iSCanGuide Help

Cell–Cell Communication Analysis on a Spatial Thymus Sample

In this case study, we perform a complete cell–cell communication (CCC) analysis on a spatial transcriptomics sample of human thymus tissue in iSCanGuide. Starting from an annotated dataset, we run the consensus CCC pipeline, inspect sender–receiver pairs, drill into ranked ligand–receptor interactions, evaluate spatial co-expression with Moran's I, characterise receiver-side inflow, and compare two runs side by side.

You can import the study in this tutorial with view-permission to your study list using the following link:

https://iscanguide.advaitabio.com/studies/import/Pg47QjGF69cYaxHJ8

Dataset

Field

Value

Study

Thymus_GSM6281320

Sample

Thymus E1

Sample type

Spatial (10x Visium-style)

Cells / spots

1,831

Annotation used

CD4_CD28 (2 cell types: CD4-CD28- T cell, unassigned)

Workflow

Open the existing CCC result

The Analysis workspace opens on the Cell-Cell Communication tab. An existing run on sample Thymus E1, labelled with the CD4_CD28 annotation against the consensus resource, is listed with 2 cell types and 1,233 interactions.

Existing CCC analysis result on the thymus dataset

Click View in the Actions column to open the results.

Cell Type Pairs — global sender/receiver view

The results open on the Cell Type Pairs tab. A chord diagram summarises sender–receiver flow between the two annotated populations (CD4-CD28- T cell and unassigned).

Cell Type Pairs chord diagram and interaction table

The companion table ranks the four directed pairs:

Sender

Receiver

# Interactions

Mean Distance

Interacting

unassigned

unassigned

343

23.5

Yes

unassigned

CD4-CD28- T cell

320

39.5

Yes

CD4-CD28- T cell

unassigned

294

26.7

Yes

CD4-CD28- T cell

CD4-CD28- T cell

276

23.5

Yes

Interpretation: homotypic signalling dominates within each compartment (343 and 276 interactions respectively), but the strongest heterotypic channel runs from the unassigned compartment to CD4-CD28- T cells (320 interactions at a slightly longer mean distance of 39.5).

Every chart supports a built-in export panel for downloading publication-ready images at multiple resolutions.

Interactions — ranked ligand–receptor pairs

Click the Interactions tab. By default, the table shows all sender → receiver pairs ranked by the Rank Aggregate Consensus.

Filter by sender

Apply a sender filter for CD4-CD28- T cell. This returns 11 outgoing pairs; the top-ranked interactions by consensus magnitude significance are LCK → CD8A_CD8B and PSAP → CD1B.

Interactions filtered by sender CD4-CD28- T cell

Filter by receiver

Switch the filter to receiver CD4-CD28- T cell. This returns 24 incoming pairs. The strongest signals are B2M → CD1B (magnitude significance 9.45×10⁻⁵), LCK → CD8A_CD8B, and several HLA class I ligands signalling onto CD8A/CD8B/CD3D.

Interactions filtered by receiver CD4-CD28- T cell

Interpretation: the unassigned compartment functions as an antigen-presenting niche (HLA-B, B2M) signalling to T-cell receptor and co-receptor machinery on the CD4-CD28- T cells. The LCK/CD8 autocrine signalling within the T-cell compartment reflects TCR-proximal kinase activity.

Spatial Co-expression — Moran's I

Click the Spatial Co-expression tab. This tab evaluates which ligand–receptor pairs are physically co-located on the tissue.

With the default filter of Moran's I ≥ 0.01, four pairs survive:

Ligand

Receptor

Moran's I

p-value

Mean Cosine

Std Dev

LCK

CD8A_CD8B

0.172

< 0.001

0.940

0.043

CD99

PILRB

0.088

< 0.001

0.593

0.102

CCL25

CCR9

0.069

< 0.001

0.770

0.084

PSAP

CD1B

0.016

< 0.001

0.942

0.039

Spatial co-expression filtered to Moran's I ≥ 0.01, showing 4 pairs

Interpretation:

  • LCK → CD8A_CD8B is the most spatially clustered axis (Moran's I = 0.172, Mean Cosine = 0.94), consistent with a localised lobular T-cell zone.

  • CCL25 → CCR9 is the canonical thymocyte-homing chemokine axis — a clear spatial signal even at moderate Moran's I.

Removing the Moran's I filter shows all 21 pairs, including those with negative Moran's I (anti-co-localised) such as HLA-E → CD8A and B2M → CD1A.

Spatial co-expression — full distribution including anti-co-localised pairs

To overlay a pair on tissue, select its row, enter a name in Export selected pairs to Features tab, and click Save export.

Inflow — receiver-side signalling pressure

Click the Inflow tab. This tab quantifies how much signalling each receiver cell receives from each ligand–receptor pair.

With sender CD4-CD28- T cell, Moran's I ≥ 0.01, and FDR p-value ≤ 0.05 applied, the bar chart ranks the top pairs by Moran's I — led by LCK^CD8A_CD8B, PSAP^CD1B, EFNA1^EPHB6, and CXCL12^CXCR4.

Inflow tab — filtered LR pair with Moran's I bar chart

Switching the right-hand visualization to a violin plot shows the distribution of Moran's I across all LR pairs per sender. Both senders produce skewed distributions with a long upper tail crossing the α = 0.01 threshold — a small subset of incoming channels are very spatially organised, while most are diffuse.

Inflow tab — violin plot of spatial variability by sender

Select rows and click Save export to push them to the Features tab for downstream tissue overlay.

Creating a new analysis with Advanced Parameters

Switch to the New Analysis tab. Configure the same run to reproduce the existing result:

Parameter

Value

Sample

Thymus E1

Cell type labels from

Cell Annotation

Cell Annotation

CD4_CD28

Ligand-receptor resource

Consensus

Expression proportion

0.10

Min cells per type

6

Permutations

1000

Spatial bandwidth

Auto-detect

Methods

CellPhoneDB, Connectome, log2FC, NATMI, SingleCellSignalR

Selecting the ligand-receptor resource in the New Analysis form

Click Run Analysis. A confirmation toast appears pointing to the study log for progress.

Job submission confirmation toast

The same workflow can be re-run against a finer annotation than the binary CD4_CD28 split — the annotations visible in the left panel (Naive T cell, Megakaryocyte, Memory B cell, Th17/Th2/Th1 cells, MSC, M1/M2 macrophage, etc.) would resolve the "unassigned" compartment into its constituent thymic populations and reveal more granular interactions.

Comparing two runs side by side

When the second job finishes, both runs appear in Existing Results with their timestamps and per-run interaction counts.

Two CCC runs listed under Existing Results

Open a Side-by-Side view to place both runs next to each other across the embedding and spatial panels, with a third panel showing the highlighted L–R pair LCK^CD8A_CD8B projected on the tissue.

Side-by-side comparison of two runs with the LCK^CD8A_CD8B feature overlaid on tissue

Summary

The CCC module in iSCanGuide produces an end-to-end view of cell–cell communication on a spatial sample:

  • The consensus engine identifies a thymic antigen-presentation ↔ T-cell signalling programme (B2M/HLA-B → CD1B/CD3D/CD8A) plus intra-T-cell TCR-proximal signalling (LCK → CD8A_CD8B) as the dominant axes.

  • Adding the spatial layer narrows the biologically interesting set: only four of 324 candidate pairs have meaningful spatial co-expression at Moran's I ≥ 0.01, led by LCK → CD8A_CD8B and the thymocyte-homing CCL25 → CCR9 axis.

  • The Inflow analysis confirms that spatial structure is sender-dependent and concentrated in a small tail of ligand–receptor pairs.

  • The whole workflow is reproducible from the Advanced Parameters panel and can be re-run on richer cell-type annotations without leaving the iSCanGuide interface.

Last modified: 21 May 2026