GitHub 链接打开该图的文件夹。

Click a card to open it in the gallery; the GitHub link below it opens the figure folder.

分组散点与回归（house 风格）Scatter with group fits (house style) correlation/house-scatter · GitHub

带置信带的折线图（house 风格）Line chart with CI ribbons (house style) correlation/house-line · GitHub

带原始点与 P 值的柱图（house 风格）Bar chart with points and P values (house style) comparison/house-bar · GitHub

雨云图（house 风格）Raincloud plot (house style) distribution/house-raincloud · GitHub

分组密度曲线与均值（house 风格）Density curves with group means (house style) distribution/house-density · GitHub

效应量森林图（house 风格）Forest plot of effects (house style) comparison/house-forest · GitHub

带数值的发散色热图（house 风格）Annotated diverging heatmap (house style) heatmap/house-heatmap · GitHub

直标环图（house 风格）Donut chart with direct labels (house style) composition/house-donut · GitHub

小多图与面板内趋势（house 风格）Small multiples with in-panel trends (house style) layout/house-small-multiples · GitHub

用法 / Usage

1. source("../../../styles/r/theme_house.R")
2. theme_house() + pv_palette("house")
3. pv_save_house(p, "figure", cells = "2x1")

中文：theme_house.R 会顺带加载 theme_viz.R。theme_house() 默认 7/8 pt；非用图例不可时用 theme_house(legend = "inside")，放框内右上。尺寸只用网格格数 cells = c(宽, 高)（或 "2x1" 这样的字符串），不在网格上的尺寸直接报错，旧写法 width = "single", height_mm = 76 现在会报错。只有单个面板存成 2×2（89 × 89 mm）时 pv_save_house() 自动 +2 pt（9/10 pt），bump = 0 关掉。多张图拼成一页用 pv_house_mosaic()，见下方第二段代码，导出时 cells 与拼页相同。pv_save_house() 会在输出目录写 <file>.pdf、<file>.png 和 preview.png，覆盖同目录里同名的 preview.png；试跑请在临时目录里进行。

English: theme_house.R also loads theme_viz.R. theme_house() defaults to 7/8 pt; when a legend is unavoidable, use theme_house(legend = "inside") to place it top-right inside the frame. Sizes are given only as grid cells, cells = c(width, height) (or a string such as "2x1"); sizes off the grid are errors, so the old width = "single", height_mm = 76 now fails. pv_save_house() adds +2 pt (9/10 pt) only to a single panel saved at 2×2 (89 × 89 mm); bump = 0 switches this off. To put several plots on one page, use pv_house_mosaic() (second code block below) and export with the same cells. pv_save_house() writes <file>.pdf, <file>.png and preview.png to the output folder and overwrites any preview.png already there, so try it in a scratch folder.

# R: figures/<category>/<slug>/plot.R
source("../../../styles/r/theme_house.R")
library(ggplot2)
df <- data.frame(x = rep(1:10, 3),
                 y = c(1:10, 0.8 * (1:10) + 1, 0.5 * (1:10) + 2),
                 group = rep(c("a", "b", "c"), each = 10))   # or utils::read.csv("data.csv")
p <- ggplot(df, aes(x, y, colour = group)) +
  geom_line(linewidth = pv_house_lw("main")) +
  scale_colour_manual(values = unname(pv_palette("house", 3))) +
  theme_house()
pv_save_house(p, "figure", cells = "2x1")   # 89 x 43 mm

# Python
import sys
from pathlib import Path
import numpy as np
sys.path.insert(0, str(Path(__file__).resolve().parents[3] / "styles" / "python"))
import house
house.apply_house_style()
x = np.arange(1, 11)
fig, ax = house.house_figure(cells="2x1")
ax.plot(x, 0.8 * x + 1, color=house.pv_palette("house")[0])
house.save_house(fig, "figure")

# R: several plots on one page (1 + 1 + 2 cells, 183 x 43 mm)
source("../../../styles/r/theme_house.R")
library(ggplot2)
set.seed(1)
df <- data.frame(x = 1:30, y = 1:30 + rnorm(30, 0, 4))
fit <- summary(lm(y ~ x, data = df))
p_lab <- pv_fmt_p(coef(fit)[2, 4])                  # "P < 0.001" at 7/8 pt
p_exact <- pv_fmt_p(coef(fit)[2, 4], exact = TRUE)  # "P = m × 10^−k", only for text ≥ 8.6 pt
pa <- ggplot(df, aes(x, y)) +
  geom_point(shape = HOUSE_POINT$shape, size = HOUSE_POINT$size, stroke = HOUSE_POINT$stroke,
             fill = pv_palette("house")[["blue"]]) +
  annotate("text", x = 1, y = 35, hjust = 0, vjust = 1, size = pv_pt2size(7),
           label = p_lab, parse = TRUE) +
  labs(x = "Dose (mg)", y = "Response", tag = "a") + theme_house()
pb <- ggplot(df, aes(y)) +
  geom_histogram(bins = 8, fill = pv_palette("house")[["sky"]], colour = "white") +
  labs(x = "Response", y = "Count", tag = "b") + theme_house()
pc <- ggplot(df, aes(x, y)) +
  geom_line(linewidth = pv_house_lw("main"), colour = pv_palette("house")[["red"]]) +
  labs(x = "Day", y = "Response", tag = "c") + theme_house()
page <- pv_house_mosaic(list(
  list(plot = pa, at = c(1, 1), cells = c(1, 1)),
  list(plot = pb, at = c(2, 1), cells = c(1, 1)),
  list(plot = pc, at = c(3, 1), cells = c(2, 1))
), cells = c(4, 1))
pv_save_house(page, "figure", cells = c(4, 1))

规范摘要（中文）

新加的图一律用 house style；已有的图不重渲染，继续用旧版 theme_viz()。完整规则以 styles/style-contract.md 为准。

默认格数

拿不准时用下表；太挤或太空可在网格内升降一档。

Style summary (English)

New figures use the house style. Existing figures are not re-rendered and keep the legacy theme_viz(). The full rules are in styles/style-contract.md.

Default cells

Use these when unsure; if a figure is crowded or empty, move one size up or down within the grid.

默认尺寸与升档阈值 / Default sizes and when to go up

• 字宽 1.30 mm/字符
• 行高 ≥ 3.5 mm
• 面板内宽 = 格跨度 − 12.7 mm
• 顺序：先缩写或换行，再加宽，最后加高

• Character width: 1.30 mm/char
• Row height: ≥ 3.5 mm
• Inner panel width = cell span − 12.7 mm
• Order: abbreviate or wrap first, then widen, lastly add height

图型 Chart	默认 Default	组 / 类别 / 集合 > Groups / categories / sets >	标签字符 > Label chars >	行 > Rows >	列 / 交集 > Columns / intersections >	点 / bins / 面板 > Points / bins / panels >	升到 Up to
Scatter 散点	1×1	3	—	—	—	2,000	2×2
Line 折线	2×1	4	—	—	—	12	4×1 / 2×2
Bar 柱图	1×1	4	4	—	—	—	2×1
Box / violin	1×1	4	—	—	—	—	2×1
Raincloud（横向 / horizontal）	1×1	4	—	—	—	—	2×2
Forest 森林图	1×2	—	17	21	—	—	2×2
Heatmap 热图（印数值 / values）	2×2	—	10	21	17	—	2×3 / 2×4
Histogram / density	1×1	3	—	—	—	37	2×1
Donut 环图	1×1	4	5	—	—	—	2×1
Stacked bar 堆叠条	1×1	4	—	—	—	—	1×2 / 2×1
Small multiples 小多图	2×2	—	—	—	—	4	4×2 / 4×3
UpSet（拼图 / mosaic）	3×1	5	10	—	15	—	4×1 / 3×2
Manhattan / genome track	4×1	19	—	—	—	—	4×1 + 2×1
示例 / Examples
house-mosaic-transcriptome	4×3
house-mosaic-aba-kinetics	2×4
house-mosaic-gwas	4×2

参考示例 / Examples: house-mosaic-transcriptome (4×3) · house-mosaic-aba-kinetics (2×4) · house-mosaic-gwas (4×2). 测量方法与每档容量 / Method and capacity per size: styles/style-contract.md.

色板 / Palettes

中文说明

• geo 提供 6 种颜色，适用于地理或群体渐变（PCA 着色按地理顺序）。
• 当 n > 6 时，pv_palette("geo", n) 会自动插值。
• chrom12 提供 12 种交替色，相邻染色体一浅一深，适用于曼哈顿图或基因组轨道。
• 若染色体数量超过 12 条，系统会报错。
• 标签对比规则：组色对白底对比度 ≥ 3:1 时标签用组色，否则用 #1A1A1A。

English Description

• The geo palette offers 6 colors for geographic or population gradients. PCA coloring follows geographic order.
• The function pv_palette("geo", n) interpolates colors when n > 6.
• The chrom12 palette offers 12 alternating light and dark colors for Manhattan plots and genome tracks.
• Requesting more than 12 colors will result in an error.
• Label contrast rule: Use group color for labels if contrast against white background is ≥ 3:1, otherwise use #1A1A1A.

# R
cols <- pv_palette("geo")              # 6 colours; pv_palette("geo", 9) ramps
chr_cols <- pv_palette("chrom12")      # 12 colours; n > 12 is an error
# Python
cols = house.pv_palette("geo")
chr_cols = house.pv_palette("chrom12")

参考 / References

只列链接与图号；参考图不进仓库。

Links and figure numbers only; no reference images are stored here.