---
title: "buildconstellation"
date: 2021-03-03
params:
  name: NULL
  args_list: NULL
  groups.col: NULL
  colors.use : NULL
  max_size: 20
  n.pcs: 20
  run.knn: TRUE
  k: 15
  frac.th: 0.05
  out.dir: '../out'
output: html_document
---

```{r}
knitr::opts_knit$set(progress = TRUE, verbose = FALSE)
```

```{r}
name <- params$name
groups.col <- params$groups.col
colors.use <- params$colors.use
max_size <- params$max_size

n.pcs <- params$n.pcs
k <- params$k
frac.th <- params$frac.th
run.knn <- params$run.knn

out.dir <- params$out.dir

args_list <- params$args_list

cells.cl.df <- args_list$cells_cl_df
rd.dat <- args_list$reductions
cl <- args_list$cl
cl.numeric <- args_list$cl_numeric
```

Extra functions

```{r functions}
source_rmd("scrattch.hicat_fxns.Rmd")
```

## 1. Build dataframes for constellation plots.

```{r cl_and_cl_numeric, eval=TRUE, include=TRUE}
# cl and cl numeric can also be passed as a param to the Rmd.

if(is.null(cl) & is.null(cl.numeric)) {

  message("No cl and c;.numeric provided")
# Cluster, or area-celltype combination.
  cl <- cells.cl.df[[groups.col]] %>% as.factor %>%
    set_names(cells.cl.df$cell.name)
  # 128 clusters in all _combo2_ cells
  # 77 clusters in all _combo2_ & ExN lineage cells.

  cl.numeric <- as.numeric(cl) %>% purrr::set_names(nm = names(cl))
}
```

2

```{r cl.df, eval=TRUE, include=TRUE}

cl.df <- get_cl_df(cl)

cl.df$area <- str_split_fixed(cl.df$cluster_label, "-", 2)[ ,2] %>% tolower

cl.df$clade <- str_split_fixed(cl.df$cluster_label, "-", 2)[ ,1] %>% tolower
# Add clade_id, clade_color to cl.df
# cl.df <- cl.df %>% left_join(clade.cols)
cl.df <- cl.df %>% left_join(colors.use, by = c("clade" = "celltype"))
# %>%  rename(cluster_color = "colors.use")
# rm(group.cols)

# cells.cl.df: Add cluster_id column from cl.df; remove unused columns.
cells.cl.df <- cells.cl.df %>% select(-cluster_label) %>% rename(cluster_label = groups.col) %>%
                        left_join(
                         # %>% select(cell.name, groups.col, combined.cluster.2),
                         cl.df, by = "cluster_label") %>%

                         # Requires cells.cl.df (metadata) to have column being used for groups
                         # named 'cluster_label' to match with cl_df during join.
                 mutate(cluster_id = as.factor(cluster_id))
```

## 4 Find cluster centers from UMAP coordinates

```{r rd.cl.center}

rd.cl.center <- get_RD_cl_center(rd.dat = rd.dat$umap, cl)
message("Got rd.cl.center")
```

```{r update-rd.cl.center}
rd.cl.center %<>%
  as.data.frame %>%
  set_names(c("x", "y")) %>%
  add_column(cl = cl.df$cluster_id, .before = "x") %>%
  # add_column preserves rownames.
  # but moving rownames to column cluster_label anyway bc of left_join below.
  # Needs to be cl (not cl_id) or else you get error:
  # Error in `$<-.data.frame`(`*tmp*`, "edge.frac.within", value = numeric(0)) :
  # replacement has 0 rows, data has 26
  rownames_to_column("cluster_label")

message("Updated rd.cl.center")
```

## 5 Join `cl.df` and `rd.cl.center` into `cl.center.df` for input into `get_KNN_graph`.

```{r cl.center.df}
cl.center.df <- left_join(rd.cl.center, cl.df,
                          by = c("cluster_label"))
```

## 6 Get knn and cluster counts

Calls `knn.cl` in `scrattch.hicat_fxns.Rmd`

```{r knn.result}
if(run.knn == TRUE) {

message("Running KNN")

knn.result <- RANN::nn2(data = rd.dat$pca[, 1:n.pcs], k = k)

} else {knn.result <- knn.result}
```

```{r knn.cl}

knn.cl <- get_knn_graph(knn.result = knn.result,
                        rd.dat = rd.dat$umap,
                        cl.df =  cl.df,
                        cl = cl.numeric,
                        cl.numeric = cl.numeric,
                        knn.outlier.th = 2,
                        outlier.frac.t = 0.5)

# rm(rd.dat, ncx.clusters)
```

# 2. Make constellation plot

```{r knn_cl_df_filter}

# Keep only cells whose $frac (fraction of cells in cluster with nearest neighbors in a different cluster) >= 0.05.
# Defined in `get_knn_graph`:
# knn.cl.df$frac = knn.cl.df$Freq / knn.cl.df$cl.from.total
# 10% : 213 edges
knn.cl.df.filter <- knn.cl$knn.cl.df %>% dplyr::filter(frac >= frac.th) %>%
                      mutate(cl.from = as.numeric(cl.from), cl.to = as.numeric(cl.to))
# cl.to, cl.from numeric or factor?
# Need to be numeric for getting the rows where cl.to == cl.from (knn.cl.df$same)
```

```{r plot-constellation}
# Plot only edges between ExN lineage clusters.
# knn.cl.df %<>% filter_at(vars(cl.from.label, cl.to.label),
#                        all_vars(str_detect(., "RG|IPC|Neuron|OPC|Dividing"))
#              )

# cl.center.df$cluster_label %<>% str_remove("_combo2")
message("Plotting")

cl.plot <- plot_constellation(knn.cl.df = knn.cl.df.filter,
                              cl.center.df = cl.center.df,
                              out.dir = out.dir,
                              node.label = "cluster_label",  # name of column in cl.center.df with cluster/node names
                              exxageration = 0.4, curved = TRUE,
                              plot.parts = FALSE, plot.hull = NULL,
                              plot.height = 40, plot.width = 40,
                              node.dodge = TRUE,
                              label.size = 3,
                              max_size = max_size)

```

```{r}
result <- lst(args_list, results = lst(cl.plot, knn.result, cl.center.df, knn.cl, knn.cl.df.filter))
# rm(cl.plot, knn.result, cl.center.df, knn.cl, knn.cl.df.filter)
st <- format(Sys.time(), "%Y%m%d_%H%M_")
write_rds(result, path = file.path(out.dir, paste0(st, name, "_constellation_tables.rds")))
```