¶ Submitting jobs
The essence of ariadne.ai SPATIAL is to provide a worry- and code-free analysis platform for spatial biology analysis. All anaylsis steps are packed into easily understandable jobs. To analyze your data submit jobs in the following order.
¶ 1. Registration
Registration is the first crucial preprocessing step to ensure best results during later analysis. Our elastic registration is scalable and highly precise.
To submit a registration job for your data click on the NEW JOB button in the main window and select Registration.
The window above will open. Now select a target layer. This can be an H&E layer or for example the DAPI layer of your first staining cyle. Afterwards select all layers you want to register onto the target layer.
¶ 2.A Cell Segmentation
The essence of spatial biology is mapping the detections of omics markers to single cells or even smaller subcellular compartments. To achieve this, a very precise segmentation of the cells is necessary.
The segmentation in SPATIAL allows precise segmentation of histology data and fluorescence microscopy data in 2 and 3 dimensions. We even support highly compley cell types like neurons.
To start a segmentation job click on the NEW JOB button in the main window and select Segmentation. The window above will open. Now select the Target layer which will be used to perform the segmentation. Additionally, you can also choose to add a Nuclear layer, which will be used together with the target layer.
¶ 2.B Nucleus Expansion
In a nucleus expansion job in SPATIAL, the process begins by segmenting the nuclei within the data. After segmentation, the algorithm expands these nuclear masks to cover the entire cell, achieving full cell segmentation.
To start, click NEW JOB and select Nucleus Expansion. Choose the Nuclear Layer to segment, and the algorithm will expand the nuclei to generate the complete cell segmentation.
¶ 3. Intensity Map
To map your marker detections to the segmented cells, start a marker mapping job. Click on the NEW JOB button in the main window and select Intensity Map. The window above will open. Now select a segmentation as target layer onto which the marker intensities from the source layers will be mapped.
¶ 4. Recurrent Neigborhoods
Recurrent cellular neighborhoods (RCN) are calculated using the cells present in the "selections". Choose the selections that you want to include in the RCN calculation. If your selections do not cover all the cells but you want to keep them in the calculation, tick Keep unassigned. This will group all the cells that are not part of the selections under the the name "NA" and treat them as another selection. Please note that the results will vary depending on the choice for keeping unassigned cells. Then, select a cutoff radius. All cells that are closer than the cutoff radius will be considered connected. Finally, choose the number of RCNs that you would like to obtain. A smaller value can give more coarse RCNs while larger numbers will yield to smaller but finer RCNs. Each recurrent neighborhood calculation can be reached via their Experiment Name
You can access to the recurrent neighborhood results using the right panel -> Click + -> Select Recurrent neighborhoods . Your experiments will be accessible via the dropdown menu. The details of each experiment is accessible via the Details tab. You can also download the results as a zip file using the download button. Moreover, the calculated recurrent neighborhoods will also be available under the "Selections" panel.
¶ 5. Neighborhood enrichment
The Neighborhood enrichment score is calculated between a set of selections in a pairwise manner. Choose the "selections" you want to calculate the neighborhood enrichment score for. If your selections do not cover all the cells but you want to include them in the computation, click Keep unassigned. This will group all the cells that are not part of the selections under the name "NA" and treat them as another selection. Please note that this does not affect the neighborhood enrichment score between any other selections. Then, select the cutoff radius. All cells that closer than the cutoff radius will be considered connected. Each neighborhood enrichment score calculation can be reached via their Experiment Name
You can access to the neighborhood enrichment score results using the right panel -> Click + -> Select Neighborhood enrichment . Your experiments will be accessible via the dropdown menu. The details of each experiment are accessible via the Details tab. You can also download the results as a zip file using the download button.
¶ 6. Automatic cell type detection
This method is currently under heavy development and will be available in the upcoming versions.
¶ 7. UMAP and T-SNE
SUbmitting UMAP and T-SNE jobs on the image features will be available in the upcoming versions.
¶ Optional custom jobs (contact us)
¶ Virtual marker segmentation
A precise cell segmentation might need not only one, but multiple markers. Get in touch with us to create a virtual marker combined of multiple markers, to get the best possible cell segmentation.
¶ Tissue segmentation
A tissue segmentation based on morphological features of the cells in your tissue provides you with a spatial marker independent tissue classification giving you a great tool at hand to find even subtle expression changes between different experiments. Depending on your tissue type the nuclei alone might be already sufficient.
¶ Stitching
Precise stitching of your image tiles is more impotant than ever, especially when working with point like signals, like in spatial transcriptomics.
We have multiple years of experience with image stitching and will deliver best results for your 2d or 3d tiles, with or without overlap.
¶ Artefact segmentation
Get rid of your imaging artefacts using our custom artefact segmentation. Common artefacts that we remove for our customers are air bubbles, fibers or hairs, dust particles and tissue folds.
¶ Ongoing jobs
To see you ongoing jobs click on the JOBS button in the navigation panel (top left)
¶ Submit batch jobs
You can apply the same job to multiple datasets.
Note that the feature is only available if sibling datasets(within the same folder) exist.