quollr3hexbin.Rmd

---
title: "3. Algorithm for binning data"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{3. Algorithm for binning data}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
options(rmarkdown.html_vignette.check_title = FALSE)
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  warning = FALSE, 
  message = FALSE
)
```

Here, we'll discuss the algorithm for binning data.

```{r setup}
library(quollr)
library(ggplot2)
library(dplyr)
```

By passing the preprocessed 2D embedding data and hexagonal grid configurations, you can obtain the hexagonal binning information like centroid coordinates, hexagonal polygon coordinates, the standardise counts within each hexagon etc.  

```{r}
r2 <- diff(range(s_curve_noise_umap$UMAP2))/diff(range(s_curve_noise_umap$UMAP1))
hb_obj <- hex_binning(data = s_curve_noise_umap_scaled, bin1 = 6, r2 = r2)
```

```{r}
## Data set with all possible centroids in the hexagonal grid
all_centroids_df <- hb_obj$centroids
glimpse(all_centroids_df)

## Generate all coordinates of hexagons
hex_grid <- hb_obj$hex_poly
glimpse(hex_grid)

## To obtain the standardise counts within hexbins
counts_df <- hb_obj$std_cts
df_bin_centroids <- extract_hexbin_centroids(centroids_df = all_centroids_df, 
                                             counts_df = counts_df) |>
      filter(drop_empty == FALSE)
```

```{r}
ggplot(data = hex_grid, aes(x = x, y = y)) + 
  geom_polygon(fill = "white", color = "black", aes(group = hex_poly_id)) +
  geom_point(data = all_centroids_df, aes(x = c_x, y = c_y), color = "red") +
  coord_fixed()
```

```{r}
ggplot(data = hex_grid, aes(x = x, y = y)) + 
  geom_polygon(fill = "white", color = "black", aes(group = hex_poly_id)) +
  geom_point(data = all_centroids_df, aes(x = c_x, y = c_y), color = "red") +
  geom_point(data = df_bin_centroids, aes(x = c_x, y = c_y), color = "purple") +
  coord_fixed()
```

```{r}
ggplot(data = hex_grid, aes(x = x, y = y)) + 
  geom_polygon(fill = "white", color = "black", aes(group = hex_poly_id)) +
  geom_point(data = s_curve_noise_umap_scaled, aes(x = UMAP1, y = UMAP2), color = "blue") +
  coord_fixed()
```

```{r}
hex_grid_with_counts <- left_join(hex_grid, counts_df, by = c("hex_poly_id" = "hb_id"))

ggplot(data = hex_grid_with_counts, aes(x = x, y = y)) +
  geom_polygon(color = "black", aes(group = hex_poly_id, fill = std_counts)) +
  geom_text(data = all_centroids_df, aes(x = c_x, y = c_y, label = hexID)) +
  scale_fill_viridis_c(direction = -1, na.value = "#ffffff") +
  coord_fixed()
```

You can also use `geom_hexgrid` to visualise the hexagonal grid rather than `geom_polygon`.

```{r}
ggplot(data = all_centroids_df, aes(x = c_x, y = c_y)) +
  geom_hexgrid() +
  coord_equal() +
  xlab("x") + ylab("y") +
  theme(axis.text = element_text(size = 5),
        axis.title = element_text(size = 7))

```