Skip to contents

How-to-use

Lmisc.Rmd

Prerequisites 

library(Lmisc)
library(dplyr)
library(ggplot2)

First we need to build two dataframes; one containing info for the nodes and one containing for the lines

The node dataframe must contain the columns below. Each row represents a node.


nodes <- data.frame(
    node_id = c("A", "C", "B", "D"),
    label = c("A", "C", "B", "D"),
    x = c(0, 14, 14, 0),
    y = c(0, 3, 0, 3)
  )

The dataframe below describes the relationship between the aforementioned nodes.

  • from and to indicate the start and the end node.

  • curvature takes two values; 1 means the line should be curved and 0 that it’s a straight line.

  • pvalue, est, ci.upper, ci.lower are the p-value, the estimate, upper confidence interval and the lower confidence interval of the estimate. These are drawn on top of the lines

  • vjust and hjust indicates the vertical and horizontal offset for the text on top of the line

  • scale_factor controls the gap between the boxes and the start/end of the straight line Values between 0 and 1, shrink the length of the line. scale_factor=0 means that the line start and ends at the exact centers of the boxes. scale_factor=1 means that the line start and ends near the borders of the boxes. Default value is 1. If you need granular control, specify the values for each relationship.

  • curvature_amount controls the curvature of the curved paths. These follow the notation of ggplot: positive values mean a right-hand curve while negative values a left-hand curve.


edges <- data.frame(
  from = c(
    "A", "C", "B", "D"
  ),
  to = c(
    "B", "D", "C", "A"
  ),
  curvature = c(0, 0, 0, 0),
  pvalue = c(0.1, 0.30, 0.5, 0.5),
  est = c(
    0.2, 0.15, -0.25, -0.3
  ),
  ci.lower = c(0.4, 0.1, -0.6, -0.3),
  ci.upper = c(1.2, 0.9, 0.0, -0.3),
  hjust = c(0.5, 0.5, 0.5, 0.5),
  vjust = c(0.5, 0.5, 0.50, 0.5),
  curvature_amount = c(
    0, 0, 0, 0
  ),
  scale_factor=c(
    1, 1, 1, 1
  )
)

Plot 

p <- plot_dag(nodes, edges, ylim = c(-0.2, 3.2), xlim = c(-0.2, 14.2), text_size = 3)
print(p)

Explanation of curvature_amount

Imagine you’re walking along the path from start to end:

Curvature Sign Curve Direction What It Looks Like
Positive (+) Right-hand curve Path bends to your right as you walk forward.
Negative (−) Left-hand curve Path bends to your left as you walk forward.
Zero (0) Straight line No bending; direct line. 
edges <- data.frame(
  from = c(
    "A", "C", "B", "D"
  ),
  to = c(
    "B", "D", "C", "A"
  ),
  curvature = c(1, 1, 1, 1),
  pvalue = c(0.1, 0.30, 0.5, 0.5),
  est = c(
    0.2, 0.15, -0.25, -0.3
  ),
  ci.lower = c(0.4, 0.1, -0.6, -0.3),
  ci.upper = c(1.2, 0.9, 0.0, -0.3),
  hjust = c(0.5, 0.5, 0.5, 0.5),
  vjust = c(0.5, 0.5, 0.50, 0.5),
  curvature_amount = c(
    0.2, -0.2, 0.2, -0.2
  )
)
p <- plot_dag(nodes, edges, ylim = c(-2, 3.2), xlim = c(-0.2, 15), text_size = 3)
print(p)

When we walk from D to A (or C to D) with a negative value (curvature_amount = -0.2), the curved path is left-handed. When we walk from B to C (or from A to B) with a positive value, the path is right-handed (curvature_amount = 0.2).

Control of scale_factor

We can control the length of the lines by using the scale_factor argument.

By using 0, we say that the lines should start from the center of the boxes. 1 is the default.

  • We can assign a value between 0 & 1 to shrink the lines
edges <- data.frame(
  from = c(
    "A", "C", "B", "D", "B", "C"
  ),
  to = c(
    "B", "D", "C", "A", "D", "A"
  ),
  curvature = c(0, 0, 0, 0, 0, 0),
  pvalue = c(0.1, 0.30, 0.5, 0.5, 0.5, 0.005),
  est = c(
    0.2, 0.15, -0.25, -0.3, -0.35, -0.4
  ),
  ci.lower = c(0.4, 0.1, -0.6, -0.3, -0.3, -0.3),
  ci.upper = c(1.2, 0.9, 0.0, -0.3, -0.3, -0.3),
  hjust = c(0.5, 0.5, 0.5, 0.5, 0.3, 0.8),
  vjust = c(0.5, 0.5, 0.50, 0.5, 0, 0),
  curvature_amount = c(
    0, 0, 0, 0,0,0
  )
)

# Set the scale_factor values
edges$scale_factor <- c(0.95,0.9,0.95,0.9,0.95,0.9)

plot_dag(nodes, edges, ylim = c(-0.2, 3.2), xlim = c(-0.2, 14.2), text_size = 3)

As you can see, those relationships that have 0.90 as their scale_factor values have their length shrinken.