Plot blocks

[DivadNojnarg]

• Motivation: We currently build plots as a single big block. This is not sustainable and hard to customise for the end user.

• Proposition:

  • We aim at building the plot by layer by layer where each layer is a block.
  • For each core geom (TBD), extract core parameters (TBD) and translate them into fields that users can configure.
  • If within a geom each parameter is a field, aes for mappings should be a composite field by leveraging new_list_field.
  • UI consideration: currently, the stack puts margin between blocks. We'd have to create a new S3 method for the plot block such that all layers belong to the same UI container, thereby avoiding to end with too many individual blocks.

• Questions:

  • How do we build a ggplot?
    • Pattern 1: data and mapping passed to the main function.

    ggplot(data = <DATA>, mapping = <MAPPING>) +
    geom_point()

    • Pattern 2: Common data but specific mapping for layers.

     ggplot(data = <DATA>) +
     geom_point(mapping = <MAPPING>) +
     geom_point(mapping = <MAPPING>)

    • Pattern 3: each layer can have its own data and mappings (more flexible)

     ggplot() +
     geom_point(data = <DATA>, mapping = <MAPPING>) +
     geom_point(data = <DATA>, mapping = <MAPPING>)

  • What are the core geoms to support?
  • How do we programmatically extract the list of aesthetics to support for each geom? For instance geom_points understands x, y, alpha, ... while geom_line also supports linetype.
  • What else should we support: coordinates, faceting, scales, labels, themes, ... ?

cc @christophsax, @JohnCoene and @nbenn.

[DivadNojnarg]

Below is an example of a plot we generate with @karmatarap, which should give use some insight about what core elements to support:

p <- ggplot(data) +
        geom_point(
          mapping = aes(
            x = .data[[x_var]],
            y = .data[[y_var]],
            color = .data[[color]],
            shape = .data[[shape]]
          ),
          size = 3 
        ) +
        geom_errorbar(
          aes(
            x = .data[[x_var]],
            y = .data[[y_var]],
            ymin = MEAN - SE,
            ymax = MEAN + SE,
            color = ACTARM
          ),
          width = 0.2
        ) +
        geom_line(
          aes(
            x = .data[[x_var]],
            y = .data[[y_var]],
            group = .data[[color]],
            color = .data[[color]]
          )
        ) +
        labs(
          title = .(title),
          x = .(x_lab),
          y = .(y_lab)
        ) +
        theme(
          axis.text.x = element_text(angle = 45, hjust = 1),
          legend.title = element_text(face = "bold"),
          legend.position = "bottom"
        ) +
        scale_color_brewer(name = "Treatment Group", palette = "Set1") +
        scale_shape_manual(
          name = "Treatment Group",
          values = c(16, 17, 18, 19, 20)
        )