Journal: Frontiers in Immunology

Article Title: CheckDyn: a multi-cohort computational framework for profiling treatment-induced immune checkpoint dynamics and predicting adaptive resistance to immune checkpoint blockade

doi: 10.3389/fimmu.2026.1847297

Figure Lengend Snippet: Study overview: multi-cohort paired transcriptomics of checkpoint dynamics. (a) Study design schematic. Paired tumor biopsies were collected before and after ICB therapy (anti-PD-1 or anti-PD-1/anti-CTLA-4), and transcriptomic profiles were used for checkpoint dynamics analysis and clinical response association. (b) Dataset overview. Bar chart showing the number of paired patients (pre + post biopsies) per cohort and cancer type. n = 42 from GSE91061 (melanoma), n = 11 from Sade-Feldman2018/ GSE120575 (melanoma), n = 11 from Yost2019/ GSE123813 (BCC); total n = 64. (c) Meta-analysis mean fold-change (Z-score) per checkpoint gene across all datasets. Asterisks indicate genes reaching padj < 0.05. Genes are colored by functional category: inhibitory checkpoints (red), co-stimulatory molecules (teal), metabolic checkpoints (salmon), exhaustion markers (blue-grey). (d) Six-step analysis pipeline: data collection (4 cohorts, n = 64 paired) → pseudo-bulk aggregation and Z-score batch correction → paired DE analysis (Stouffer Z meta-analysis) → network rewiring (co-expression, hub shift) → ensemble model (AUC = 0.812, n = 38 genes) → clinical implication (combination therapy).

Article Snippet: The CheckDyn pipeline integrates these datasets through pseudo-bulk aggregation, empirical Bayes batch correction, and Stouffer Z-score meta-analysis to quantify treatment-induced changes across a curated 38-gene checkpoint and exhaustion marker panel.

Techniques: Transcriptomics, Functional Assay, Expressing

Journal: Frontiers in Immunology

Article Title: CheckDyn: a multi-cohort computational framework for profiling treatment-induced immune checkpoint dynamics and predicting adaptive resistance to immune checkpoint blockade

doi: 10.3389/fimmu.2026.1847297

Figure Lengend Snippet: Meta-analysis volcano and cross-dataset expression heatmap. (a) Meta-analysis volcano plot (3 cohorts). Genes reaching meta-padj < 0.05 (red): LAG3, PDCD1, TOX2 (top right). CD274, CD80 also approach significance. Dashed lines indicate significance thresholds. (b) Log 2 FC heatmap of significant genes per dataset. Color scale as in Figure 2a . Values are annotated per cell. LAG3 shows uniformly high log 2 FC across all three cohorts (0.49/0.76/0.74).

Article Snippet: The CheckDyn pipeline integrates these datasets through pseudo-bulk aggregation, empirical Bayes batch correction, and Stouffer Z-score meta-analysis to quantify treatment-induced changes across a curated 38-gene checkpoint and exhaustion marker panel.

Techniques: Expressing

Journal: Frontiers in Immunology

Article Title: CheckDyn: a multi-cohort computational framework for profiling treatment-induced immune checkpoint dynamics and predicting adaptive resistance to immune checkpoint blockade

doi: 10.3389/fimmu.2026.1847297

Figure Lengend Snippet: Study overview: multi-cohort paired transcriptomics of checkpoint dynamics. (a) Study design schematic. Paired tumor biopsies were collected before and after ICB therapy (anti-PD-1 or anti-PD-1/anti-CTLA-4), and transcriptomic profiles were used for checkpoint dynamics analysis and clinical response association. (b) Dataset overview. Bar chart showing the number of paired patients (pre + post biopsies) per cohort and cancer type. n = 42 from GSE91061 (melanoma), n = 11 from Sade-Feldman2018/ GSE120575 (melanoma), n = 11 from Yost2019/ GSE123813 (BCC); total n = 64. (c) Meta-analysis mean fold-change (Z-score) per checkpoint gene across all datasets. Asterisks indicate genes reaching padj < 0.05. Genes are colored by functional category: inhibitory checkpoints (red), co-stimulatory molecules (teal), metabolic checkpoints (salmon), exhaustion markers (blue-grey). (d) Six-step analysis pipeline: data collection (4 cohorts, n = 64 paired) → pseudo-bulk aggregation and Z-score batch correction → paired DE analysis (Stouffer Z meta-analysis) → network rewiring (co-expression, hub shift) → ensemble model (AUC = 0.812, n = 38 genes) → clinical implication (combination therapy).

Article Snippet: We integrated publicly available RNA-seq and scRNA-seq data from 64 paired tumor samples spanning melanoma, basal cell carcinoma, and non-small-cell lung cancer ( GSE91061 , GSE120575 , GSE123813 , GSE176021 ), applying pseudo-bulk aggregation, Z-score batch correction, and Stouffer meta-analysis for cross-cohort harmonization.

Techniques: Transcriptomics, Functional Assay, Expressing

Journal: Frontiers in Immunology

Article Title: CheckDyn: a multi-cohort computational framework for profiling treatment-induced immune checkpoint dynamics and predicting adaptive resistance to immune checkpoint blockade

doi: 10.3389/fimmu.2026.1847297

Figure Lengend Snippet: Meta-analysis volcano and cross-dataset expression heatmap. (a) Meta-analysis volcano plot (3 cohorts). Genes reaching meta-padj < 0.05 (red): LAG3, PDCD1, TOX2 (top right). CD274, CD80 also approach significance. Dashed lines indicate significance thresholds. (b) Log 2 FC heatmap of significant genes per dataset. Color scale as in Figure 2a . Values are annotated per cell. LAG3 shows uniformly high log 2 FC across all three cohorts (0.49/0.76/0.74).

Article Snippet: We integrated publicly available RNA-seq and scRNA-seq data from 64 paired tumor samples spanning melanoma, basal cell carcinoma, and non-small-cell lung cancer ( GSE91061 , GSE120575 , GSE123813 , GSE176021 ), applying pseudo-bulk aggregation, Z-score batch correction, and Stouffer meta-analysis for cross-cohort harmonization.

Techniques: Expressing