Simulation-based Bayesian deep learning enables uncertainty-aware tumor fraction estimation in cell-free DNA
5 affiliations
- 1Gray Faculty of Medical and Health Sciences, Tel Aviv University, Tel Aviv, 39040 Israel
- 2Edmond J. Safra Center for Bioinformatics, Tel Aviv University, Tel Aviv, 39040 Israel
- 3Department of Oncology, Chaim Sheba Medical Center, Ramat Gan, 52621 Israel
- 4Department of Oncology, Tel Aviv Sourasky Medical Center, Tel Aviv, 39040 Israel
- 5Department of Statistics and Operations Research, Tel Aviv University, Tel Aviv, 39040 Israel
Abstract
Background: Estimating tumor fraction from whole-genome cell-free DNA sequencing is critical for liquid biopsy, but is hampered by weak signals and baseline noise at low tumor fractions. Existing computational methods often require matched controls or large labeled datasets for training and lack uncertainty quantification. To address these gaps, we developed purNPE, a Bayesian deep-learning framework trained without labeled cancer cell-free DNA samples. Specifically, purNPE leverages a two-part generative model: one component simulates diverse tumor copy-number profiles based on evolutionary genealogies, while a second, data-driven component learns and replicates realistic sequencing background patterns from cancer-free cell-free DNA. By training a Neural Posterior Estimator on synthetic tumor profiles augmented with learned noise, purNPE performs amortized inference in milliseconds without needing a reference sample set at inference.
Results: In a real-world pan-cancer cohort, purNPE achieved comparable performance with existing methods against an orthogonal mutant-allele-fraction proxy (MAE = 0.066). In semi-synthetic and in silico experiments, purNPE showed calibrated uncertainty estimates and a dose-response across low tumor-fraction spike-ins, with separation of approximately 1% TF spike-ins from blanks under the evaluated conditions.
Conclusions: This work provides a proof-of-concept framework for using simulation-based inference to derive calibrated, uncertainty-aware TF estimates. Larger prospective cohorts with stronger orthogonal ground truth are required before analytical sensitivity or clinical utility can be established.
Supplementary Information: The online version contains supplementary material available at https://doi.org/10.1186/s13040-026-00595-5.
Related
- Ejection fraction on a budget: mapping the accuracy-compute trade space for video-based ejection fraction estimation
- Correction of arterial input-function saturation for extraction fraction estimation in retinal fluorescein videoangiography
- Bayesian Parameter Estimation of Cannibalism Dynamics in Larval Atractosteus tropicus Under Different Feeding Conditions
- CP-BayesNE: Uncertainty-Aware Personalized Nutrient-Density Estimation from Food Images Using Conformal Prediction and Prior-Corrected Dietary Adaptation
- Deep learning cardiac motion analysis reveals the dynamic pathophysiology and genetic architecture of heart failure with preserved ejection fraction
BibTeX
@article{volkov2026simulationbased,
title = {{Simulation-based Bayesian deep learning enables uncertainty-aware tumor fraction estimation in cell-free DNA}},
author = {Hadas Volkov and Maria Raitses-Gurevich and Meitar Grad and Rani Shlayem and Artem Danilevsky and Tami Rubinek and Malka Gorfine and Noam Shomron},
journal = {BioData Mining},
year = {2026},
month = aug,
doi = {10.1186/s13040-026-00595-5},
url = {https://doi.org/10.1186/s13040-026-00595-5},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Simulation-based Bayesian deep learning enables uncertainty-aware tumor fraction estimation in cell-free DNA
AU - Hadas Volkov
AU - Maria Raitses-Gurevich
AU - Meitar Grad
AU - Rani Shlayem
AU - Artem Danilevsky
AU - Tami Rubinek
AU - Malka Gorfine
AU - Noam Shomron
PY - 2026
DA - 2026/08/06
JO - BioData Mining
DO - 10.1186/s13040-026-00595-5
UR - https://doi.org/10.1186/s13040-026-00595-5
AB - Background: Estimating tumor fraction from whole-genome cell-free DNA sequencing is critical for liquid biopsy, but is hampered by weak signals and baseline noise at low tumor fractions. Existing computational methods often require matched controls or large labeled datasets for training and lack uncertainty quantification. To address these gaps, we developed purNPE, a Bayesian deep-learning framework trained without labeled cancer cell-free DNA samples. Specifically, purNPE leverages a two-part generative model: one component simulates diverse tumor copy-number profiles based on evolutionary genealogies, while a second, data-driven component learns and replicates realistic sequencing background patterns from cancer-free cell-free DNA. By training a Neural Posterior Estimator on synthetic tumor profiles augmented with learned noise, purNPE performs amortized inference in milliseconds without needing a reference sample set at inference. Results: In a real-world pan-cancer cohort, purNPE achieved comparable performance with existing methods against an orthogonal mutant-allele-fraction proxy (MAE = 0.066). In semi-synthetic and in silico experiments, purNPE showed calibrated uncertainty estimates and a dose-response across low tumor-fraction spike-ins, with separation of approximately 1% TF spike-ins from blanks under the evaluated conditions. Conclusions: This work provides a proof-of-concept framework for using simulation-based inference to derive calibrated, uncertainty-aware TF estimates. Larger prospective cohorts with stronger orthogonal ground truth are required before analytical sensitivity or clinical utility can be established. Supplementary Information: The online version contains supplementary material available at https://doi.org/10.1186/s13040-026-00595-5.
SN - 1756-0381
ER -