Acute COVID-19 cohort - peripheral blood mononuclear cells across disease severity and time

CD8/NK-CD8 GZMH CD52

Ontology Term: effector memory CD8-positive, alpha-beta T cell, terminally differentiatedCL:0001062
Synonyms Terms that have the same, or nearly the same, meaning as the selected cell type.: CD8-3, GZMH CD52, TEMRA, GZMH+ CD8 T cell
Category Class of which the cell type is a subclass.: CD8-positive, alpha-beta T cellCL:0000625
Ontology Assessment Disagreements with any aspect of the Cell Ontology.: —
Marker Gene Evidence Gene markers used by researcher to assign this cell annotation.: CD8A, CD3D, GZMH, CD8B, CD52, S100A4, B2M, CD3G, CD3E, SH3BGRL3
Canonical Marker Genes: CD3E, CD8A, GZMH, GZMB, PRF1, KLRG1, CX3CR1, B3GAT1
Negative Marker Genes Genes expected to be absent, or expressed at a notably low level, in these cells, and whose absence supports the assignment of this cell annotation.: TYROBP
Rationale: Cluster 3 of the CD8 T/NK cell lineage (paper subset CD8-3). Terminally differentiated effector memory CD8 T cell (TEMRA): CD8A [0.80], CD8B [0.74], GZMH [0.75], KLRG1 [0.61], CD52 [0.73], with low CCR7 [0.48], SELL [0.43] and CD27 [0.45]; the original annotation notes record CD45RA protein (CITE-seq). More abundant after SARS-CoV-2 infection; 19% of cells carried public TCRs. Numbers in brackets are one-vs-rest AUCs within the CD8 T/NK cell lineage.
Rationale DOI: 10.1101/2025.09.19.25334095

Inherited from COVID_annotation_07062023

Set of Labels Description: Manual annotations of Leiden clusters computed separately within each of four lineages (B cells, CD4 T cells, CD8 T and NK cells, mononuclear phagocytes). For each lineage, PCA scores were aligned across batches with Harmony, each cell was connected to its 50 nearest neighbors, and the Leiden algorithm was run on that graph. Clusters were annotated by hand from marker genes (one-vs-rest AUC of 0.7 or more and pseudobulk FDR below 0.05), cross-referenced with published immune cell signatures and with CITE-seq surface protein data. Each label combines the lineage with the marker genes chosen when the labels were made; some subsets were renamed later in the paper, so the synonyms field lists the paper's subset ID (e.g., MP-12) and later name for each label. Cell Ontology terms were assigned in October 2026 from the marker statistics and the paper's interpretation of each subset.
Annotation Method: manual

CD8/NK-CD8 GZMH CD52

Ontology Term

effector memory CD8-positive, alpha-beta T cell, terminally differentiatedCL:0001062

Synonyms

Terms that have the same, or nearly the same, meaning as the selected cell type.

CD8-3, GZMH CD52, TEMRA, GZMH+ CD8 T cell

Category

Class of which the cell type is a subclass.

CD8-positive, alpha-beta T cellCL:0000625

Ontology Assessment

Disagreements with any aspect of the Cell Ontology.

—

Marker Gene Evidence

Gene markers used by researcher to assign this cell annotation.

CD8A, CD3D, GZMH, CD8B, CD52, S100A4, B2M, CD3G, CD3E, SH3BGRL3

Canonical Marker Genes

CD3E, CD8A, GZMH, GZMB, PRF1, KLRG1, CX3CR1, B3GAT1

Negative Marker Genes

Genes expected to be absent, or expressed at a notably low level, in these cells, and whose absence supports the assignment of this cell annotation.

TYROBP

Rationale

Cluster 3 of the CD8 T/NK cell lineage (paper subset CD8-3). Terminally differentiated effector memory CD8 T cell (TEMRA): CD8A [0.80], CD8B [0.74], GZMH [0.75], KLRG1 [0.61], CD52 [0.73], with low CCR7 [0.48], SELL [0.43] and CD27 [0.45]; the original annotation notes record CD45RA protein (CITE-seq). More abundant after SARS-CoV-2 infection; 19% of cells carried public TCRs. Numbers in brackets are one-vs-rest AUCs within the CD8 T/NK cell lineage.

Rationale DOI

10.1101/2025.09.19.25334095

Inherited from COVID_annotation_07062023

Set of Labels Description

Manual annotations of Leiden clusters computed separately within each of four lineages (B cells, CD4 T cells, CD8 T and NK cells, mononuclear phagocytes). For each lineage, PCA scores were aligned across batches with Harmony, each cell was connected to its 50 nearest neighbors, and the Leiden algorithm was run on that graph. Clusters were annotated by hand from marker genes (one-vs-rest AUC of 0.7 or more and pseudobulk FDR below 0.05), cross-referenced with published immune cell signatures and with CITE-seq surface protein data. Each label combines the lineage with the marker genes chosen when the labels were made; some subsets were renamed later in the paper, so the synonyms field lists the paper's subset ID (e.g., MP-12) and later name for each label. Cell Ontology terms were assigned in October 2026 from the marker statistics and the paper's interpretation of each subset.

Annotation Method

manual