Is there a nicer way to plot this PCAPlot in ggplot after doing plotPCA in DESeq2?

Question

My data has the following structure:

structure(list(PC1 = c(0.575275450335122, -0.271560036518229, 
0.953309669344739, -0.0973831071329307, 0.871105953866523, 0.0379904534598842, 
0.737223421023106, -1.07991788617048, 0.977937610804854, 0.206927407135419, 
0.846524295528142, -0.743507855124003, 0.229716783731562, -1.708551233374, 
0.623630074143358, -1.22006221488515, 0.624234285603959, -1.56289307177187
), PC2 = c(0.912546111986394, -0.247117897558157, 1.15416425034184, 
-0.0618766648767635, 1.30003524991646, -0.339609948322352, 0.12755285074635, 
0.130688683339775, -0.596424639876554, -1.31752263137418, -0.390273336420353, 
-0.562559844453795, -0.129516525520303, 0.407888572167762, -0.545085195561956, 
0.177266986014392, -0.343592143975577, 0.323436123427009), group = structure(c(1L, 
2L, 7L, 8L, 13L, 14L, 3L, 4L, 9L, 10L, 15L, 16L, 5L, 6L, 11L, 
12L, 17L, 18L), .Label = c("High:D10:Basal", "High:D10:Stim", 
"High:D14:Basal", "High:D14:Stim", "High:D18:Basal", "High:D18:Stim", 
"Low:D10:Basal", "Low:D10:Stim", "Low:D14:Basal", "Low:D14:Stim", 
"Low:D18:Basal", "Low:D18:Stim", "Medium:D10:Basal", "Medium:D10:Stim", 
"Medium:D14:Basal", "Medium:D14:Stim", "Medium:D18:Basal", "Medium:D18:Stim"
), class = "factor"), CellType = structure(c(2L, 2L, 1L, 1L, 
3L, 3L, 2L, 2L, 1L, 1L, 3L, 3L, 2L, 2L, 1L, 1L, 3L, 3L), .Label = c("Low", 
"High", "Medium"), class = "factor"), Donor = structure(c(1L, 
1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 
3L), .Label = c("D10", "D14", "D18"), class = "factor"), Day = structure(c(1L, 
2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 
2L), .Label = c("Basal", "Stim"), class = "factor"), name = c("D10_High_d0", 
"D10_High_d7", "D10_Low_d0", "D10_Low_d7", "D10_Medium_d0", "D10_Medium_d7", 
"D14_High_d0", "D14_High_d7", "D14_Low_d0", "D14_Low_d7", "D14_Medium_d0", 
"D14_Medium_d7", "D18_High_d0", "D18_High_d7", "D18_Low_d0", 
"D18_Low_d7", "D18_Medium_d0", "D18_Medium_d7")), class = "data.frame", row.names = c("D10_High_d0", 
"D10_High_d7", "D10_Low_d0", "D10_Low_d7", "D10_Medium_d0", "D10_Medium_d7", 
"D14_High_d0", "D14_High_d7", "D14_Low_d0", "D14_Low_d7", "D14_Medium_d0", 
"D14_Medium_d7", "D18_High_d0", "D18_High_d7", "D18_Low_d0", 
"D18_Low_d7", "D18_Medium_d0", "D18_Medium_d7"), percentVar = c(0.386943314549313, 
0.207313334715293))

The data was retrieved after doing plotPCA in the DESeq2 package. I have "three levels" of attributes that I'd like to explore in this PCA, so I figured that using shape, fill and text to label them was enough (with the following code). But for some reason, the legend for the fill is not showing the correct colours. The PCA is OK, but the colours in the legend are all black. Is there a better way to plot my dummy data?

percentVar <- round(100 * attr(pcaData, "percentVar"))
ggplot(pcaData, aes(x= PC1, y = PC2))+
  geom_point(size= 3, aes(shape=CellType, fill=Day)) +
  scale_fill_manual(values = c("#E69F00", "#56B4E9"))+
  scale_shape_manual(values=c(21, 23, 25))+
  geom_text_repel(size= 3.5, aes(label=Donor, colour=Donor))+ 
  scale_color_discrete()+
  xlab(paste0("PC1: ", percentVar[1], "% variance")) +
  ylab(paste0("PC2: ", percentVar[2], "% variance")) +
  coord_fixed() +
  ggtitle("PCA with Scaled data")

Answer 1

The problem is that while fill legend is showing, it automatically sets a shape for the legend that doesn't have a fill; which is kind of useless indeed.

To correct this, you can manually specify/override the shape of the fill legend like so:

  scale_fill_manual(values = c("#E69F00", "#56B4E9"),
                    guide = guide_legend(override.aes = list(shape = 22)))

I'd recommend a shape that you're not using already so there is no confusion between the fill and shape legends.