more_ggplots.Rmd

# More ggplots

## Histograms and density plots

### Histogram

We can use one of our previous objects to create a histogram. You can get it back with:

```{r}
rnaseq2 <- read_csv("DataViz_source_files-main/files/GSE150029_rnaseq_log2_long.csv")
```

```{r}
ggplot(rnaseq2, aes(x=log2_counts)) + 
  geom_histogram()
```

We can split the histogram per sample:

```{r}
ggplot(rnaseq2, aes(x=log2_counts, color=sample)) + 
  geom_histogram() 
```

Set position to "identity", so histograms will not be on top of each other:

```{r}
ggplot(rnaseq2, aes(x=log2_counts, color=sample)) + 
  geom_histogram(position="identity")
```

Set alpha (transparency to 0.2):

```{r}
ggplot(rnaseq2, aes(x=log2_counts, color=sample)) + 
  geom_histogram(position="identity", alpha=0.2)
```

Express as density instead of counts:

```{r}
ggplot(rnaseq2, aes(x=log2_counts, color=sample)) + 
  geom_histogram(aes(y=after_stat(density)), position="identity", alpha=0.2) +
  theme_classic()
```


### Density plot

The structure is pretty much the same:

```{r}
ggplot(rnaseq2, aes(x=log2_counts, color=sample)) + 
  geom_density(alpha=0.2) +
  theme_classic()
```

You can use fill instead:

```{r}
ggplot(rnaseq2, aes(x=log2_counts, fill=sample)) + 
  geom_density(alpha=0.2) +
  theme_classic()
```

### Histogram + density

As we combined geom_boxplot and geom_violin in a previous exercise, we can also combine geom_histogram and geom_density:

```{r}
ggplot(rnaseq2, aes(x=log2_counts, color=sample)) + 
  geom_histogram(aes(y=after_stat(density)), position="identity", alpha=0.2) +
  geom_density(alpha=0.2) +
  theme_classic()
```


## Pie chart

We will plot a pie chart representing the number of genes from **gtf** found in each chromosome.

Prepare data so as to obtain counts (that is more advanced dplyr manipulation - we will not go through details):

```{r}
gtf_count_chr <- gtf %>%
  group_by(chr) %>% # group by allows to make the next calculations by the group specified
  summarise(counts=n())
```

The **coord_polar()** is needed here: pie plots are stacked barplots in polar coordinates.

```{r}
ggplot(data=gtf_count_chr, mapping=aes(x="", y=counts, fill=chr)) + 
  geom_bar(stat="identity") +
  coord_polar("y", start=0)
```

```{r}
ggplot(data=gtf_count_chr, mapping=aes(x="", y=counts, fill=chr)) + 
  geom_bar(stat="identity") +
  coord_polar("y", start=0) +
  theme_void()
```

Change the color scale:

```{r}
ggplot(data=gtf_count_chr, mapping=aes(x="", y=counts, fill=chr)) + 
  geom_bar(stat="identity") +
  coord_polar("y", start=0) +
  scale_fill_brewer(palette="Paired") +
  theme_void()
```

Here is a [nice page](https://genchanghsu.github.io/ggGallery/posts/2021-06-18-post-3-pie-charts-with-ggplot/) about more pie plot customization.

## Marginal plots

Function [**ggMarginal**](https://www.rdocumentation.org/packages/ggExtra/versions/0.10.1/topics/ggMarginal) from package {ggExtra} allows to easily add histograms, boxplots or density plots to the margins of a scatter plot.

```{r}
library(ggExtra)

# Save the classic plot to an object.
p <- ggplot(data=geneexp, mapping=aes(x=sample1, y=sample2)) +
      geom_point()
```

```{r}
# add marginal histogram
ggMarginal(p, type="histogram")
```

```{r}
# add marginal density
ggMarginal(p, type="density")
```

```{r}
# add marginal boxplot
ggMarginal(p, type="boxplot")
```

The **mapping of variables** (for example, using **color** or **fill**) can also be **inherited** by the marginal plots.

```{r}
# add mapping of points to "DE" column :
p <- ggplot(data=geneexp, mapping=aes(x=sample1, y=sample2, color=DE)) +
  geom_point()

# tell ggMarginal to color / split the boxplot according to variable mapped in color/colour
ggMarginal(p, type="boxplot", groupColour = TRUE)
```

Let's add more features that we have learned in this workshop:

```{r, fig.height=12, fig.width=10}
# add mapping of points to "DE" column :
p <- ggplot(data=geneexp, mapping=aes(x=sample1, y=sample2, color=DE)) +
  geom_point() +
  theme_minimal(base_size = 16) +
  geom_rug(alpha=0.4, sides = "tr") +
  geom_density_2d(alpha=0.2, colour = "pink") +
  ggtitle("Example of ggMarginal") +
  theme(legend.position="bottom", plot.title = element_text(hjust = 0.5))

# tell ggMarginal to color / split the boxplot according to variable mapped in color/colour
ggMarginal(p, type="boxplot", groupColour = TRUE)
```