Clonal Compass

Single-cell immune-repertoire co-pilot — paired scRNA-seq + TCR-seq. This is an exploratory research analysis, not a diagnostic one.

2,992Cells
14,077Genes
13Clusters
6Cell types
2,992TCR+ cells

1. Overview figures

Leiden clusters on the transcriptome (UMAP).
Leiden clusters on the transcriptome (UMAP).
Marker-based cell-type / T-cell subset labels.
Marker-based cell-type / T-cell subset labels.
TCR clone size (log1p of cells per clonotype).
TCR clone size (log1p of cells per clonotype).
Clonal-expansion category per cell.
Clonal-expansion category per cell.

2. Interpretation of the most clonally-expanded clusters

Each interpretation is tagged with the path that produced it. Claims are hedged and cite the specific metric behind them.

Cluster 4: CD8 cytotoxic effector

Claude API
Provisional labelCD8 cytotoxic effector
Cells in cluster337
Top marker genes (rank_genes_groups score)NKG7 (+20.83), GZMH (+18.76), COX3 (+18.07), GZMA (+17.30), ATP6 (+16.64), ND5 (+16.47), RPL37A (+15.90), COX1 (+14.89)
TCR coverage337 / 337 (100%)
Clonal expansion170 cells in clones > 1 (50% of TCR+ cells)
Largest clone sizes here20, 20, 20, 20, 20
Interpretation generated by: Claude API

The provisional "CD8 cytotoxic effector" label is consistent with the cytotoxic marker profile: NKG7 (+20.83), GZMH (+18.76), and GZMA (+17.30) are all strongly enriched, which is suggestive of an effector cytotoxic program. I'd note a caveat: several top "markers" are mitochondrial or ribosomal genes (COX3 +18.07, ATP6 +16.64, ND5 +16.47, COX1 +14.89, RPL37A +15.90), which can reflect metabolic activity but may also indicate stress or technical/quality effects, so the cytotoxic signal should be weighted toward NKG7/GZMH/GZMA rather than these. TCR recovery is complete (337/337 cells, 100%), giving high confidence in the clonality metrics. Clonal expansion is moderate-to-substantial, with 170/337 cells (50%) in clones of size > 1 and several sizable clones (five clones of size 20), which is consistent with antigen-driven expansion but does not identify the antigen. No VDJdb epitope match was found for this cluster's expanded clones — this is a normal, valid result and simply means no specificity can be predicted here, not that the analysis failed. Overall the data are suggestive of an expanded CD8 cytotoxic effector population, but this is an exploratory research interpretation only, with no functional or clinical claim implied.

No VDJdb match for this cluster’s expanded clones. Absence of a match is a normal, valid outcome — not a failure.

Cluster 6: CD8 cytotoxic effector

Claude API
Provisional labelCD8 cytotoxic effector
Cells in cluster447
Top marker genes (rank_genes_groups score)CCL5 (+20.89), CCL4 (+16.43), DUSP2 (+16.14), NKG7 (+15.42), CST7 (+14.84), TMSB4X (+14.43), CCL4L1 (+14.25), GZMA (+13.97)
TCR coverage447 / 447 (100%)
Clonal expansion149 cells in clones > 1 (33% of TCR+ cells)
Largest clone sizes here10, 10, 10, 8, 8
Interpretation generated by: Claude API

The marker profile is consistent with a CD8 cytotoxic effector phenotype, supported by high scores for cytotoxicity/granule genes NKG7 (+15.42), GZMA (+13.97) and CST7 (+14.84), alongside the chemokines CCL5 (+20.89) and CCL4 (+16.43). These are activation/effector-associated transcripts, so the provisional "cytotoxic effector" label is reasonable but remains marker-based rather than functionally proven. TCR recovery is complete (447/447 cells, 100%), so clonality estimates here are well-powered. There is moderate clonal expansion — 149 cells (33% of TCR+ cells) sit in clones larger than one — which is suggestive of an antigen-experienced, partially expanded population rather than a monoclonal outgrowth. The largest clones are modest in size (10, 10, 10, 8, 8), consistent with several parallel expansions instead of one dominant clone. Two cells carry CDR3 sequences that are exact VDJdb matches to Influenza A epitopes (VSDGGPNLY, 1 cell; GILGFVFTL, 1 cell); these are predicted/possible specificities only and are not confirmed functional reactivities. Note that only 2 of 447 cells have any epitope match, so the antigen specificity of this cluster is largely undetermined — an expected and valid outcome, not a gap in the data. Any functional reading here is exploratory and research-level only, with no clinical or patient-level interpretation implied.

Predicted epitope matches (VDJdb, exact CDR3 aa identity) — predicted specificity, not confirmed reactivity:
  • InfluenzaA: VSDGGPNLY (1 cell)
  • InfluenzaA: GILGFVFTL (1 cell)

Cluster 5: CD8 cytotoxic effector

Claude API
Provisional labelCD8 cytotoxic effector
Cells in cluster182
Top marker genes (rank_genes_groups score)NKG7 (+16.93), CST7 (+15.06), ND4 (+14.87), HLA-C (+13.90), GZMH (+13.68), CCL4 (+13.05), PRF1 (+12.70), PNRC1 (+12.58)
TCR coverage182 / 182 (100%)
Clonal expansion108 cells in clones > 1 (59% of TCR+ cells)
Largest clone sizes here13, 13, 13, 13, 13
Interpretation generated by: Claude API

The provisional "CD8 cytotoxic effector" label is consistent with the marker profile: this cluster shows high cytotoxicity- and effector-associated scores including NKG7 (+16.93), CST7 (+15.06), GZMH (+13.68), PRF1 (+12.70), and CCL4 (+13.05), which together are suggestive of a cytotoxic program rather than proof of a defined functional state. Note that ND4 (+14.87) is mitochondrially encoded and HLA-C (+13.90) is broadly expressed, so neither strongly supports a specific effector identity and both warrant caution. TCR coverage is complete (182/182 cells, 100%), which strengthens confidence in the clonal metrics below. Clonal expansion is moderate: 108 of 182 TCR+ cells (59%) fall in clones of size > 1, consistent with an antigen-experienced, expanded population. Several sizeable clones are present (multiple clones of size 13), which is compatible with prior antigen-driven proliferation but does not identify the antigen. No VDJdb exact-CDR3 match was returned for the expanded clones — this is a normal, valid result and simply means no predicted epitope can be offered here; it neither confirms nor excludes any specificity. Overall the evidence is suggestive of an expanded, cytotoxic-leaning CD8 population, but this is an exploratory research interpretation only, with no functional or clinical conclusion implied.

No VDJdb match for this cluster’s expanded clones. Absence of a match is a normal, valid outcome — not a failure.

3. Predicted epitope specificities (VDJdb)

42 TCR+ cell(s) carry a CDR3 that exactly matches a VDJdb entry. These are predicted specificities (database hits), not confirmed reactivity:

Matches are exact CDR3 amino-acid hits against VDJdb and indicate possible/predicted antigen specificity only; they require functional validation. A “no match” result is expected and valid.