Shengbin Ye

A bstract: Single-cell DNA sequencing (scDNA-seq) and single-cell RNA sequencing (scRNA-seq) enable the inference of copy number clones and transcriptional states in tumor studies. However, these two assays usually profile different cells of the same tumor, requiring computational assignment of unpaired scRNA-seq cells to clones identified by scDNA-seq to link clonal genotypes with transcriptional profiles. This task is challenging because the only connection between the two modalities is a weak, gene-dependent dosage effect, and the absence of clone labels for RNA cells precludes supervised learning of this relationship. Moreover, clonal distributions may differ between the two modalities because the cells are profiled independently, while the inherent clonal imbalance of real tumors can leave minority subclones underrepresented, further compromising assignment accuracy. To address these challenges, we present KLONE, a per-clone k -nearest-neighbor ( k -NN) ensemble that assigns RNA cells directly to copy-number clones based on cross-modal Pearson similarity, without the need for a parametric model or clone labels for RNA cells. To ensure a fair assessment of each clone’s compatibility with a given RNA cell, KLONE allocates an identical neighborhood budget to every clone, preventing majority clones from dominating the affinity assessment solely through their greater abundance. We compare KLONE with six existing methods on three colorectal cancer patients, using cells profiled in both modalities as ground truth. KLONE is the most accurate method and recalls minority clones at 100%. Despite discrepancies in clonal distribution between the two modalities, KLONE recovers the clonal distribution of RNA cells almost perfectly ( ρ = 0.99).

Sample-level cell counts of CD8 clusters

Statistical test for association of lymphocyte cluster proportion with progression

Fig. S1. Patient disposition and characteristics. Fig. S2. Timing of OI biopsy collection did not correlate with response. Fig. S3. Major cell types identified in the tumor immune microenvironment. Fig. S4. CD4+ T cells identified in the tumor immune microenvironment. Fig. S5. Effector Treg cells in the HPV-positive OPC TIME highly express activation markers. Fig. S6. Treatment-induced changes in Treg proportions and clonotype frequencies. Fig. S7. CD8+ T cells in MHRs have different transcriptional programs vs those in nMHRs. Fig. S8. CD8+ T cells identified in the tumor immune microenvironment. Fig. S9. T cell clusters did not vary by anatomic site. Fig. S10. TRM CD8+ T cells had high expression of tumor-reactive, cytotoxic and checkpoint genes. Fig. S11. Fluorescent imaging of antigen-experienced, memory CD8+ T cells in baseline tumors. Fig. S12. Clonality of CD4+ T-cell TCR repertoire. Fig. S13. CD8+ T-cell TCR repertoire clonality and dynamics. Fig. S14. Phenotypes of treatment-altered CD8 clonotypes varied by response. Fig. S15. Predominant phenotypes of treatment-responsive CD8+ T-cell clonotypes. Fig. S16. Identification of HPV-specific TCRs. Fig. S17. NK cells that express Fc gamma receptor IIIa (CD16) are more activated than those that do not. Fig. S18. Increase in proportion of Effector1(IFNG+) cells associated with reduction in tumor viability.

Table S1. Patient and tumor characteristics. Table S2. Representativeness of study participants. Table S3. Clinical and histopathologic characteristics of individual patients. Table S4. Treatment related adverse events that occurred after first dose. Table S5. Adverse events related to immune checkpoint blockade. Table S6. Serious adverse events.

Clinical and single-cell features used in multivariable regression models

Single-cell RNA and TCR sequencing information and metrics

Differentially expressed genes of CD8+ T cells from MHRs vs non-MHRs at baseline

Differentially expressed genes of CD8 clusters

Statistical test for proportions of CD4+ T cell clusters in the tumor immune microenvironment stratified by histological response and timepoint