Cluster analysis workflow overview: cell population identification¶

"Analysis" is the process of identifying cell populations or labelling cells of interest with some cell population label. The analysis comprises of the steps:

• A. "Workflow" specification.
• B. "Preprocessing" of all files.
• C. "Cleaning" of irrelevant particles. For example, you may want to remove the doublets, and dead cells, and isolate only the leukocytes.
• D. "Clustering" or labelling of cell populations. For example, if you isolated the lymphocytes in the previous step, this step further clusters the leukocytes such that you can label its sub-populations as T-cells and B-cells.

---

A. Workflow specification¶

Aim: You will specify the parameters for the next three steps in the analysis.

Go to your Experiments page, Files tab. See section "To view your Experiment" @ Files .

1. Select (click on checkboxes) the file(s) you want to analyze.
2. Click "Set up Workflow". In the pop-up:
   • Section "Type analysis name": Enter a workflow name; this will help you keep track of each analysis as you will be able to analyze files multiple times.
   • Section "Select reference file": Select one file that will be used to batch normalize the other files. Batch normalization is only applied to create dimensionality-reduced visualizations when all files are concatenated together in the meta-analysis step i.e. we preserve the original file throughout the analysis and for calculating statistics.
   • Section "Select cytometer you used": Select the machine that processed the files you uploaded for an optimized analysis experience. Metaflow can handle files from different types of cytometers.
   • Section "Select pre-processing or transformation": This section corresponds to step B. "Preprocessing".
     • "Normalization": Select this option if you want your heatmap plot to have a normalized colour range across all channels.
     • "Identification and removal of anomalies derived from abrupt changes in flow rate or instability of signal": Select this option if you want Metaflow to automatically remove anomalies as specified.
     • "Select display transformation": Select "None" or one of the transformation options to transform your data into a linear scale.
   • Section "Select channels for cleaning": This section corresponds to step C. "Cleaning". Select the morphological and fluorescent channels you want to use to remove particles not of interest to you.
   • Section "Select channels for clustering": This section corresponds to step D. "Clustering". Select the morphological and fluorescent channels you want to use to identify cell populations of interest to you. Steps C. and D. are separated to ensure that only relevant channels are used. This prevents irrelevant channels from affecting your analysis.
   • Click "Start Workflow".

Once you have started a workflow, Metaflow will run the preprocessing and clusterings for steps B., C., and D. in the background.

To check the status of your Workflow: 1. Click on "Workflows". This will bring you to the Workflows section of your Experiment page . 2. Click on the icon on the right end of your Workflows' row.

Click to see example

---

B. Preprocessing¶

Preprocessing is done automatically without the need for human intervention.

---

C. Cleaning¶

Aim: You will identify the particles you want to analyze from your files.

Before starting, see the section "To check the status of your Workflow" in the previous section. After opening up your Workflow, under the "Status" column,

• you will see the status "Processing" if Metaflow is still processing your workflow.
• If Metaflow is finished processing, you will see the status "Clustering done".
• Once you see the "Clustering done" status, your files are ready for you to analyze. After you analyze the file and are ready to move to the next step, the status will change to "Analysis done".

Cell population identification: Once you see the "Clustering done" status for your files in the "Cleaning" step,

1. Click on the name of your first/reference file to start the analysis. Doing so will take you to your files' analysis page .
2. The analysis page provides a familiar interface for those who are used to manual 2D gating but with the added advantage of multi-dimensional clustering-guided cell population identification. For example, to isolate singlets:
   • Adjust aes: On your first 2D plot/graph, click on the axes names to select the axes you want to visualize your particles with. For example, you can select axes FSC-A and FSC-H.
   • Create lasso: Click on the lasso icon on the top right corner of the graph and encircle the points that resemble singlets. Give your lasso a name e.g. "Singlets".
   • Isolate cells: You can right-click on the completed lasso and double-click OR right-click and select "Show clusters on new graph" to create a new graph containing only the points that belong to the clusters (indicated by the different colours) whose centre belong inside your lasso.
3. Repeat the previous step to isolate live and CD45+ cells.
   • Editing your lasso: If you feel the clusters are too large, you can drag the slider under your graph to the right to increase the "depth" or the number of clusters you see.
   • Sub-analysis: If not even the largest depth can identify the cell population you want to find, click on "Sub-analysis" on the top ribbon menu to initiate clustering on a specific subset of clusters (in section "Select clusters") and channels (in section "Select channels").
4. If you feel that the CD45+ cells are the cells you want to carry forward to step D.,
   • scroll to the bottom of the page and click "Select clusters of interest and channels for the next analysis step."
   • Scroll down and make sure the events on the newly displayed graph are the ones you want to further analyze in the next step. If not, go to the "Cluster list" below to adjust event selection.
   • On the right, select the channels you want to use for the next analysis step. Only select the ones you need for best results.
   • Click "Save the analysis" and you are done!

Edit-protect your analysis: At any stage, if you do not want anyone to edit your analysis, click the "Template unprotected" button. Click on the button again to continue editing.

Transfer analysis¶

If you have more than one file,

• you can analyze the next file by clicking "Next" OR
• Click on "Transfer" to transfer your current analysis to all your other files. You will still want to click through your files and confirm the transfer by clicking "Save the analysis" for all of your files. This is to ensure that you have confirmed the transfer applied.

Initiate step D. Clustering¶

After you have finished cleaning all of your files:

1. Go back to the Workflows page by clicking on the icon on the top left corner of the analysis page. You should see, under the Status column, that all your files have had their "Analysis done".
2. Click on "Cleaned, start clustering" to initiate step D.

Export analysis¶

If you want to export your analysis graphs, clusters, cell population statistics, etc.,

1. Click "Export" and select the export file you want to download. Metaflow will start preparing the exported file for you.
2. Click on your name in the top right corner of the page.
3. Click "Exports".
4. After your file's export "Status" displays "Success", click on the icon and select "Download" to download the file, or select "Delete" to remove the file from your Exports page.

You can also go to the Workflow page and click "Batch statistic export" to export statistics for all the files at once.

---

D. Clustering¶

Repeat the processes done in step C.. Make sure to give all your cell populations a name by either saving your lasso or renaming your cluster.