Running the optimize archR Workflow

Jen Garbarino
21 steps
2 minutes

Latch

Atlasxomics

Optimize archR is used to test clustering parameters and filtering before performing the full analysis. It will generate plots showing the clustering spatially and in UMAP space as well as give show the TSS and number of fragments on average and spatially. This workflow allows for the quickly testing of different clustering parameters and filtering with minimal credits by not performing many of the much more computationally expensive outputs of the create archRProject workflow such as peak calling and motif enrichment.

Click on the workflows button in LatchBio to go to the workflows tab

Click on the optimize archr workflow

To add your sample click add a row

Enter the Run_id here

Run_id should not have any spaces and cannot start with or be a single number ex) 1 or Sample 1 would cause errors but Sample_1 will work

Next associate the fragments file for this Run_id

To find the fragment file in LatchBio, click select file. Navigate to the fragment files and click "Select".

Service run fragment files will be located in the Raw_Data folder.

Enter Condition if applicable to samples. You can leave this blank.

No spaces allowed within condition. If you have two conditions you can separate with a dash (dash cannot be used within condition). example: "post_treatment-non_resonder" will give you two conditions: post_treatment and non-responder. "post treatment-non-responder" will lead to errors.

Next we will input the location of the spatial directory. This folder is generated by AtlasXbrowser and is needed to connect the data to the imaging done of the sample.

Click "Select Folder"

Click on the folder labelled "spatial" within the folder for the Run ID of interest.

Click the "Select" button when you have chosen the spatial folder

Service spatial folders will be located in the Raw_Data folder. If AtlasXBrowser completed in your workspace they will be located in the folder spatials

Set genome for experiment (mm10-mouse, hg38-human) contact AtlasXomics if you need an additional reference genomes added. If using the tutorial or demo datasets, use the mouse genome

Enter a project name to call this workflow execution. Project name cannot contain spaces and will be the name of the output folder for this execution.

Next we will test different clustering parameters to choose the best clustering for our experiment. There are three parameters to adjust which are explained in steps 14-16. For more details on how clustering works in archR go to: <https://www.archrproject.com/bookdown/iterative-latent-semantic-indexing-lsi.html>

LSI (Latent Semantic Indexing) resolution helps determine how the points in UMAP space are distributed. This is from a scale of 0-1 with lower numbers being more conservative. Selecting the right resolution is needed for capturing dataset complexity without oversimplifying, often requiring experimentation to find the optimal balance. You can try multiple values by pressing the + LSI resolution button

LSI varFeatures is how many variable features from the sample are being used for the cluserting. 25,000 is our standard value but you can use more or less features here to add in more information or reduce noise respectively. For very small libraries (typically in Cut&Tag) we recommend using 100,000 features or higher to help maximize the data in the clustering. Click + LSI varFeatures to add an additional parameters to test here.

Clustering resolution dictates which regions of the UMAP are called different clusters. The UMAP will not change but how many regions it identifies within it is determined here. This is on a scale of 0-1 with 1 maximizing the amount of clusters identified. Click + clustering resolution to add more parameters here.

Each combination of factors will be performed so keep in mind how many different options you try or it will take a long time to process.

For additional filtering options click "Hidden Parameters"

This workflow defaults to filtering out points with a TSS less than 2. To raise or lower this cutoff you can change this field.

This workflow defaults to having no minimum fragments count per point. If you would like to increase this to remove low count regions, edit this field.

This guide was done for 1 sample. if you would like to combine multiple samples together click add Row to fill in the information for additional samples.

When you have finished adding all the parameters click launch workflow. You will be taken to the execution page where you can track the progress of your workflow.