Introduction to bridging Olink® NPX datasets

Olink® NPX datasets are normalized datasets using either plate control normalization or intensity normalization methods. Intensity normalization method assumes that all samples within a project are fully randomized.

The joint analysis of two or more Olink® NPX datasets often requires an additional batch correction step to remove technical variations, which is referred to as bridging.

To bridge two or more Olink® NPX datasets, bridging samples are needed to calculate the assay-specific adjustment factors between datasets. Bridging samples are shared samples among datasets - that is samples that are analysed in both datasets. The recommended number of bridging samples are shown in the table below. Olink® NPX datasets without shared samples should not be combined using the bridging approach described below.

The following tutorial is designed to give you an overview of the kinds of data combining methods that are possible using the Olink® bridging procedure. Before starting bridging, it is important to check if the same sample IDs were assigned to the bridging samples.

Recommended number of bridging samples for Olink platforms
Platform	BridgingSamples
Target 96	8-16
Explore 1536	8-16
Explore Expansion	16-24

Setup bridging datasets

Bridging datasets are standard Olink® NPX tables. They can be loaded using read_NPX() function with NPX manager output file as input.

data1 <- read_NPX("~/NPX_file1_location.xlsx")
data2 <- read_NPX("~/NPX_file2_location.xlsx")

To demonstrate how bridging works, we will use the example datasets (npx_data1 and npx_data2) from Olink Analyze package. This workflow also uses functions from the dplyr, stringr, and ggplot2 packages.

Check bridging datasets

Explore and have an overview of the datasets that are going to be bridged. For example, plot and compare NPX distribution between datasets.

# Load datasets
npx_1 <- npx_data1 %>%
  filter(!str_detect(SampleID, "CONTROL_SAMPLE")) %>% #Remove control samples
  mutate(dataset = "data1")
npx_2 <- npx_data2 %>%
  filter(!str_detect(SampleID, "CONTROL_SAMPLE")) %>% #Remove control samples
  mutate(dataset = "data2")

npx_df <- bind_rows(npx_1, npx_2)


# Plot NPX density before bridging normalization
npx_df %>%
  mutate(Panel = gsub("Olink ", "", Panel)) %>%
  ggplot(aes(x = NPX, fill = dataset)) +
  geom_density(alpha = 0.4) +
  facet_grid(~Panel) +
  olink_fill_discrete(coloroption = c("red", "darkblue")) +
  set_plot_theme() +
  ggtitle("Before bridging normalization: NPX distribution") +
  theme(axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        strip.text = element_text(size = 16),
        legend.title = element_blank(),
        legend.position = "top")

Use PCA plot to visualize sample-to-sample distance before bridging. Typically the project dataset accounts for most of the observed variation within the combined datasets at this point.

## before bridging

#### Extract bridging samples
npx_1 <- npx_data1 %>%
  mutate(dataset = "data1")
npx_2 <- npx_data2 %>%
  mutate(dataset = "data2")

overlap_samples <- intersect(npx_1$SampleID, npx_2$SampleID) %>%
  data.frame() %>%
  filter(!str_detect(., "CONTROL_SAMPLE")) %>% #Remove control samples
  pull(.)


### Generate unique SampleIDs
npx_1 <- npx_data1 %>%
  mutate(dataset  = "data1",
         SampleID = paste0(dataset, PlateID, SampleID))

br_1 <- npx_data1 %>%
  mutate(dataset = "data1") %>%
  filter(SampleID %in% overlap_samples) %>%
  mutate(SampleID = paste0(dataset, PlateID, SampleID))

npx_2 <- npx_data2 %>%
  mutate(dataset  = "data2",
         SampleID = paste0(dataset, PlateID, SampleID))

br_2 <- npx_data2 %>%
  mutate(dataset = "data2") %>%
  filter(SampleID %in% overlap_samples) %>%
  mutate(SampleID = paste0(dataset, PlateID, SampleID))

npx_before_br <- rbind(npx_1, npx_2) %>%
  mutate(datatype = ifelse(SampleID %in% c(br_1$SampleID, br_2$SampleID),
                           "bridges", dataset))



### PCA plot
OlinkAnalyze::olink_pca_plot(df          = npx_before_br,
                             color_g     = "datatype",
                             byPanel     = TRUE)

PCA plot of combined datasets before bridging

Perform bridging between two data sets

We can use olink_normalization() function to bridge two datasets. The bridging procedure is to first calculate the median of the paired NPX differences per assay between the bridging samples as adjustment factor then use these adjustment factors to adjust NPX values between two datasets. In this process, one dataset is considered the reference dataset (df1) and its NPX values remain unaltered. The other dataset is considered the new dataset (df2) and is adjusted to the reference dataset based on the adjustment factors.

The output from olink_normalization() function is a NPX table with adjusted NPX value in the column NPX.

olink_normalization() uses the information from column Project to distinguish between reference dataset from the other dataset. It is up to the user to define which dataset is the reference dataset and specify the names of the bridge samples.

# Find shared samples
npx_1 <- npx_data1 %>%
  mutate(dataset = "data1")
npx_2 <- npx_data2 %>%
  mutate(dataset = "data2")

overlap_samples <- intersect(npx_1$SampleID, npx_2$SampleID) %>%
  data.frame() %>%
  filter(!str_detect(., "CONTROL_SAMPLE")) %>% #Remove control samples
  pull(.)

# Perform Bridging normalization
npx_br_data <- olink_normalization(df1                     = npx_1,
                                   df2                     = npx_2,
                                   overlapping_samples_df1 = overlap_samples,
                                   df1_project_nr          = "20200001",
                                   df2_project_nr          = "20200002",
                                   reference_project       = "20200001")
glimpse(npx_br_data)
#> Rows: 61,824
#> Columns: 19
#> $ SampleID      <chr> "A1", "A2", "A3", "A4", "A5", "A6", "A7", "A8", "CONTROL…
#> $ Index         <int> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 1…
#> $ OlinkID       <chr> "OID01216", "OID01216", "OID01216", "OID01216", "OID0121…
#> $ UniProt       <chr> "O00533", "O00533", "O00533", "O00533", "O00533", "O0053…
#> $ Assay         <chr> "CHL1", "CHL1", "CHL1", "CHL1", "CHL1", "CHL1", "CHL1", …
#> $ MissingFreq   <dbl> 0.01875, 0.01875, 0.01875, 0.01875, 0.01875, 0.01875, 0.…
#> $ Panel_Version <chr> "v.1201", "v.1201", "v.1201", "v.1201", "v.1201", "v.120…
#> $ PlateID       <chr> "Example_Data_1_CAM.csv", "Example_Data_1_CAM.csv", "Exa…
#> $ QC_Warning    <chr> "Pass", "Pass", "Pass", "Pass", "Pass", "Pass", "Pass", …
#> $ LOD           <dbl> 2.368467, 2.368467, 2.368467, 2.368467, 2.368467, 2.3684…
#> $ NPX           <dbl> 12.956143, 11.269477, 25.451070, 14.453038, 7.628712, 6.…
#> $ Subject       <chr> "ID1", "ID1", "ID1", "ID2", "ID2", "ID2", "ID3", "ID3", …
#> $ Treatment     <chr> "Untreated", "Untreated", "Untreated", "Untreated", "Unt…
#> $ Site          <chr> "Site_D", "Site_D", "Site_D", "Site_C", "Site_C", "Site_…
#> $ Time          <chr> "Baseline", "Week.6", "Week.12", "Baseline", "Week.6", "…
#> $ Project       <chr> "20200001", "20200001", "20200001", "20200001", "2020000…
#> $ Panel         <chr> "Olink Cardiometabolic", "Olink Cardiometabolic", "Olink…
#> $ dataset       <chr> "data1", "data1", "data1", "data1", "data1", "data1", "d…
#> $ Adj_factor    <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…

Perform evaluation of bridging normalization

First, check NPX distribution in datasets after bridging normalization.

# Plot NPX density after bridging normalization

npx_br_data %>%
  mutate(Panel = gsub("Olink ", "", Panel)) %>%
  ggplot(aes(x = NPX, fill = dataset)) +
  geom_density(alpha = 0.4) +
  facet_grid(~Panel) +
  olink_fill_discrete(coloroption = c("red", "darkblue")) +
  set_plot_theme() +
  ggtitle("After bridging normalization: NPX distribution") +
  theme(axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        strip.text = element_text(size = 16),
        legend.title = element_blank(),
        legend.position = "top")

Then summarize number of assays that have adjustment factors in certain ranges. High adjustment factors can result from variations between projects, such as panel versions or technical modifications. Such assays can be visualized individually with violin plots and may warrant further investigation to confirm they are still comparable between projects.

breaks <- c(-Inf, -1, -0.25, 0.25, 1, Inf)
labels <- c("[< -1]", "(-1,-0.25]", "(-0.25,0.25]", "(0.25,1]", "(>1]")

# reference project defined in bridging normalization
reference_project <- "20200001"
npx_br_data %>%
  mutate(Panel = gsub("Olink ", "", Panel)) %>%
  # remove adjustment factors from reference project.
  filter(Project != reference_project) %>%
  mutate(group = cut(Adj_factor, breaks = breaks,
            labels = labels)) %>%
  select(OlinkID, Adj_factor, Panel, group) %>%
  distinct() %>%
  group_by(Panel, group) %>%
  tally() %>%
  tidyr::pivot_wider(names_from = group, values_from = n) %>%
  tibble::column_to_rownames(var = "Panel") %>%
  relocate(any_of(labels)) %>%
  kbl(booktabs = TRUE,
      digits = 2,
      caption = paste("Distribution of assays in different ranges of",
                      "adjustment factors")) %>%
  kable_styling(bootstrap_options = "striped", full_width = FALSE,
                position = "center", latex_options = "HOLD_position")

Distribution of assays in different ranges of adjustment factors
	[< -1]	(-1,-0.25]	(-0.25,0.25]	(0.25,1]	(>1]
Cardiometabolic	6	15	36	18	17
Inflammation	13	18	39	17	5

Finally, use PCA plot to check whether bridging normalization has effect in correcting batch effects. In the example below, it is clear that before bridging samples from data 1 and 2 are divided into separate clusters due to the batch effects, but after bridging they are shown as one cluster in the PCA plot. Bridging normalization has sufficiently removed the batch effects between two data sets.

## After bridging

### Generate unique SampleIDs
br_ids <- npx_before_br %>%
  select(SampleID, OlinkID, datatype)
npx_after_br <- npx_br_data %>%
  mutate(SampleID = paste0(dataset, PlateID, SampleID)) %>%
  left_join(br_ids, by = c("SampleID", "OlinkID"))
### PCA plot
OlinkAnalyze::olink_pca_plot(df          = npx_after_br,
                             color_g     = "datatype",
                             byPanel     = TRUE)

PCA plot of combined datasets after bridging

Introduction to bridging Olink® NPX datasets

Compiled: November 16, 2022

Setup bridging datasets

Check bridging datasets

Perform bridging between two data sets

Perform evaluation of bridging normalization