Merge pull request #271 from Olink-Proteomics/optimization_develop

Optimization develop to develop
Olink-Proteomics · Jul 26, 2023 · 589e180 · 589e180
2 parents 34d9a51 + 9773f09
commit 589e180
Show file tree

Hide file tree

Showing 39 changed files with 1,794 additions and 522 deletions.
diff --git a/.github/workflows/R-CMD-check.yaml b/.github/workflows/R-CMD-check.yaml
@@ -134,7 +134,7 @@ jobs:
         shell: Rscript {0}
 
       - name: Check package
-        run: |
+        run: |                             
           check_results <- rcmdcheck::rcmdcheck(error_on = "error",
                                                 path = "OlinkAnalyze", 
                                                 check_dir = "check")

diff --git a/OlinkAnalyze/DESCRIPTION b/OlinkAnalyze/DESCRIPTION
@@ -55,7 +55,7 @@ Description: A collection of functions to facilitate analysis of proteomic
 License: AGPL (>= 3)
 Depends: 
     R (>= 4.1.0)
-Imports: 
+Imports:
     broom,
     car,
     dplyr,
@@ -83,6 +83,7 @@ Imports:
     utils,
     zip
 Suggests: 
+    arrow,
     clusterProfiler,
     extrafont,
     FSA,

diff --git a/OlinkAnalyze/R/Olink_anova.R b/OlinkAnalyze/R/Olink_anova.R
@@ -199,7 +199,7 @@ olink_anova <- function(df,
       }
 
       number_of_samples_w_more_than_one_level <- df %>%
-        dplyr::group_by(SampleID, Index) %>%
+        dplyr::group_by(SampleID) %>%
         dplyr::summarise(n_levels = dplyr::n_distinct(!!rlang::ensym(effect), na.rm = TRUE)) %>%
         dplyr::ungroup() %>%
         dplyr::filter(n_levels > 1) %>%

diff --git a/OlinkAnalyze/R/Olink_bridgeselector.R b/OlinkAnalyze/R/Olink_bridgeselector.R
@@ -41,11 +41,11 @@ olink_bridgeselector<-function(df, sampleMissingFreq, n){
 
   #Outlier calculation as in qc_plot for filtering
   qc_outliers <- df %>%
-    dplyr::group_by(Panel, SampleID, Index) %>%
+    dplyr::group_by(Panel, SampleID) %>%
     dplyr::mutate(IQR = IQR(NPX, na.rm = TRUE),
                   sample_median = median(NPX, na.rm = TRUE)) %>%
     dplyr::ungroup() %>%
-    dplyr::select(SampleID, Index, Panel, IQR, sample_median) %>%
+    dplyr::select(SampleID, Panel, IQR, sample_median) %>%
     dplyr::distinct() %>%
     dplyr::group_by(Panel) %>%
     dplyr::mutate(median_low = mean(sample_median, na.rm = TRUE) - 3*sd(sample_median, na.rm = TRUE),
@@ -58,10 +58,42 @@ olink_bridgeselector<-function(df, sampleMissingFreq, n){
                                       IQR > iqr_low &
                                       IQR < iqr_high,
                                     0, 1)) %>%
-    dplyr::select(SampleID, Index, Panel, Outlier)
+
+    dplyr::select(SampleID, Panel, Outlier)
+
+
+  # Alternative LODs for when LOD is not present
+  if(!("LOD" %in% names(df))){
+    if("Max LOD" %in% names(df)){
+      df <- df |> 
+        dplyr::mutate(LOD = `Max LOD`)
+      message("Using Max LOD as filter criteria...")  
+    } else if ("Plate LOD" %in% names(df)){
+      df <- df |> 
+        dplyr::mutate(LOD = `Plate LOD`)
+
+      message("Using Plate LOD as filter criteria...")
+    } else if ("Plate_LOD" %in% names(df)){
+      df <- df |> 
+        dplyr::mutate(LOD = Plate_LOD)
+
+      message("Using Plate_LOD as filter criteria...")
+    } else {
+      df <- df |> 
+        dplyr::mutate(LOD = -Inf)
+
+      message("LOD not available. No filtering by LOD...")
+    }
+  }
+
+  if("SampleQC" %in% names(df)){
+    df <- df |> 
+      dplyr::mutate(QC_Warning = SampleQC)
+  }
+
 
   df_1 <- df %>%
-    dplyr::left_join(qc_outliers, by = c('SampleID', 'Index', 'Panel')) %>%
+    dplyr::left_join(qc_outliers, by = c('SampleID', 'Panel')) %>%
     dplyr::mutate(NPX = ifelse(NPX <= LOD, NA, NPX)) %>%
     dplyr::group_by(SampleID) %>%
     dplyr::mutate(QC_Warning = dplyr::if_else(all(toupper(QC_Warning) == 'PASS'),

diff --git a/OlinkAnalyze/R/Olink_normalization.R b/OlinkAnalyze/R/Olink_normalization.R
@@ -196,8 +196,27 @@ olink_normalization <- function(df1,
         df_adjusted_data <- df1 %>%
           dplyr::mutate(Panel=stringr::str_replace(Panel,'\\(.+', '')) %>%
           dplyr::left_join(adj_factor_df,by='OlinkID') %>%
-          dplyr::mutate(NPX = NPX + Adj_factor) %>%
-          dplyr::mutate(LOD = LOD + Adj_factor)
+          dplyr::mutate(NPX = NPX + Adj_factor) 
+
+        if("LOD" %in% names(df_adjusted_data)){
+          df_adjusted_data <- df_adjusted_data %>%
+            dplyr::mutate(LOD = LOD + Adj_factor)
+        }
+
+        if("Max LOD" %in% names(df_adjusted_data)){
+          df_adjusted_data <- df_adjusted_data %>%
+            dplyr::mutate(`Max LOD` = `Max LOD` + Adj_factor)
+        }
+
+        if("Plate LOD" %in% names(df_adjusted_data)){
+          df_adjusted_data <- df_adjusted_data %>%
+            dplyr::mutate(`Plate LOD` = `Plate LOD` + Adj_factor)
+        }
+
+        if("Plate_LOD" %in% names(df_adjusted_data)){
+          df_adjusted_data <- df_adjusted_data %>%
+            dplyr::mutate(Plate_LOD = Plate_LOD + Adj_factor)
+        }
 
         return(df_adjusted_data)
 
@@ -331,8 +350,27 @@ olink_normalization <- function(df1,
     dplyr::mutate(Panel=stringr::str_replace(Panel,'\\(.+', '')) %>%
     dplyr::left_join(adj_factor_df,by='OlinkID') %>%
     dplyr::mutate(Adj_factor = dplyr::if_else(Project == reference_project,0,Adj_factor)) %>%
-    dplyr::mutate(NPX = NPX + Adj_factor) %>%
-    dplyr::mutate(LOD = LOD + Adj_factor)
+    dplyr::mutate(NPX = NPX + Adj_factor) 
+
+  if("LOD" %in% names(df_adjusted_data)){
+    df_adjusted_data <- df_adjusted_data %>%
+      dplyr::mutate(LOD = LOD + Adj_factor)
+  }
+
+  if("Max LOD" %in% names(df_adjusted_data)){
+    df_adjusted_data <- df_adjusted_data %>%
+      dplyr::mutate(`Max LOD` = `Max LOD` + Adj_factor)
+  }
+
+  if("Plate LOD" %in% names(df_adjusted_data)){
+    df_adjusted_data <- df_adjusted_data %>%
+      dplyr::mutate(`Plate LOD` = `Plate LOD` + Adj_factor)
+  }
+
+  if("Plate_LOD" %in% names(df_adjusted_data)){
+    df_adjusted_data <- df_adjusted_data %>%
+      dplyr::mutate(Plate_LOD = Plate_LOD + Adj_factor)
+  }
 
 
   return(df_adjusted_data)

diff --git a/OlinkAnalyze/R/Olink_one_non_parametric.R b/OlinkAnalyze/R/Olink_one_non_parametric.R
@@ -131,7 +131,7 @@ olink_one_non_parametric <- function(df,
       }
 
       number_of_samples_w_more_than_one_level <- df %>%
-        dplyr::group_by(SampleID, Index) %>%
+        dplyr::group_by(SampleID) %>%
         dplyr::summarise(n_levels = dplyr::n_distinct(!!rlang::ensym(effect), na.rm = T),.groups = "drop") %>%
         dplyr::filter(n_levels > 1) %>%
         nrow(.)

diff --git a/OlinkAnalyze/R/Olink_ordinalRegression.R b/OlinkAnalyze/R/Olink_ordinalRegression.R
@@ -165,7 +165,7 @@ olink_ordinalRegression <- function(df,
       }
 
       number_of_samples_w_more_than_one_level <- df %>%
-        dplyr::group_by(SampleID, Index) %>%
+        dplyr::group_by(SampleID) %>%
         dplyr::summarise(n_levels = n_distinct(!!rlang::ensym(effect), na.rm = T)) %>%
         dplyr::ungroup() %>%
         dplyr::filter(n_levels > 1) %>%