docs: demo S7 vs seurat memory usage

vignettes/memory.Rmd

@@ -69,16 +69,39 @@ subset_mem <- lapply(c(10, 50, 100, 200, nrow(renee_ds@sample_meta)), function(n
   bind_rows()
 ```
 
+```{r single_cell}
+to_bytes <- function(x, unit, base = 1024) {
+  bytes_units <- list(
+    K = 1,
+    M = 2,
+    G = 3,
+    T = 4
+  )
+  return(x * (base^bytes_units[[unit]]))
+}
+sc_dat <- tibble::tribble(
+  ~"project", ~"n_cells", ~"n_samples", ~"n_genes", ~"object_size_GB",
+  "CCBR1329/CCBR1243", 64642, 10, 26359, 9.35,
+  "CCBR1297", 9991, 2, 20989, 1.78,
+  "CCBR1035", 170789, 19, 32100, 31.54,
+  "CCBR1203", 208169, 35, 30858, 37.38
+) %>%
+  mutate(
+    object_size = to_bytes(object_size_GB, "G", base = 1000),
+    dataset_type = "single-cell (Seurat)"
+  )
+```
 
 
 ```{r plot_memory}
 # what is the resource spec of default NIDAP allocation?
-# TODO: compare to seurat single cell object
+palette_name <- "Set2"
 subset_mem %>%
-  ggplot(aes(n_samples, object_size)) +
+  ggplot(aes(n_samples, object_size, color = "")) +
   geom_point() +
   geom_line(linewidth = 0.3) +
   scale_y_continuous(labels = scales::label_bytes(units = "GB", accuracy = 0.1)) +
+  scale_color_manual(values = RColorBrewer::brewer.pal(3, palette_name)[[1]]) +
   labs(
     title = "Memory usage of S7 object",
     x = "Number of samples",
@@ -89,5 +112,27 @@ subset_mem %>%
       format(length(renee_ds@counts$raw$gene_id), big.mark = ",")
     )
   ) +
+  theme_bw() +
+  theme(legend.position = "none")
+```
+
+```{r plot_memory_comp}
+dat_comp <- bind_rows(
+  subset_mem %>% mutate(dataset_type = "bulk (S7)", object_size = as.double(object_size)),
+  sc_dat
+)
+dat_comp %>%
+  ggplot(aes(n_samples, object_size, colour = dataset_type)) +
+  geom_point() +
+  geom_line(linewidth = 0.3) +
+  scale_y_continuous(labels = scales::label_bytes(units = "GB", accuracy = 0.1)) +
+  scale_color_brewer(palette = palette_name) +
+  guides(colour = guide_legend(
+    title = "",
+    position = "top",
+    reverse = TRUE,
+    theme = theme(legend.text.position = "top")
+  )) +
+  labs(title = "Memory usage of bulk & single-cell RNA-seq objects", x = "Number of samples", y = "Memory usage") +
   theme_bw()
 ```