Running AtlasXBrowser

Jen Garbarino
31 steps
6 minutes

Latch

AtlasXbrowser was built to help facilitate the image processing aspect of the DBiT-seq workflow. Taking in both a brightfield and epifluorescent image of the chipped tissue, AtlasXbrowser guides the user through the steps necessary to generate a “Spatial Folder”. This folder contains the output of the image processing in 10x Visium image data format, allowing for easy incorporation into bioinformatics packages used in spatial seq analysis.

Click on the pods icon

Click start on the AtlasXBrowser pod

Once available click the jupyter labs icon

Type "abrowser [[enter]]" in to the terminal seen below.

It will take a couple seconds for the app to begin loading but once you see text generating you can close this tab and go back to the latch workspace

Click on the open custom app button to open AtlasXBrowser in a new tab.

Click on the magnifying glass to select your sample

Your image should be in its own folder in the spatials folder in your workspace for AtlasXbrowser to access

Click on the sample you are interested in or type the name in the line at the top to filter the drop-down results

Select the appropriate barcode file for your assay.

bc50v2-24: portal50 barcode orientation (portal50 no longer available). bc96.txt: barcode orientation (portal96 is no longer available for purchase). fg96_11DEC.txt: FlowGel96 barcode orientation (20um resolution) bc22005-OCT.txt: FlowGel220 barcode orientation (works for both 10um and 15um)

You can fill in optional metadata about your sample here. This is not required.

Click "Confirm" when you are ready to start image processing.

Step 1 rotation: Make sure to rotate the image so that the green BSA channels are in the top left of your screen. This accounts for orientation in the microscope\ For portal96 use the 45 degree rotation to make the image a square instead of a diamond.

This image notation assumes that you are imaging the slide face up (as this is how the barcodes are deposited) if you flip the slide upside down for imaging these instructions will change based on how you flipped the slide. if flipping horizontally: green BSA is top right if flipping vertically: green BSA is bottom left

Click "Activate" when you have orientated the slide to the correct orientation.

Cropping:\ Move the red circles to make the blue lines just outside your BSA lines.

Click "Confirm" once you are happy with the crop window.

Next click "Activate" to start the ROI selection

To help select the ROI or any other step you can zoom in and out with the scale bar at the top

For ROI selection move each corner of the red box to the corner of the ROI\ click here.

Click "Display Grid" to check how well the ROI is aligned to the BSA lines.

If you don't like the alignment, click hide grid to bring you back to the ROI selection box. Repeat until desired alignment achieved.

Next click threshold to adjust the image for automatic on/off tissue filtering. Neighborhood and C parameters are used to calculate the local pixel intensity threshold which divides pixels to in datapoints that are on tissue (black) vs off tissue (white)

C value: a constant subtracted from this local mean regulating how far above or below the mean a pixel must be to receive a white classification. Larger values of C directly decrease the threshold value, meaning more pixels will be above this value, and thus more pixels will be classified as white. As such, a C value too large runs the risk of making the image too light, resulting in the classification of tissue as non-tissue. Smaller values of C cause a smaller proportion of the pixels to be above this threshold, leading to many pixels being designated as black. As such, a C value too small runs the risk of misclassifying off-tissue sites as on-tissue.

Neighborhood: How many pixels are taken into account when calculating the threshold value for a given region. A neighborhood value too large takes into account too much of the image, and when factors such as lighting are non-uniform, this can produce nonoptimal results. A neighborhood value too small does not take enough of the image into account and so the threshold value is somewhat “overfit” to the region.

Once you have a threshold that well represents the tissue click autofill to automatically fill in regions determined to be tissue

To see the original image after autofilling click on postB or BSA. To return to the thresholded image click on BW.

click fill or erase to add or remove on tissue regions to the image. The BrSize determines the brush size used for adding or removing multiple datapoints.

Brush size of 20 can be seen below in red circle.

When you are done filling in the regions click generate spatial folder.

The loading bar will come up to indicate progress in folder creation.

Once completed the spatial folder successfully uploaded will pop up in the bottom right of the screen. Close the tab when you are done.

When you are done creating spatial folders, close the app tab and turn off the AtlasXBrowser Pod in the workspace

The spatial folder you created will be located in your workspace. These can be found in the spatials folder. In these folders the tissue_position_list.csv indicates what barcode is located where as well as whether it is considered on tissue. The other files are the actual image used for the analysis and the scaling needed for superimposing the data onto the image.