Somdutta Saha
Research Subjects
Somdutta Saha has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- Amit M. Deshpande
Mythic Therapeutics Inc., Waltham, Massachusetts
33 shared publications - Brian Rabinovich
Research Unit Oncology, EMD Serono, Billerica, Massachusetts
33 shared publications - Chunxiao Xu
EMD Serono Res Ctr, Res Unit Oncol, 45 Middlesex Turnpike, Billerica, MA 01821 USA
33 shared publications - Laura Helming
Research Unit Oncology, EMD Serono, EMD Serono Research & Development Institute
33 shared publications - Lindsay Webb
Research Unit Oncology, EMD Serono, Billerica, Massachusetts
33 shared publications - Marat Alimzhanov
Research Unit Oncology, EMD Serono, Billerica, Massachusetts
33 shared publications
Latest Publications
- Open Access
Synthetic RNA-seq cohorts for data sharing: a discovery-aware benchmark at transcriptome scale
Background: Sharing patient-level gene expression data is essential for translational discovery but carries documented re-identification risks. Bulk RNA-seq count matrices can retain genotypic signals and paired clinical metadata compounds this through quasi-identifier matching. Synthetic RNA-seq cohorts offer a complementary path for privacy-preserving data sharing, but the field lacks a multi-axis benchmark that probes biological fidelity and empirical privacy risk at transcriptome scale. Here we present a multi-axis benchmark framework that reflects how transcriptomic cohorts are used in translational practice. Methods: We benchmarked three generative models across four cohorts drawn from datasets spanning oncology (TCGA-LUAD), sepsis (GSE184900), and pediatric IBD (RISK/GSE57945): dbTwin (a non-deep-learning, target-conditioned method that operates natively at RNA-seq scale), class-MVN (a low-rank target-conditioned multivariate Gaussian model), and PCA-CTGAN (a tabular GAN trained in PCA-compressed space). Synthetic cohorts were generated from training folds of a five-fold stratified design. We evaluated DE genes recovery, log2FC and significance (padj) concordance, held-out AUC (TSTR) and SHAP concordance and distance-based memorization risk. Results: class-MVN recovered 64.8% and 43.1% of real DE genes in the two binary cohorts with high fold-change correlation but lower significance concordance (r = 0.24-0.68) and inflated DE gene counts. dbTwin recovered 78.7% and 91.8% of real DE genes in the same cohorts, with high fold-change correlation and stronger significance concordance (r > 0.88). Both methods matched held-out real AUC under TSTR, but SHAP agreement differed substantially: dbTwin preserved feature attribution patterns across cohorts (SHAP top-50 genes r = 0.84-0.99 across two binary and two multiclass cohorts), whereas class-MVN showed moderate performance for majority classes but degraded in multiclass and imbalanced settings (SHAP r = 0.31-0.79). PCA-CTGAN performed poorly across most DE and ML metrics. Distance-to-closest-record analysis did not indicate memorization by any of the models. Conclusions: We introduced a multi-axis, transcriptome-scale, discovery-aware benchmark to validate synthetic RNA-seq cohorts for translational workflows and evaluated three generative models across four real-world cohorts. These results support the use of synthetic RNA-seq cohorts for exploratory analysis and method development, while emphasizing the need for careful validation before use in higher-stakes applications. All benchmark code and data are available via GitHub: "https://github.com/Nanda-Aditya/rna-syn-bench".
Open AccessSynthetic RNA-seq cohorts for data sharing: a discovery-aware benchmark at transcriptome scale
Background: Sharing patient-level gene expression data is essential for translational discovery but carries documented re-identification risks. Bulk RNA-seq count matrices can retain genotypic signals and paired clinical metadata compounds this through quasi-identifier matching. Synthetic RNA-seq cohorts offer a complementary path for privacy-preserving data sharing, but the field lacks a multi-axis benchmark that probes biological fidelity and empirical privacy risk at transcriptome scale. Here we present a multi-axis benchmark framework that reflects how transcriptomic cohorts are used in translational practice. Methods: We benchmarked three generative models across four cohorts drawn from datasets spanning oncology (TCGA-LUAD), sepsis (GSE184900), and pediatric IBD (RISK/GSE57945): dbTwin (a non-deep-learning, target-conditioned method that operates natively at RNA-seq scale), class-MVN (a low-rank target-conditioned multivariate Gaussian model), and PCA-CTGAN (a tabular GAN trained in PCA-compressed space). Synthetic cohorts were generated from training folds of a five-fold stratified design. We evaluated DE genes recovery, log2FC and significance (padj) concordance, held-out AUC (TSTR) and SHAP concordance and distance-based memorization risk. Results: class-MVN recovered 64.8% and 43.1% of real DE genes in the two binary cohorts with high fold-change correlation but lower significance concordance (r = 0.24-0.68) and inflated DE gene counts. dbTwin recovered 78.7% and 91.8% of real DE genes in the same cohorts, with high fold-change correlation and stronger significance concordance (r > 0.88). Both methods matched held-out real AUC under TSTR, but SHAP agreement differed substantially: dbTwin preserved feature attribution patterns across cohorts (SHAP top-50 genes r = 0.84-0.99 across two binary and two multiclass cohorts), whereas class-MVN showed moderate performance for majority classes but degraded in multiclass and imbalanced settings (SHAP r = 0.31-0.79). PCA-CTGAN performed poorly across most DE and ML metrics. Distance-to-closest-record analysis did not indicate memorization by any of the models. Conclusions: We introduced a multi-axis, transcriptome-scale, discovery-aware benchmark to validate synthetic RNA-seq cohorts for translational workflows and evaluated three generative models across four real-world cohorts. These results support the use of synthetic RNA-seq cohorts for exploratory analysis and method development, while emphasizing the need for careful validation before use in higher-stakes applications. All benchmark code and data are available via GitHub: "https://github.com/Nanda-Aditya/rna-syn-bench".
- Open Access
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Figure S1 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
M9567 mAb2 TM structure and molecules used for primary research
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Open AccessFigure S1 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
M9567 mAb2 TM structure and molecules used for primary research
- Open Access
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Figure S13 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
Tumor-targeted immune activation does not result in increased liver inflammation
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Open AccessFigure S13 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
Tumor-targeted immune activation does not result in increased liver inflammation
- Open Access
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Figure S2 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
M9657 is not appreciably internalized in MSLN+ cells
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Open AccessFigure S2 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
M9657 is not appreciably internalized in MSLN+ cells
- Open Access
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Figure S10 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
FS122m generates dose-dependent antitumor efficacy and long-term tumor antigen-specific immune protection.
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Open AccessFigure S10 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
FS122m generates dose-dependent antitumor efficacy and long-term tumor antigen-specific immune protection.
- Open Access
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Figure S7 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
FS122m characterization on cellular binding, in vitro agonism and in vivo distribution
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Open AccessFigure S7 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
FS122m characterization on cellular binding, in vitro agonism and in vivo distribution
- Open Access
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Figure S8 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
FS122m surrogate and M9657 chimeric antibodies display significant antitumor efficacy in syngeneic models
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Open AccessFigure S8 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
FS122m surrogate and M9657 chimeric antibodies display significant antitumor efficacy in syngeneic models
- Open Access
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Figure S8 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
FS122m surrogate and M9657 chimeric antibodies display significant antitumor efficacy in syngeneic models
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Open AccessFigure S8 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
FS122m surrogate and M9657 chimeric antibodies display significant antitumor efficacy in syngeneic models
- Open Access
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Figure S2 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
M9657 is not appreciably internalized in MSLN+ cells
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Open AccessFigure S2 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
M9657 is not appreciably internalized in MSLN+ cells
- Open Access
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Figure S11 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
FS122m increases infiltration of CD8+ T cells but not other immune subsets in the TME and spleen of the CT26-KSA tumor model
- 2 February 2024
- Published by American Association for Cancer Research (AACR)
Open AccessFigure S11 from M9657 Is a Bispecific Tumor-Targeted Anti-CD137 Agonist That Induces MSLN-Dependent Antitumor Immunity without Liver Inflammation
FS122m increases infiltration of CD8+ T cells but not other immune subsets in the TME and spleen of the CT26-KSA tumor model