Subsampling is a way to limit downstream analysis to events in a particular filter (gate) to your workflow. This is commonly used to down sample the number of events available for downstream tasks.
1. Add a Subsampling Task to your Workflow
In the example below, a clean-up gating task was performed to identify Live cells. We will now apply Subsampling to only have Live cells available in our workflow.
Click Add new child task and select Subsampling from the task selector.
2. Perform Automatic Subsampling
Automatic mode Subsampling allows OMIQ to automatically set sampling counts based on how the task is set-up.
2.1 Equal Sampling Method
Select the Filter that you want to base your subsampling on.
Select Automatic Mode.
Select a Counting Method.
-
Per File applies the Target Count to individual files.
- Total applies the Target Count to the total events of all files.
Input a Target Count.
Select a Sampling Method.
- If Equal is chosen, the sampling count will be limited by the file with the lowest number of events Available in Filter.
In the example above, Equal is chosen. Sample 6 Drug B has the lowest available in filter at 17,769. OMIQ will use this as maximum sampling count per sample.
2.2 Max Equal Sampling Method
- If Max Equal is selected, files with events over the Target Count will be subsampled to the Target Count while files with events below the Target Count will be subsampled to the maximum events Available in Filter.
In the example above, the Target Count is 60,000. Files with more than 60,000 have this value as their Sampling Count while files with less than 60,000 events Available in Filter has all the events as Sampling Count.
3. Perform Subsampling by Metadata
In the Automatic Subsampling mode, you can choose to group the files and subsample by metadata. This allows you to visualize how metadata groups compare to each other when subsampled.
Subsampling by Metadata is only available in Automatic Mode.
Select the Filter that you want to base your subsampling on.
Select Automatic Mode.
Select Group Files by Metadata.
This will allow you to choose a Metadata Dimension. Once selected, a Metadata Column (arrow) is appended to the table. A Summary Table of counts appears on the top right (box with arrow).
The Summary Table is useful to see the count total of each metadata group and the number of files in associated with each metadata group.
In the example above, we have set the Target Count per group to be intentionally high to show the total events for each group as shown on the Summary Table (Red Box With Arrow). This is important to understand how the Sampling Method Per Group and Sampling Method for Files in Group work. You can see the Metadata Dimension the files belong to as denoted by the red arrow.
The Sampling Method Per Group and the Sampling Method for Files in Group influence each other. The reason for this is that OMIQ will take the total in the metadata group and divide this total according to the number of files within the group. So, for example, if you have 6 files in the group, the total target count per group will be divided by the number of files. Depending on the selection in Sampling Method for Files in Group, the total target count per group might not be reached.
Select a Counting Method.
-
Per Group applies the Target Count to each metadata group.
- Total applies the Target Count to the total events of all files.
Input a Target Count.
Select a Sampling Method Per Group.
Select a Sampling Method for Files in Group.
3.1 Select a Sampling Method Per Group
3.1.1 Equal Sampling Method Per Group
If Equal is selected, OMIQ will will take the lowest metadata group total while trying to reach the Target Count.
In the example above, we have set the Target Count per group at 420,000 events. In this particular dataset, the lowest total events was 146,809 events (Drug B), which OMIQ will use for all other groups. The Drug B metadata group has 6 files, which means that 146,809 will be divided by 6, so each file in the Drug B group should have 24,468 events each. The reason for the lower count is that some files in the Drug B group have lower than 24,468 events.
3.1.2 Max Equal Sampling Method Per Group
If Max Equal is selected, OMIQ will attempt to reach the target count set per group. If a particular metadata group has a lower total, then OMIQ will take this as the target count for the group.
In the example above, we have set the Target Count per group at 420,000 events. Looking at the total counts, you would notice that Drug B, DMSO, and Drug A are reflecting their total counts (as shown previously). The Media metadata group has 8 files, which means that 420,000 will be divided by 8, so each file in the Media group should have 52,500 events each. The reason for the lower count is that some files in the Media group have lower than 52,500 events.
3.2 Select a Sampling Method for Files in Group
3.2.1 Equal Sampling Method for Files in Group
If Equal is selected, the sampling count will be limited by the file with the lowest number of events Available in Filter in the group.
In the example above, we have set the Target Count to 420,000 events. If we look at the Metadata Group "Drug B", which has 6 files, 420,000 will be divided equally between the 6 files, which means that each file within Drug B group should have 70,000 events. The lowest Available in Filter for Drug B files is Sample 6 Drug B file with 17,769 events. The Equal method will take this count and apply it to all files in the Drug B group.
3.2.2 Max Equal Sampling Method for Files in Group
If Max Equal is selected, OMIQ will take the maximum number of events Available in Filter in the group to try and reach the Target Count for the group.
In the example above, we have set the Target Count to 420,000 events. If we look at the Metadata Group "Drug B", which has 6 files, 420,000 will be divided equally between the 6 files, which means that each file within Drug B treatment group should have 70,000 events. Since none of the files reach 70,000, OMIQ will take the maximum events Available in Filter for each file.
4. Perform Custom Subsampling
Custom mode Subsampling allows you to set the counts per file.
Select the Filter you want to base your subsample on.
Select Custom Mode.
Click on Copy table as a spreadsheet.
Open a spreadsheet editor and paste the copied table spreadsheet.
Input Counts in the Sampling Count column.
Click Paste table from spreadsheet.
Paste the spreadsheet with custom sampling counts in the box provided. Click Update.
OMIQ will take the maximum available in the filter as the sampling count if a file has a lower max available in filter compared to the custom count inputted.