Cell–Cell Communication

Form fields

Mandatory fields are indicated with a red asterisk.

• Sample (required): Select the processed sample to analyze. Only pre-processed samples are available for selection, and each analysis run supports a single sample.

• Cell type labels from (required): Specify the source of the cell type assignments.

  • Sample metadata — Cell type labels are retrieved from a column in the sample's cell metadata (the AnnData obs table). Use this option if cell type annotations were included during sample import (for example, a cell_type column).

  • Cell Annotation — Cell type labels are retrieved from an annotation saved within iSCanGuide, such as results generated by SingleR or manual annotations.

  Selecting an appropriate label source is important, as inaccurate or low-quality annotations may lead to weak or difficult-to-interpret results.

• Cell Annotation or Metadata column (required): Select the specific annotation or metadata column containing the cell type labels.

• Ligand–receptor resource: Choose the reference database of known ligand–receptor pairs used for scoring interactions. Default: Consensus.

Advanced Parameters

Click the > button next to Advanced Parameters to expand the additional settings. For most analyses, the default values are recommended.

• Expression proportion (default: 0.10): Defines the minimum fraction of cells within a cell type that must express a gene for that gene to be considered expressed in the population. Ligands or receptors below this threshold are excluded for that cell type, and the corresponding ligand–receptor interaction is not scored.

  • Increase the threshold (e.g., 0.25) to apply stricter filtering and reduce noise.

  • Decrease the threshold (e.g., 0.05) when working with rare cell populations or sparse datasets.

• Min cells per type (default: 5): Cell types containing fewer cells than this threshold are excluded from the analysis. This helps avoid unstable scores driven by very small populations.

  • Increase the threshold (e.g., 20–50) to exclude very small clusters that may represent doublets or low-quality populations.

• Permutations (default: 1000): Number of random permutations used to calculate p-values for non-spatial analyses. Increasing the number of permutations improves p-value precision but also increases runtime.

  The default value of 1000 is appropriate for most analyses. This parameter applies only to non-spatial datasets; spatial analyses use a different statistical approach and ignore this setting.

• Spatial bandwidth (default: Auto-detect; spatial datasets only): Controls the size of the spatial neighbourhood (Gaussian kernel) used for Moran's I and Inflow score calculations, measured in coordinate units.

  In most cases, the recommended setting is Auto-detect, which selects an appropriate value based on the median nearest-neighbour distance within the dataset.

  • Smaller values emphasize fine-scale local signaling patterns.

  • Larger values capture broader spatial gradients.

• Methods (default: all five): Specifies the individual scoring methods combined to generate the consensus ranking. Keeping all five methods enabled is recommended for the most robust and stable results.

Available methods:

Submitting

Click the Run Analysis button. The job runs in the background; when it finishes it appears under Existing Results. Visit Study Logs to monitor job progress.

Exporting

Hover over the chord diagram to reveal export options. Click the export button to download the chart at multiple resolutions.

Click the export button above the table to download the results as a CSV.

Key columns

• Sender/Receiver: The cell types participating in the interaction, representing the signaling and receiving populations.

• Ligand/Receptor: The ligand–receptor gene complexes involved in the interaction.

• Magnitude/Magnitude rank: Measures the expression-based strength of the interaction. Higher magnitude values indicate stronger signaling activity, while lower magnitude ranks indicate interactions that are stronger relative to all other tested pairs.

• Specificity/Specificity rank: Reflects the permutation-based significance of the interaction within a specific sender → receiver context. Lower specificity ranks indicate interactions that are more specific to that particular cell-type pair.

Filtering and Sorting

• Use the column filter icons on Sender, Receiver, Ligand, or Receptor to focus on specific interactions.

• Adjust the shared filters at the top of the table. The default thresholds are:

  • Magnitude rank ≤ 0.05

  • Specificity rank ≤ 0.05

  If no interactions are returned, consider relaxing the thresholds to ≤ 0.1.

• Use the method dropdown on the right to display scores from a specific analysis method:

  • CellPhoneDB

  • Connectome

  • log2FC

  • NATMI

  • SingleCellSignalR

• Sort results by:

  • Magnitude Significance — lower values indicate more statistically credible interactions.

  • Magnitude — higher values indicate stronger inferred signaling.

Click the export button to download the table as a CSV file.

Columns

• Moran's I: Measures the spatial autocorrelation of the ligand × receptor expression product. Higher values indicate stronger spatial clustering. Default filter: Moran's I ≥ 0.01.

• Moran's p-value: Permutation-based p-value associated with Moran's I.

• Mean Cosine: Measures the cosine similarity between ligand and receptor expression patterns across neighbouring spots or cells. Higher values indicate more similar spatial expression patterns.

• Std Dev: Standard deviation of the cosine similarity scores.

Scatter Plot

The scatter plot on the right displays:

• x-axis: Moran's I

• y-axis: Mean Cosine

Interactions appearing in the upper-right region of the plot are both strongly spatially clustered and highly co-expressed.

Filtering

Enable Use interactions tab filters to apply the filtering criteria from the Interactions tab. This is recommended to restrict results to ligand–receptor pairs that are both statistically significant and spatially co-expressed.

Export Co-expression Pairs to Features

Select rows using the checkboxes, enter a name, and click Save export to push the pairs to the Features explorer in the left panel.

Each exported pair becomes a per-cell feature with value equal to the cosine similarity score for that ligand–receptor pair at each cell. Feature label format: ligand^receptor (e.g., IL1A^IL1R1).

Export Inflow Pairs to Features

Select rows using the checkboxes, enter a name, and click Save export to push pairs to the Features explorer. Two row types can be exported:

• Parent row (unmasked): per-cell value is the inflow score for that sender–ligand–receptor axis across all receiver cells. Label format: Sender: ligand^receptor (e.g., Immune: IL1A^IL1R1).

• Child row (receiver-masked): same inflow score scoped to a specific receiver cell type. Label format: Sender: ligand^receptor → ReceiverCellType (e.g., Immune: IL1A^IL1R1 → Fibroblast).