Merge pull request #13 from cytomining/updates

Handle rep_group correctly
cytomining · Mar 13, 2023 · 8fa534a · 8fa534a
2 parents 5e39445 + e8623b2
commit 8fa534a
Show file tree

Hide file tree

Showing 4 changed files with 54 additions and 13 deletions.
diff --git a/matric/2.calculate_index.Rmd b/matric/2.calculate_index.Rmd
@@ -81,16 +81,27 @@ profiles <-
 ```{r}
 log_info("Calculating index ...")
 
-if (is.null(params$sim_params$any_different_cols_non_rep)) {
-  log_info("Reducing similarity calculations because `any_different_cols_non_rep` is NULL ...")
-  x_all_same_cols_rep <- params$sim_params$all_same_cols_rep
+if (is.null(params$sim_params$any_different_cols_non_rep) &
+    xor(
+      !is.null(params$sim_params$all_same_cols_rep),
+      !is.null(params$sim_params$all_same_cols_group)
+    )) {
+  log_info(
+    "Reducing similarity calculations because `any_different_cols_non_rep` is NULL and only one of `all_same_cols_rep` and `all_same_cols_group` are specified..."
+  )
+  if (!is.null(params$sim_params$all_same_cols_rep)) {
+    x_all_same_cols_rep_or_group <- params$sim_params$all_same_cols_rep
+  } else {
+    x_all_same_cols_rep_or_group <-
+      params$sim_params$all_same_cols_group
+  }
   x_all_same_cols_ref <- params$sim_params$all_same_cols_ref
   x_reference_df <- reference_df
 } else {
   log_info(
-    "Performing all similarity calculations because `any_different_cols_non_rep` is not NULL ..."
+    "Performing all similarity calculations because `any_different_cols_non_rep` is not NULL or both `all_same_cols_rep` and `all_same_cols_group` are specified...."
   )
-  x_all_same_cols_rep <- NULL
+  x_all_same_cols_rep_or_group <- NULL
   x_all_same_cols_ref <- NULL
   x_reference_df <- NULL
 }
@@ -102,7 +113,7 @@ sim_df <-
       population = profiles,
       method = NULL,
       lazy = TRUE,
-      all_same_cols_rep_or_group = x_all_same_cols_rep,
+      all_same_cols_rep_or_group = x_all_same_cols_rep_or_group,
       all_same_cols_ref = x_all_same_cols_ref,
       reference = x_reference_df
     )

diff --git a/matric/3.calculate_metrics.Rmd b/matric/3.calculate_metrics.Rmd
@@ -123,7 +123,7 @@ if (!is.null(attr(collated_sim, "params")$calculate_index$sim_params$any_differe
     sim_metrics(
       collated_sim,
       sim_type_background = "non_rep",
-      calculate_grouped = FALSE,
+      calculate_grouped = calculate_grouped,
       use_furrr = TRUE,
       calculate_pvalue = TRUE
     )

diff --git a/matric/5.inspect_metrics.Rmd b/matric/5.inspect_metrics.Rmd
@@ -71,6 +71,8 @@ plot_metric <-
 
 # Read
 
+## Level 1_0
+
 ```{r}
 metric_set <- glue("level_1_0_{type}")
 
@@ -84,11 +86,9 @@ log_info("Reading {parquet_file} ...")
 
 level_1_0_metrics <-
   arrow::read_parquet(glue(parquet_file))
-
-all_same_cols_rep <- attr(level_1_0_metrics, "all_same_cols_rep")
 ```
 
-After reading level_1, drop duplicates that may result from annotating level 1_0 entities
+## Level 1
 
 ```{r}
 metric_set <- glue("level_1_{type}")
@@ -102,11 +102,41 @@ parquet_file <-
 log_info("Reading {parquet_file} ...")
 
 level_1_metrics <-
-  arrow::read_parquet(glue(parquet_file)) %>%
+  arrow::read_parquet(glue(parquet_file))
+```
+
+After reading level_1, drop duplicates that may result from annotating level 1_0 entities
+
+```{r}
+all_same_cols_rep <- attr(level_1_0_metrics, "all_same_cols_rep")
+
+level_1_metrics <-
+  level_1_metrics %>%
   select(all_of(all_same_cols_rep), matches("^sim_")) %>%
   distinct()
 ```
 
+
+## Level 2_1
+
+```{r}
+metric_set <- glue("level_2_1_{type}")
+
+parquet_file <-
+  with(
+    params,
+    glue("{input_metrics_file_prefix}_{metric_set}.parquet")
+  )
+
+if (file.exists(parquet_file)) {
+  log_info("Reading {parquet_file} ...")
+  
+  level_2_1_metrics <-
+    arrow::read_parquet(glue(parquet_file))
+}
+```
+
+
 # Plot metrics
 
 ## Average Precision

diff --git a/renv.lock b/renv.lock
@@ -1314,7 +1314,7 @@
       "RemoteRepo": "matric",
       "RemoteUsername": "shntnu",
       "RemoteRef": "null_df",
-      "RemoteSha": "419ed83c194624a5a5393f6c8765fae9c2940630",
+      "RemoteSha": "a2adb70e437cd1f07020ada415292590220d4ff1",
       "Requirements": [
         "R",
         "arrow",
@@ -1336,7 +1336,7 @@
         "tidyr",
         "yardstick"
       ],
-      "Hash": "653b4266a16595683c83c021c59f030e"
+      "Hash": "ab6353150e1beacd9443747b47539fcd"
     },
     "memoise": {
       "Package": "memoise",