...

Annotating samples with attributes

The Data tab displays the samples in the project - six Astrocytoma and four Oligodendroglioma tumor samples - with the number of cells in each sample (Figure 5). One of the goals of this analysis will be to compare gene expression in a cell type between the two Glioma subtypes. For this, we need to add an annotation indicating the subtype of each sample. 

...

There is new column, Subtype, in the Data tab, but every samples a value of N/A. Next, we will assign each sample to a subtype. 

• Click Edit attributes 

• Use the drop-down menus to assign each sample to its corresponding subgroup (Figure 8)
  
  

  Sample Name	Subtype
  MGH36	Oligodendroglioma
  MGH42	Astrocytoma
  MGH45	Astrocytoma
  MGH53	Oligodendroglioma
  MGH54	Oligodendroglioma
  MGH56	Astrocytoma
  MGH60	Oligodendroglioma
  MGH64	Astrocytoma

  
  
  

• Once each sample has been assigned to a subgroup, click Apply changes to proceed

...

Filtering cells in single cell RNA-Seq data

With samples imported and annotated, we can begin analysis. 

...

...

The Normalization task dialog will open with available normalization methods in the left-hand panel and a blank right-hand panel that will list our selected normalization steps in order of operation (Figure 11). 

The tutorial data set is taken from a published study and has already been normalized using TPM (Transcripts per million), which normalizes for length of feature and total reads (Wagner et al. 2012). This normalization method is also available in Partek Flow, along with other commonly used RNA-Seq data normalization methods. For more information on TPM and other normalization options, please see the Normalize Counts section of the user manual. In the published study using this data set, after TPM normalization, the authors performed three additional transformations, which we can easily replicate using Partek Flow. 

• Drag Divide by from the left panel to the right panel
• Select Custom value from the Divide by drop-down menu
• Set the Custom value to 10
•  Drag Add from the left panel to the right panel
• Drag Log from the left panel to the right panel 

The normalization dialog is now configured to divide the TPM values of each gene by 10, add 1, then perform a log2 transformation (Figure 12). This will replicate the normalization method in the published study, log2([TPM/10] +1). 

• Select Finish to perform normalization

A Normalize counts task node and a Normalized count data node will be added to the Analyses tab. Initially, the nodes will be semi-transparaent to indicate that they have been queued, but not completed. A progress bar will appear on the Normalize counts task node to indicate that the task is running (Figure 13).

Most tasks can be queued up on data nodes that have not yet been generated, so you can wait for normalization step to complete, or proceed to the next section. 

Filtering cells in single cell RNA-Seq data

An important step in analyzing single cell RNA-Seq data is to filer out low - quality cells. These include doublets and cells damaged during cell isolation. 

• Click on the Normalized counts data nodeClick on on QA/QC section  section of the task menu
• Click on Single cell QA/QC (Figure 1410)
  
  

...

...

...

Image Removed

...

A task node, Single cell QA/QC, is produced.  

...

A task node, Single cell QA/QC, is produced. Initially, the node will be semi-transparaent to indicate that it has been queued, but not completed. A progress bar will appear on the Single cell QA/QC task node to indicate that the task is running. 

• Click the Single cell QA/QC node once it finishes running
• Click Task report on the task menu (Figure 11)
  
  

The Single cell QA/QC report includes interactive violin plots showing the value of every cell in the project on several quality measures (Figure 1712). 

 For For this data set, there are two plots: number of reads per cell and number of detected genes per cell. Typically, there is a third plot showing the percentage of mitochondrial reads per cell, but mitochondrial transcripts were not included in the data set by the study authors.

Each point on the plots is a cells cell and the violins illustrate the distribution of cell values for the y-axis metric. Cells can be filtered either by drawing a gate clicking and dragging to select a region on one of the plots or by setting thresholds using the filters below the plots. Here, we will apply a filter for the number of read counts.

...

The plot will be shaded to reflect the gate. Cells that are excluded will be shown as black dots on both plots (Figure 1813). 

 Because Because this data set was already filtered by the study authors to include only high-quality cells, this read counts filter is sufficient for this tutorial. 

...

A new task, Filter cells, is added to the Analyses tab. This task produces a new Single cell data node (Figure 1914).

...

Most tasks can be queued up on data nodes that have not yet been generated, so you can wait for filtering step to complete, or proceed to the next section. 

Filtering genes in single cell RNA-Seq data

...

• Click the Single cell data node produced by the Filter cells task
• Click Filtering in the task menu
• Click Filter features (Figure 2015)
  
  

There are three categories of filter available - Noise reduction filters, Statitics bsaed filters, and Feature list filters (Figure 2116).

...

• Click the Noise reduction filter check box 
• Set the Noise reduction filter to Exclude features where expression value == 0 in 100% 99% of cells using the drop-down menus and text boxes (Figure 2216)
• Click Finish to apply the filter
  

This produces a Filtered counts data node. This will be the starting point for the next stage of analysis - identifying cell types in the data using the interactive t-SNE plot. 

Normalizing single cell RNA-Seq data

We are omitting normalization in ths tutorial because the data has already been normalized. 

The tutorial data set is taken from a published study and has already been normalized using TPM (Transcripts per million), which normalizes for length of feature and total reads, then transformed as log2(TPM/10+1). This normalization and transformation can be performed in Partek Flow, along with other commonly used RNA-Seq data normalization methods. 

For more information on normalization in Partek Flow, please see the Normalize Counts section of the user manual.  

...