Add Viridis like compatibilities for multi feature map

[AndrewLyon]

When using FeaturePlot3 with multiple features, it gets difficulty to observe the differences between colors at higher resolutions.

Is it possible to add support for other palettes than the standard options, perhaps even the option to choose the specific colors for the chosen features?

[AndrewLyon]

Here are some examples:

It gets difficult to see where any overlap lies, or even where there is no expression vs low.

[huayc09]

Thank you for raising this issue! In your example, you're using the red + yellow combination. Yellow can indeed be difficult to see at higher resolutions due to its high brightness/luminance.

My recommendations:

1. For white/light backgrounds: Use RYB color model (simulates subtractive paint mixing). For two features, I suggest red + blue instead of red + yellow, as the yellow is too bright and hard to see.

2. For dark/black backgrounds: Use RGB color model (simulates additive light mixing). The red + green combination provides more noticeable contrast on black backgrounds.

Example code:

library(Seurat)
library(SeuratExtend)

# For white/light background: use RYB with red+blue
# (RYB mode: feature.1 = red, feature.2 = yellow, feature.3 = blue)
FeaturePlot3(
  pbmc,
  feature.1 = "CD3D",   # Red
  feature.2 = NA,        # Skip yellow (too bright)
  feature.3 = "CD14",   # Blue
  color = "ryb",
  pt.size = 0.5
)

# For dark/black background: use RGB with red+green
# (RGB mode: feature.1 = red, feature.2 = green, feature.3 = blue)
FeaturePlot3(
  pbmc,
  feature.1 = "CD3D",   # Red
  feature.2 = "CD14",   # Green
  feature.3 = NA,        # Skip blue
  color = "rgb",
  dark.theme = TRUE,
  pt.size = 1,
  color.range = c(0.1, 1)
)

Regarding custom palette mixing (e.g., Viridis):

Unfortunately, custom palette mixing is not currently supported because it's difficult to model a generalized color blending system that works predictably across different color spaces.

Advanced customization:

For the RYB color system, if you're comfortable with internal code modifications, you can customize the color blending weights by modifying the internal ryb_ref variable in R/Colors.R:142-148. This defines the RYB->RGB color mapping, but you'll need to experiment and debug to achieve your desired effect.

# Internal reference (for advanced users only)
ryb_ref <- data.frame(
  x = c(0, 1, 2, 3, 4, 5, 6),          # Color wheel positions
  r = c(1, 1, 1, 0, 0, 0.625, 1),      # Red component
  y = c(0.125, 0.5, 0.75, 0.75, 0.5, 0.125, 0.125),  # Yellow
  b = c(0.125, 0, 0, 0, 1, 0.9375, 0.125)           # Blue
)

Let me know if you need further clarification!