(C0-C13)
Subtypes
Cells
Cells
1. Overview & UMAP Clustering
B cells were extracted from the integrated LPL/WM scRNA-seq dataset (6 patients, PRE and POST idiotype DNA vaccination) and re-clustered independently using Harmony batch correction. A total of 6,213 B cells were identified and grouped into 14 clusters (C0–C13), which were annotated into 11 B cell subtypes using the scType framework (Ianevski et al., Nat Commun 2022) based on the Yang et al. 18-subtype B cell classification.
UMAP: B Cell Clusters
Figure 1. UMAP visualization of B cell clusters (C0-C13) after Harmony integration across 6 patients. Colors represent distinct clusters identified by Louvain community detection.
UMAP: B Cell Subtypes (scType Annotation)
Figure 2. UMAP colored by scType-annotated B cell subtypes. 11 subtypes identified based on Yang et al. 18-subtype framework: Immature B, FOShi naive B, Memory B, Plasma Cell, Plasmablast, FTLhi immature B, S100A8hi immature B, Naive B, Pre B, Pro B, Cycling pro/pre B.
UMAP: By Patient
Figure 3. UMAP colored by patient origin, showing successful batch integration via Harmony. Cells from all 6 patients are well-intermixed across clusters.
UMAP: By Timepoint (Pre vs Post)
Figure 4. UMAP colored by vaccination timepoint (Pre = red, Post = blue). Notable reduction of C3 (FOShi naive B) population in POST samples.
t-SNE: B Cell Subtypes
Figure 5. t-SNE visualization of B cell subtypes as an alternative dimensionality reduction to UMAP.
Elbow Plot: PC Selection
Figure 6. Elbow plot showing the standard deviation of principal components used to determine the optimal number of PCs for clustering.
2. Cluster Annotation & Marker Genes
2a. Cluster-Subtype Mapping
| Cluster | Subtype | Key Markers | Characteristics |
|---|---|---|---|
| C0 | Immature B | MS4A1, CD37, TSC22D3, ZFP36L2 | Largest cluster; transitional B cells |
| C1 | Memory B | CD27, IGHG1, IGHG3, AIM2 | Class-switched memory; CD27-negative subset |
| C2 | Plasma Cell | MZB1, JCHAIN, XBP1, IGHG1 | Antibody-secreting; Ig heavy chain high |
| C3 | FOShi naive B | FOS, JUN, NR4A2, DUSP1 | Activation markers high; most vaccine-responsive |
| C4 | Plasmablast | MZB1, TXNDC5, IGHG1, XBP1 | Actively differentiating to plasma cells |
| C5 | Pre B | VPREB1, IGLL1, CD34 | Pre-B cell receptor expressing |
| C6 | Cycling pro/pre B | MKI67, TOP2A, CDK1 | Proliferating progenitor B cells |
| C7 | Immature B | MS4A1, CD79A, TCL1A | Secondary immature B cluster |
| C8 | Naive B | IGHD, IGHM, FCER2 | Mature naive B cells |
| C9 | Pro B | CD34, MME, DNTT | Earliest B cell progenitors |
| C10 | Immature B | MS4A1, CD79A | Minor immature B cluster |
| C11 | FTLhi immature B | FTL, FTH1 | Ferritin-high immature B cells |
| C12 | S100A8hi immature B | S100A8, S100A9 | Alarmin-expressing immature B cells |
| C13 | S100A8hi immature B | S100A8, S100A9, S100A12 | Minor alarmin-expressing cluster |
Overview Panel
Figure 7. Multi-panel overview of B cell analysis: UMAP clusters, subtype annotations, patient distribution, and timepoint comparison.
2b. Marker Gene Expression
Dot Plot: B Cell Markers by Subtype
Figure 8. Dot plot showing expression of canonical B cell subtype markers. Dot size = percentage of cells expressing the gene; Color intensity = average expression level.
Dot Plot: Markers by Cluster
Figure 9. Dot plot of marker genes across all 14 clusters (C0-C13).
Feature Plot: LPL-Specific Markers
Figure 10. UMAP feature plots showing expression of LPL/WM-relevant markers across B cell clusters.
Heatmap: Top Cluster Markers
Figure 11. Heatmap of top differentially expressed marker genes per cluster, identified by FindAllMarkers (Wilcoxon rank-sum test).
3. Lymphoma vs Normal B Cell Identification
Lymphoma cells were identified by matching B cell receptor (BCR) CDR3 sequences to patient-specific clonotype data from VDJ sequencing. Cells with CDR3 sequences matching the dominant lymphoma clone were classified as Lymphoma; all others as Normal B.
Lymphoma vs Normal B: UMAP
Figure 12. UMAP showing lymphoma (red) vs normal B (blue) cells identified by CDR3 matching.
Lymphoma vs Normal B: By Subtype
Figure 13. Lymphoma and Normal B cell distribution across B cell subtypes.
3a. Lymphoma Fraction by Subtype
| B Cell Subtype | Total Cells | Lymphoma Cells | Lymphoma % |
|---|---|---|---|
| FOShi naive B | 687 | 581 | 84.6% |
| Plasma Cell | 788 | 665 | 84.4% |
| Plasmablast | 389 | 318 | 81.7% |
| Memory B | 1342 | 780 | 58.1% |
| S100A8hi immature B | 140 | 78 | 55.7% |
| Immature B | 1763 | 785 | 44.5% |
| FTLhi immature B | 91 | 33 | 36.3% |
| Naive B | 226 | 8 | 3.5% |
| Pro B | 97 | 3 | 3.1% |
| Pre B | 365 | 10 | 2.7% |
| Cycling pro/pre B | 325 | 8 | 2.5% |
Lymphoma Percentage by Subtype
Figure 14. Bar plot showing the percentage of lymphoma cells within each B cell subtype. Mature/differentiated subtypes harbor the highest lymphoma fractions.
Lymphoma Cluster Composition
Figure 15. Stacked bar plot showing lymphoma vs normal B cell composition per cluster.
3b. Lymphoma Distribution by Patient
| Patient | Pre (Lymph/Normal, %Lymph) | Post (Lymph/Normal, %Lymph) | Change |
|---|---|---|---|
| Patient1 | 873/127 (87.3%) | 386/22 (94.6%) | ↑ +7.3% |
| Patient2 | 416/247 (62.7%) | 355/951 (27.2%) | ↓ -35.5% |
| Patient3 | 355/299 (54.3%) | 505/469 (51.8%) | ↓ -2.5% |
| Patient4 | 100/322 (23.7%) | 37/259 (12.5%) | ↓ -11.2% |
| Patient9 | 76/139 (35.3%) | 166/109 (60.4%) | ↑ +25.1% |
Lymphoma by Patient
Figure 16. Lymphoma vs Normal B cell distribution per patient.
Lymphoma Pre vs Post by Patient
Figure 17. Lymphoma fraction changes between Pre and Post vaccination per patient. Patients 2 and 4 show the most dramatic reduction in lymphoma fraction.
4. Pre vs Post Vaccination Changes
4a. Cluster Proportion Changes
| Cluster: Subtype | Pre Mean % | Post Mean % | Change | Sig |
|---|---|---|---|---|
| C0: Immature B | 8.39% | 8.26% | ↓ Decreased | ns |
| C1: Memory B | 12.93% | 14.11% | ↑ Increased | ns |
| C2: Plasma Cell | 11.86% | 10.91% | ↓ Decreased | ns |
| C3: FOShi naive B | 10.44% | 2.5% | ↓ Decreased | ns |
| C4: Plasmablast | 3.97% | 6.68% | ↑ Increased | ns |
| C5: Pre B | 0.11% | 0.26% | ↑ Increased | ns |
| C6: Cycling pro/pre B | 0.05% | 0.23% | ↑ Increased | ns |
| C7: Immature B | 2.68% | 2.26% | ↓ Decreased | ns |
| C8: Naive B | 0.15% | 0.12% | ↓ Decreased | ns |
| C9: Pro B | 0% | 0.16% | ↑ Increased | ns |
| C10: Immature B | 0.78% | 0.79% | ↑ Increased | ns |
| C11: FTLhi immature B | 0.51% | 0.9% | ↑ Increased | ns |
| C12: S100A8hi immature B | 0.62% | 1.36% | ↑ Increased | ns |
| C13: S100A8hi immature B | 0.19% | 0.77% | ↑ Increased | ns |
- C3 (FOShi naive B): 10.44% → 2.50% (↓76% reduction) — most responsive to vaccine
- C4 (Plasmablast): 3.97% → 6.68% (↑68% increase) — resistant, clone replacement
- C1 (Memory B): 12.93% → 14.11% (modest increase)
- C2 (Plasma Cell): 11.86% → 10.91% (stable)
Pre vs Post: Cluster Proportion Summary
Figure 18. Stacked bar plot showing B cell cluster proportions before (Pre) and after (Post) vaccination.
Pre vs Post: Change Plot
Figure 19. Waterfall plot showing the direction and magnitude of cluster proportion changes between Pre and Post vaccination. Negative = decreased POST; Positive = increased POST.
Pre vs Post: Box Plot
Figure 20. Box plots comparing cluster proportions between Pre (red) and Post (blue) samples across patients. Each dot represents one patient.
B Cell Proportions by Patient
Figure 21. Stacked bar plot showing B cell subtype proportions per patient-timepoint sample.
5. Summary & Key Findings
6,213 B cells from 6 patients resolved into 14 clusters and 11 subtypes. The five major populations are: Immature B (C0, largest), Memory B (C1), Plasma Cell (C2), FOShi naive B (C3), and Plasmablast (C4).
3,269 lymphoma cells (52.6%) identified via CDR3 matching. Lymphoma is concentrated in differentiated B cell subtypes (FOShi naive B: 84.6%, Plasma Cell: 84.4%, Plasmablast: 81.7%) while progenitor B cells remain largely normal (<4% lymphoma).
FOShi naive B cells (C3) show the most dramatic reduction post-vaccination (10.44% → 2.50%, ~76% reduction), consistent with their high HLA class II expression making them vulnerable to vaccine-induced T cell immunity. In contrast, plasmablasts (C4) expanded (3.97% → 6.68%), indicating immune evasion and potential clone replacement.
The data reveals a clear hierarchy: early/mid-differentiated lymphoma B cells (C0, C3) are vaccine-responsive, while terminally differentiated antibody-secreting cells (C2, C4) are resistant. This provides a mechanistic rationale for combining idiotype DNA vaccines with anti-plasma cell agents (e.g., daratumumab, venetoclax).
6. Methods
Preprocessing: scRNA-seq data from 6 LPL/WM patients (Pre and Post vaccination) processed with Cell Ranger. Quality control: nFeature_RNA 200–5000, nCount_RNA < 20000, percent.mt < 15%.
Integration: Harmony batch correction (Korsunsky et al., Nat Methods 2019) across patient samples. 30 PCs used for dimensionality reduction.
Clustering: Louvain community detection (resolution = 0.8) on shared nearest neighbor (SNN) graph. UMAP and t-SNE for visualization.
Annotation: scType (Ianevski et al., Nat Commun 2022) with custom B cell marker database based on Yang et al. 18-subtype classification. Cluster-level annotation using AverageExpression scores.
Lymphoma Identification: CDR3 sequence matching from paired 5' VDJ BCR sequencing. Patient-specific dominant clonotypes used to classify lymphoma vs normal B cells.
Statistical Analysis: Wilcoxon signed-rank test (paired) for Pre vs Post comparisons across patients. All p-values Bonferroni-corrected.
Software: Seurat v5, Harmony, scType, R 4.3.2. Analysis executed on Modal cloud computing.