User Tutorial

Setup¶

In the rawData directory for the project, create a directory titled “h5”. Navigate to the directory and create softlinks for the h5 files (absolute file path is generally helpful). From inside the directory, this can be done for each file with the command:
ln -s absolute/path/to/sample/outs/filtered_feature_bc_matrix.h5 renamed_sample.h5

If there are too many samples to repeatedly type this command, a bash script can be generated and run to create the softlinked files (assuming that these paths are generated from CellRanger outputs):

echo '#!/bin/bash' > make_softlinks.bash;
for file in /path/to/experiment/*/outs/filtered_feature_bc_matrix.h5; do
    sample=$(echo $file|sed "s~.*experiment/~~g"|sed "s~/outs.*~~g");
    echo ln -s $file "$sample".h5;
    done >> make_softlinks.bash

The h5 directory should now contain softlinks for all the h5 files, pointing to their "true" location. This can be checked with the ls -lh command from inside the h5 directory.

lrwxrwxrwx. 1 user Group  16 Jul  2  2020 sample1_NR.h5 -> /path/to/sample1_NR/outs/filtered_feature_bc_matrix.h5
lrwxrwxrwx. 1 user Group  15 Jul  2  2020 sample1_R.h5 -> /path/to/sample1_R/outs/filtered_feature_bc_matrix.h5
lrwxrwxrwx. 1 user Group  16 Jul  2  2020 sample2_NR.h5 -> /path/to/sample2_NR/outs/filtered_feature_bc_matrix.h5
lrwxrwxrwx. 1 user Group  15 Jul  2  2020 sample2_R.h5 -> /path/to/sample2_R/outs/filtered_feature_bc_matrix.h5

In the GUI, follow the standard usage to select the scRNASeq pipeline and an organism genome of interest. Current supported genomes are GRCh38 (human) and mm10 (mouse).

Set the data directory the parent directory containing the h5 directory, not the h5 directory directly. You should see the h5 directory in the selection window:

Create the working directory as normal and “Initialize Directory”. If set up properly, symlinks to the h5 files should be created within the working directory.

Initial Quality Control¶

Clustering¶

The clustering algorithm can be chosen from one of three options:
1. Smart Local Moving Algorithm (default)
2. Original Louvain algorithm
3. Louvain algorithm with multilevel refinement

Multiple clustering resolutions can be used, ranging from 0 to 2, where lower resolution values will result in larger and fewer clusters. The default values of 0.4, 0.6, 0.8, 1.0, and 1.2 should cover the majority of clustering resolutions necessary.

Cell Type Annotation¶

The annotation databases provided by SingleR are species dependent and are populated in response to the species selected. Five human databases are included by default : * Human Primary Cell Atlas (non-specific) * Blueprint and ENCODE (non-specific) * Database for Immune Cell Expression (DICE) (immune) * Monaco Immune Cell Data (immune) * Novershtern Hematopoietic Cell Data (hematopoietic & immune)

Two mouse databases are also provided: * ImmGen (immune) * mouseRNASeq – A collection of mouse data (non-specific) All relevant databases are used for cell type identification; the choice of database selects the default reference used in the preliminary QC plots.

Sample Tab Files¶

As before, groups.tab contains three columns for each sample:

FileNameHeader(noExtension)    GroupID    SampleAlias

contrasts.tab contains two columns:

Group1    Group2

Contrasts are calculated as Group1-Group2.

For example, if there are 4 samples in 2 groups, with file names Sample1.h5, Sample2.h5, etc., groups.tab would contain the following:

Sample1    Group1    S1
Sample2    Group2    S2
Sample3    Group1    S3
Sample4    Group2    S4

and to determine Group2-Group1, contrasts.tab would contain:

Group2    Group1

CITESeq¶

CITESeq is a recent addition to single-cell technologies where antibody capture is used in conjunction with scRNASeq. This allows for more direct correlation to traditional cell sorting techniques, where surface markers are used to identify cells. By selecting this option, it retrieves the CITESeq data from the h5 object and performs relevant scaling and normalization, per Seurat recommendations.

Dry run the Pipeline¶

After setting up the data directory, the working directory, the groups.tab file, the contrasts.tab file, and all necessary options, click the Dry Run button. This will launch a preliminary pipeline check to ensure that all necessary files are present and accessible. A new window will open showing the steps that will be run in the pipeline. Scroll to the end of the dry run to confirm that the process names and number of processes run are identical at the beginning and end.

Top of Dry Run	End of Dry Run

Run the scRNASeq Pipeline¶

If the dry run checks out, click the Run button. This produces the following popup:

Click OK to launch the pipeline. Users will be notified by email when the run is completed.