Pith. sign in

REVIEW 3 major objections 6 minor 62 references

Data-Driven Soft Labeling Scales DNA Read Classification to Whole-Body Cell-Type Deconvolution

T0 review · 3 major / 6 minor · reviewed 2026-07-11 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Data-driven soft labels let DNA-read classifiers deconvolve 39 whole-body cell types, cutting error 2.56× versus prior methods.

desk verdict Solid engineering that makes read-level methylome deconvolution work at 39 classes via data-driven soft labels; the 2.56× MSE gain is real on their setup, but rests on pseudobulk + TCS proxies. read the letter →

arxiv 2607.04987 v2 pith:QBEK7RXN submitted 2026-07-06 cs.LG q-bio.GNq-bio.QM

Dmytro Rizdvanetskyi , Nathan Roos , Pavlo Lutsik This is my paper · ORCID
classification cs.LGq-bio.GNq-bio.QM
keywords cell-typedeconvolutionDNAmethylationread-levelclassificationsoftlabelslabelenhancementwhole-bodyatlascalibrationmany-to-manymapping
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Cell-type deconvolution estimates how much of each cell type is present in a mixed biological sample. Most methylation-based methods throw away the pattern carried by each individual DNA read and work only on averages; the few methods that keep read-level patterns never scaled past a handful of classes. Hard one-hot labels clash with biology: the same methylation pattern can arise from several cell types, and off-target reads dominate once dozens of types are included, so classifiers never converge. The paper shows that replacing hard labels with data-driven soft labels—empirical conditional distributions over cell types for each observed methylation signature—removes the conflict. These soft labels are embedded in Syto, a modular pipeline of classifier, deconvolver and calibrator. On a 39-cell-type whole-body atlas the best Syto configuration reduces mean-squared error 2.56-fold relative to the previous state of the art; the same gains appear on an out-of-distribution tissue panel. The result opens the door to larger reference panels and treats tumor purity and classification as special cases of deconvolution.

What carries the argument

Data-driven soft labels: for each unique methylation signature the empirical frequency of cell types among reads that share that signature (with class reweighting and nearest-neighbor pooling for low coverage). These soft targets replace one-hot labels inside an entropy-weighted cross-entropy loss, enabling the modular Syto pipeline of classifier, deconvolver and linear calibrator.

What would settle it

Run the same best Syto configuration on a set of real bulk samples whose cell-type proportions have been independently measured by single-cell or flow-cytometry ground truth; if the reported 2.56-fold MSE reduction disappears, the central claim fails.

Watch

Extended reading notes

Core claim

Data-driven soft labels that estimate the true conditional cell-type distribution for each DNA-read methylation signature allow read-level classifiers to converge and deconvolve mixtures of 39 whole-body cell types, cutting mean-squared error by a factor of 2.56 relative to the prior state of the art and transferring the improvement to an out-of-distribution 16-tissue panel.

Load-bearing premise

Performance is measured on artificial mixtures made by mixing purified cell-type libraries and on a proxy score that only checks whether expected tissue cell types appear, not on samples whose true multi-type proportions are known.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper introduces Syto, a modular read-level classification framework for methylome-based cell-type deconvolution that scales to dozens of classes. The central technical contribution is data-driven soft labeling: for each methylation signature the soft label is the (reweighted, Jaccard-pooled) empirical conditional distribution over cell types, presented as the MLE of P(c|s). This is combined with extensions of CancerDetector, Dismir and MethylBERT, five deconvolvers (including PSLS/NNLS and learned models), and a hyperparameter-free linear calibrator for simplex outputs. On 39-class WGBS pseudobulks from the Loyfer atlas, the best configuration (Dismir + soft labels with pooling + PSLS + linear simplex calibration) reports a 2.56 imes MSE reduction versus UXM (1.17e-4 vs 3.00e-4); gains transfer to an OOD RRBS tissue atlas under a Tissue Concordance Score (TCS) proxy. Soft labeling is argued to be generally useful for many-to-many signal-to-label maps.

Significance. If the soft-label construction and modular pipeline hold under stronger evaluation, the work would remove a genuine bottleneck: existing read-level classifiers (CancerDetector, Dismir, MethylBERT) have been limited to few classes because hard labels conflict with non-discriminative off-target reads. Scaling to whole-body panels and framing tumor purity/classification as special cases of deconvolution would matter for liquid-biopsy and multi-omics applications. Strengths that should be credited include the exhaustive 360-permutation design with BCa bootstrap CIs, the explicit demonstration that MLE deconvolvers fail at this scale (Tables S1–S3), the feature-selection ablations, the linear calibrator comparison against Dirichlet/vector/temperature scaling, and the candid Limitations section. The soft-label MLE argument and entropy-weighted CE are clean and potentially portable beyond methylation.

major comments (3)
  1. The headline 2.56× MSE claim (Abstract; Table 1 best row: Dismir + DD soft labels w/ pool. + PSLS + Lin. simplex, 1.17e-4 vs UXM 3.00e-4) is measured exclusively on Algorithm-2 pseudobulks subsampled from purified WGBS libraries (N=475k, K=100k; §4.1, A.4.2). Soft labels themselves are empirical frequencies of those same purified labels (Eq. 4 + Alg. 1). No public multi-type bulk with independent known proportions is used. Section 5 acknowledges this, yet the strongest claim is still framed as a deconvolution accuracy gain. Residual purification noise, coverage bias, or DMR-selection artifacts shared between training soft labels and test pseudobulks can inflate the reported ratio relative to real heterogeneous tissue. Either (i) add at least one external known-proportion or spike-in multi-type evaluation, or (ii) restate the central claim as a large-scale pseudobulk result and demote the
  2. OOD transfer is scored only by Tissue Concordance Score (TCS = sum of predicted mass on Tabula-Sapiens-mapped expected types, §A.4.5; Fig. 4, Tables S17–S21). TCS is an upper-bound proxy that does not penalize misallocation among expected types and depends on a CL/UBERON mapping plus a ≥1% threshold. The claim that gains “transfer” to 16 tissues is therefore weaker than a proportion-level claim. Report TCS alongside at least one secondary metric (e.g., expected-type KL or rank correlation with Tabula Sapiens fractions where available) and state explicitly that TCS is not a substitute for known-proportion MSE.
  3. UXM is given a structural advantage: its atlas is built from all splits (§4.1), while Syto classifiers are trained on the 62% train split only. The 2.56× comparison therefore understates Syto’s relative gain if anything, but the absolute SoTA framing is slightly unfair. Either retrain UXM on the same train-only atlas or report both “UXM-all” and “UXM-train” baselines so the reader can calibrate the gap.
minor comments (6)
  1. Figure 3 shows non-convergence of hard labels but does not report the corresponding soft-label training curves; adding them would make the labeling comparison self-contained.
  2. Notation: Δ^{C−1} is used for both C-class soft labels and C+1 hard-with-background labels; clarify when the background class expands the simplex.
  3. Algorithm 1: the asymmetric Jaccard pooling and inverse-frequency reweighting are important; a short sensitivity table for τ and δ_max would strengthen reproducibility.
  4. Table 1 caption and §A.4.4: “top 156 features” and “† = diagonal + background” are easy to miss; consider a one-line legend in the main table.
  5. Related work on label enhancement (Xu et al., Wang & Geng) is cited; a sentence contrasting entropy-upweighting in prior LE with the paper’s low-entropy upweighting would help readers outside the methylation community.
  6. Typos / polish: “Sytolays”, “Sytoconsistently”, “Sytoreduces” (missing spaces after Syto in Abstract/Introduction); “easurable” → “measurable” (A.4.3); “assesed” → “assessed” (A.3.2).

Circularity Check

0 steps flagged · score 1.0 of 10

No load-bearing circularity: soft labels are empirical MLEs of training conditionals, deconvolvers/calibrators are fit on held-out pseudobulks, and the 2.56× MSE claim is an external comparison to UXM, not forced by construction.

full rationale

The derivation chain is ordinary supervised estimation. Data-driven soft labels (Eq. 4, Alg. 1) are the MLE of the empirical conditional P(c|s) computed solely from training-set class counts (with optional Jaccard pooling and inverse-frequency reweighting); they are not defined from the test MSE or TCS. Classifiers are trained with entropy-weighted CE on those labels; deconvolvers (XGB/MLP/SWN/NNLS/PSLS) are fit on K=100k pseudobulk prediction matrices generated from the training split (Alg. 2–3); linear calibrators are fit on validation proportions and applied post-hoc. Test metrics (Table 1) are computed on a held-out split (and an OOD RRBS set scored by the independently defined TCS). UXM is an external unsupervised baseline (even though its atlas is built on all splits, an acknowledged advantage). No equation equates the reported MSE ratio to a fitted input; no uniqueness theorem or ansatz is imported from the authors’ prior work to force the result. The only minor self-reference is the modular reuse/extension of MethylBERT (prior work co-authored by Lutsik), which is not load-bearing for the soft-label claim. Pseudobulk and TCS proxies are validity limitations (explicitly noted in §5), not circular reductions. Score 1 for the ordinary self-extension of one classifier; central claims remain independently evaluated.

Assumptions & free parameters 7 free parameters · 6 assumptions · 4 invented entities

The central claim rests on standard probabilistic MLE for categorical soft labels, domain biology of imperfect methylation maintenance, public atlases, and several engineering choices (pooling thresholds, feature masks, pseudobulk size, TCS definition) that are free or ad hoc. No new physical entities; invented constructs are methodological.

free parameters (7)
  • soft-label pooling min count τ and max Jaccard distance δ_max
    Control how far nearest-signature counts are transferred; chosen for coverage, not derived from first principles.
  • canonical label-smoothing ε
    Smoothing strength on hard labels with background; standard free hyperparameter.
  • top-156 feature mask for deconvolvers
    Selected by max/mean scores on pure profiles; computational convenience that discards most of the G×C matrix.
  • pseudobulk total read count N=475000 and K=100000 mixtures
    Heuristic balance of coverage, noise, and compute; affects deconvolver training distribution.
  • entropy-weight formula (max(q_i)−C⁻¹)/(1−C⁻¹) and NN loss MSE+0.5 KL
    Hand-designed loss weights not cross-validated exhaustively.
  • linear calibrator slopes a_c, intercepts b_c per cell type
    Fit on validation predicted vs true proportions; post-hoc correction parameters.
  • TCS expected-cell-type threshold (≥1% in Tabula Sapiens)
    Defines which cell types count as 'expected' for the OOD proxy metric.
assumptions (6)
  • domain assumption Methylation pattern to cell-type mapping is many-to-many due to imperfect maintenance, cell-cycle variation, purification noise, and technical noise.
    Biological rationale §3.2.2; justifies soft labels over hard labels.
  • domain assumption Reads are i.i.d.; classifier outputs can be aggregated by CpG-weighted averages per GR group.
    Standard in cited read-level methods; used in likelihood and aggregation (Eq. 2, Alg. 3).
  • standard math Empirical class frequencies among reads sharing signature s are the MLE of P(c|s).
    Classical categorical MLE (Murphy); Eq. 4.
  • domain assumption Pseudobulk mixtures from purified libraries approximate real bulk deconvolution difficulty.
    Standard proxy; Limitations §5 admits it is imperfect.
  • ad hoc to paper Cell types not expected in a tissue (via Tabula Sapiens CL/UBERON mapping) should receive near-zero predicted mass; TCS is a valid accuracy upper-bound proxy.
    OOD metric definition §4.1 / A.4.5 when ground truth is unavailable.
  • domain assumption DMR groups of Loyfer et al. (≤25 per cell type) are sufficiently informative for 39-way deconvolution.
    Region selection omitted from Syto; authors note strong dependence on regions (§5).
invented entities (4)
  • Syto modular framework (classifier + deconvolver + calibrator)
    purpose: Harmonize and scale read-level classification-based deconvolution.
    Organizational contribution; components are extensions of existing models plus new deconvolvers/calibrator.
  • Data-driven soft labels with asymmetric Jaccard pooling
    purpose: Estimate conditional cell-type distribution per omics signature so multi-class training converges.
    Core methodological invention; MLE construction is standard but pooling/reweighting/entropy weighting are paper-specific.
  • Linear calibration with clip-normalize or simplex projection for probability-simplex outputs
    purpose: Post-hoc correction of deconvolver bias without temperature/Dirichlet hyperparameters.
    Simple new calibrator; compared to Dirichlet/vector/temperature scaling in supplement.
  • Tissue Concordance Score (TCS)
    purpose: Proxy metric for OOD deconvolution when true proportions are unknown.
    Defined for this paper’s OOD evaluation; not an external benchmark.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Data-Driven Soft Labeling Scales DNA Read Classification to Whole-Body Cell-Type Deconvolution." pith.science (2026). https://pith.science/paper/QBEK7RXN

@misc{pith2026260704987,
  author       = {Pith},
  title        = {Pith review of: Data-Driven Soft Labeling Scales DNA Read Classification to Whole-Body Cell-Type Deconvolution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QBEK7RXN}},
  note         = {Machine review of arXiv:2607.04987}
}
abstract

Cell-type deconvolution, the task of estimating the proportions of constituent cell types in a heterogeneous biological sample, is a core problem in computational biology. Methods that rely on epigenetic marks such as DNA methylation typically operate on aggregated methylation estimates, discarding the pattern-level information carried by individual DNA reads. Existing read-level approaches that exploit this information are scarce, and all remain restricted to few-class settings; scaling them further is an open problem because, at scale, non-discriminative reads dominate and hard labels conflict with the many-to-many mapping between methylation patterns and cell types, preventing classifier convergence. To overcome this, we propose data-driven soft labels that estimate the conditional cell-type distribution for each read, and integrate this scheme into Syto, a new modular framework for read-level classification-based deconvolution. On a whole-body atlas of 39 human cell types, Syto reduces MSE by 2.56$\times$ over SoTA, with gains transferring to an out-of-distribution dataset spanning 16 tissues. Syto lays the foundation for modeling increasingly large cell-type panels, with improved applications in biology and healthcare. The proposed soft-labeling scheme is further translatable to any setting with a many-to-many signal-to-label mapping.

Figures

Figures reproduced from arXiv: 2607.04987 by the authors.

Figure 1
Figure 1. Principal scheme of the methylome-based cell-type deconvolu￾tion Epigenetic marks, such as histone modifications, chro￾matin accessibility and DNA methylation, collectively define cell identity and can be used to distinguish between different normal and malignant cells [1, 2]. Among these, DNA methylation stands out as the most stable and cost￾effective to profile at scale. Advances in methylome profiling have allow… view at source ↗
Figure 2
Figure 2. Proposed framework (Syto) for read-level classification-based cell-type deconvolution. We deliberately omit the selection of informative regions to emphasize that Syto can be applied out of the box with different GR groups, as illustrated in Figure S1. In particular, Syto can be applied to long reads by treating the whole genome as a single GR group, as in [17]. 3.2 Labeling 3.6 3.8 0 1000 2000 3000 4000 Steps 0.0 0… view at source ↗
Figure 3
Figure 3. Validation cross-entropy loss for MethylBERT using hard labeling with or without background class. Using standard one-hot encoded labels (or hard labels) as the classification target when the number of cell types is large impedes model convergence. Indeed, when decon￾volving C cell-types with an atlas where each GR group g is specific for a single cell type c (g) , only about 1/C (≈ 3% when C = 39) of reads originat… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Pairwise mean differences in TCS (row minus column, best configu￾ration per category) with p-values of the two-sided t-test in parentheses. SL: soft labels; DD: data-driven; HL: hard labels. OOD RRBS dataset. Each labeling scheme’s best classifier-deconvolver-calibrato…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

62 extracted references · 3 canonical work pages

  1. [1]

    Amount and distribution of 5-methylcytosine in human dna from different types of tissues or cells.Nucleic acids research, 10(8):2709–2721, 1982

    Melanie Ehrlich, Miguel A Gama-Sosa, Lan-Hsiang Huang, Rose Marie Midgett, Kenneth C Kuo, Roy A McCune, and Charles Gehrke. Amount and distribution of 5-methylcytosine in human dna from different types of tissues or cells.Nucleic acids research, 10(8):2709–2721, 1982

    1982

  2. [2]

    The 5-methylcytosine content of dna from human tumors

    Miguel A Gama-Sosa, Valerie A Slagel, Ronald W Trewyn, Ronald Oxenhandler, Kenneth C Kuo, Charles W Gehrke, and Melanie Ehrlich. The 5-methylcytosine content of dna from human tumors. Nucleic acids research, 11(19):6883–6894, 1983

    1983

  3. [3]

    Dna methylation-based classification of central nervous system tumors.Nature, 555(7697):469–474, 2018

    David Capper, David TW Jones, Martin Sill, V olker Hovestadt, Daniel Schrimpf, Dominik Sturm, Christian Koelsche, Felix Sahm, Lukas Chavez, David E Reuss, et al. Dna methylation-based classification of central nervous system tumors.Nature, 555(7697):469–474, 2018

    2018

  4. [4]

    Ultra-fast deep-learned cns tumor classification during surgery.Nature, 622(7984):842–849, 2023

    C Vermeulen, M Pagès-Gallego, L Kester, MEG Kranendonk, P Wesseling, N Verburg, P de Witt Hamer, EJ Kooi, L Dankmeijer, J Van Der Lugt, et al. Ultra-fast deep-learned cns tumor classification during surgery.Nature, 622(7984):842–849, 2023

    2023

  5. [5]

    crossnn is an explainable framework for cross-platform dna methylation-based classification of tumors.Nature cancer, 6(7):1283– 1294, 2025

    Dongsheng Yuan, Robin Jugas, Petra Pokorna, Jaroslav Sterba, Ondrej Slaby, Simone Schmid, Christin Siewert, Brendan Osberg, David Capper, Skarphedinn Halldorsson, et al. crossnn is an explainable framework for cross-platform dna methylation-based classification of tumors.Nature cancer, 6(7):1283– 1294, 2025

    2025

  6. [6]

    Systematic assessment of tumor purity and its clinical implications.JCO precision oncology, 4:995–1005, 2020

    Syed Haider, Svitlana Tyekucheva, Davide Prandi, Natalie S Fox, Jaeil Ahn, Andrew Wei Xu, Angeliki Pantazi, Peter J Park, Peter W Laird, Chris Sander, et al. Systematic assessment of tumor purity and its clinical implications.JCO precision oncology, 4:995–1005, 2020

    2020

  7. [7]

    Hallmarks of cancer: new dimensions.Cancer discovery, 12(1):31–46, 2022

    Douglas Hanahan. Hallmarks of cancer: new dimensions.Cancer discovery, 12(1):31–46, 2022

    2022

  8. [8]

    Hot and cold tumors: Immunological features and the therapeutic strategies

    Lianjie Wang, Hui Geng, Yujie Liu, Lei Liu, Yanhua Chen, Fanchen Wu, Zhiyi Liu, Shiliang Ling, Yan Wang, and Lihong Zhou. Hot and cold tumors: Immunological features and the therapeutic strategies. MedComm, 4(5):e343, 2023

    2023

Show all 62 references
  1. [9]

    Dna methylation arrays as surrogate measures of cell mixture distribution.BMC bioinformatics, 13:1–16, 2012

    Eugene Andres Houseman, William P Accomando, Devin C Koestler, Brock C Christensen, Carmen J Marsit, Heather H Nelson, John K Wiencke, and Karl T Kelsey. Dna methylation arrays as surrogate measures of cell mixture distribution.BMC bioinformatics, 13:1–16, 2012

    2012

  2. [10]

    Shuo Li, Weihua Zeng, Xiaohui Ni, Qiao Liu, Wenyuan Li, Mary L Stackpole, Yonggang Zhou, Arjan Gower, Kostyantyn Krysan, Preeti Ahuja, et al. Comprehensive tissue deconvolution of cell-free dna by deep learning for disease diagnosis and monitoring.Proceedings of the National A...

    2023

  3. [11]

    Comprehensive cell type decomposition of circulating cell-free dna with celfie.Nature communications, 12(1):2717, 2021

    Christa Caggiano, Barbara Celona, Fleur Garton, Joel Mefford, Brian L Black, Robert Henderson, Catherine Lomen-Hoerth, Andrew Dahl, and Noah Zaitlen. Comprehensive cell type decomposition of circulating cell-free dna with celfie.Nature communications, 12(1):2717, 2021

    2021

  4. [12]

    Cell type deconvolution of methylated cell-free dna at the resolution of individual reads.NAR Genomics and Bioinformatics, 5(2):lqad048, 2023

    Pia Keukeleire, Stavros Makrodimitris, and Marcel Reinders. Cell type deconvolution of methylated cell-free dna at the resolution of individual reads.NAR Genomics and Bioinformatics, 5(2):lqad048, 2023

    2023

  5. [13]

    A dna methylation atlas of normal human cell types.Nature, 613(7943):355–364, 2023

    Netanel Loyfer, Judith Magenheim, Ayelet Peretz, Gordon Cann, Joerg Bredno, Agnes Klochendler, Ilana Fox-Fisher, Sapir Shabi-Porat, Merav Hecht, Tsuria Pelet, et al. A dna methylation atlas of normal human cell types.Nature, 613(7943):355–364, 2023

    2023

  6. [14]

    Computational deconvolution of dna methylation data from mixed dna samples.Briefings in Bioinformatics, 25(3):bbae234, 2024

    Maísa R Ferro dos Santos, Edoardo Giuili, Andries De Koker, Celine Everaert, and Katleen De Preter. Computational deconvolution of dna methylation data from mixed dna samples.Briefings in Bioinformatics, 25(3):bbae234, 2024

    2024

  7. [15]

    Cost-effective methylome sequencing of cell-free dna for accurately detecting and locating cancer.Nature communications, 13(1):5566, 2022

    Mary L Stackpole, Weihua Zeng, Shuo Li, Chun-Chi Liu, Yonggang Zhou, Shanshan He, Angela Yeh, Ziye Wang, Fengzhu Sun, Qingjiao Li, et al. Cost-effective methylome sequencing of cell-free dna for accurately detecting and locating cancer.Nature communications, 13(1):5566, 2022

    2022

  8. [16]

    Dna hypomethylation in cancer cells.Epigenomics, 1(2):239–259, 2009

    Melanie Ehrlich. Dna hypomethylation in cancer cells.Epigenomics, 1(2):239–259, 2009

    2009

  9. [17]

    Genome-wide classification of tumor-derived reads from bulk long-read sequencing.bioRxiv, pages 2026–03, 2026

    Toby M Baker, Nedas Matulionis, Cassidy Andrasz, Dan Gerke, Natalia Garcia-Dutton, David Atkinson, Kami Chiotti, Selina Wu, Suchita Lulla, Jieun Oh, et al. Genome-wide classification of tumor-derived reads from bulk long-read sequencing.bioRxiv, pages 2026–03, 2026. 10

    2026

  10. [18]

    Wenyuan Li, Qingjiao Li, Shuli Kang, Mary Same, Yonggang Zhou, Carol Sun, Chun-Chi Liu, Lea Matsuoka, Linda Sher, Wing Hung Wong, et al. Cancerdetector: ultrasensitive and non-invasive cancer detection at the resolution of individual reads using cell-free dna methylation seque...

    2018

  11. [19]

    Jiaqi Li, Lei Wei, Xianglin Zhang, Wei Zhang, Haochen Wang, Bixi Zhong, Zhen Xie, Hairong Lv, and Xiaowo Wang. Dismir: Deep learning-based noninvasive cancer detection by integrating dna sequence and methylation information of individual cell-free dna reads.Briefings in bioinf...

    2021

  12. [20]

    Methylbert enables read-level dna methylation pattern identification and tumor deconvolution using a transformer-based model.Nature Communications, 16(1):788, 2025

    Yunhee Jeong, Clarissa Gerhäuser, Guido Sauter, Thorsten Schlomm, Karl Rohr, and Pavlo Lutsik. Methylbert enables read-level dna methylation pattern identification and tumor deconvolution using a transformer-based model.Nature Communications, 16(1):788, 2025

    2025

  13. [21]

    Bert: Pre-training of deep bidi- rectional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidi- rectional transformers for language understanding. InProceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human lang...

    2019

  14. [22]

    Focal loss for dense object detection

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Dollár. Focal loss for dense object detection. InProceedings of the IEEE International Conference on Computer Vision, pages 2980–2988, 2017

    2017

  15. [23]

    Label distribution learning.IEEE Transactions on Knowledge and Data Engineering, 28(7): 1734–1748, 2016

    Xin Geng. Label distribution learning.IEEE Transactions on Knowledge and Data Engineering, 28(7): 1734–1748, 2016

    2016

  16. [24]

    Label enhancement for label distribution learning.IEEE Transactions on Knowledge and Data Engineering, 33(4):1632–1643, 2021

    Ning Xu, Yun-Peng Liu, and Xin Geng. Label enhancement for label distribution learning.IEEE Transactions on Knowledge and Data Engineering, 33(4):1632–1643, 2021. doi: 10.1109/TKDE.2019. 2947040

    2021 doi

  17. [25]

    Classification with label distribution learning

    Jing Wang and Xin Geng. Classification with label distribution learning. InIJCAI, volume 1, page 2, 2019

    2019

  18. [26]

    Rethinking the inception architecture for computer vision

    Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jon Shlens, and Zbigniew Wojna. Rethinking the inception architecture for computer vision. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 2818–2826, 2016

    2016

  19. [27]

    When does label smoothing help?Advances in neural information processing systems, 32, 2019

    Rafael Müller, Simon Kornblith, and Geoffrey E Hinton. When does label smoothing help?Advances in neural information processing systems, 32, 2019

    2019

  20. [28]

    On calibration of modern neural networks

    Chuan Guo, Geoff Pleiss, Yu Sun, and Kilian Q Weinberger. On calibration of modern neural networks. In International conference on machine learning, pages 1321–1330. PMLR, 2017

    2017

  21. [29]

    Beyond temperature scaling: Obtaining well-calibrated multi-class probabilities with dirichlet calibration

    Meelis Kull, Miquel Perello Nieto, Markus Kängsepp, Telmo Silva Filho, Hao Song, and Peter Flach. Beyond temperature scaling: Obtaining well-calibrated multi-class probabilities with dirichlet calibration. Advances in neural information processing systems, 32, 2019

    2019

  22. [30]

    Fidelity of the methylation pattern and its variation in the genome.Genome Research, 13(5): 868–874, 2003

    Toshikazu Ushijima, Naoko Watanabe, Eriko Okochi, Atsushi Kaneda, Takashi Sugimura, and Kazuaki Miyamoto. Fidelity of the methylation pattern and its variation in the genome.Genome Research, 13(5): 868–874, 2003. doi: 10.1101/gr.969603

    2003 doi

  23. [31]

    Single-cell multi-omic detection of dna methylation and histone modifications reconstructs the dynamics of epigenomic maintenance.Nature Methods, 22(10):2042–2051, Oct 2025

    Christoph Geisenberger, Jeroen van den Berg, Vincent van Batenburg, Buys de Barbanson, Anna Lyubi- mova, Joe Verity-Legg, Xiufei Chen, Yibin Liu, Chun-Xiao Song, Jeroen de Ridder, and Alexander van Oudenaarden. Single-cell multi-omic detection of dna methylation and histone mo...

    doi

  24. [32]

    Drautz-Moses, Stephan C

    Li Zhou, Hong Kiat Ng, Daniela I. Drautz-Moses, Stephan C. Schuster, Stephan Beck, Changhoon Kim, John Campbell Chambers, and Marie Loh. Systematic evaluation of library preparation methods and sequencing platforms for high-throughput whole genome bisulfite sequencing.Scientif...

    2019 doi

  25. [33]

    Matthias Zilbauer, Tim F Rayner, Christine Clark, Alison J Coffey, Chris J Joyce, Priit Palta, Aarno Palotie, Paul A Lyons, and Kenneth GC Smith. Genome-wide methylation analyses of primary human leukocyte subsets identifies functionally important cell-type–specific hypomethyl...

    2013

  26. [34]

    Dna-binding factors shape the mouse methylome at distal regulatory regions.Nature, 480(7378):490–495, 2011

    Michael B Stadler, Rabih Murr, Lukas Burger, Robert Ivanek, Florian Lienert, Anne Schöler, Erik van Nimwegen, Christiane Wirbelauer, Edward J Oakeley, Dimos Gaidatzis, et al. Dna-binding factors shape the mouse methylome at distal regulatory regions.Nature, 480(7378):490–495, 2011

    2011

  27. [35]

    Function and information content of dna methylation.Nature, 517(7534):321–326, 2015

    Dirk Schübeler. Function and information content of dna methylation.Nature, 517(7534):321–326, 2015

    2015

  28. [36]

    Murphy.Machine learning : a probabilistic perspective

    Kevin P. Murphy.Machine learning : a probabilistic perspective. Adaptive computation and machine learning series. MIT, 2012. ISBN 9780262018029 0262018020

    2012

  29. [37]

    A study of the robustness of knn classifiers trained using soft labels

    Neamat El Gayar, Friedhelm Schwenker, and Günther Palm. A study of the robustness of knn classifiers trained using soft labels. InIAPR Workshop on Artificial Neural Networks in Pattern Recognition, pages 67–80. Springer, 2006

    2006

  30. [38]

    Transformers: State-of-the-art natural language processing

    Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Rémi Louf, Morgan Funtowicz, et al. Transformers: State-of-the-art natural language processing. InProceedings of the 2020 conference on empirical methods in nat...

    2020

  31. [39]

    Xgboost: A scalable tree boosting system

    Tianqi Chen and Carlos Guestrin. Xgboost: A scalable tree boosting system. InProceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, pages 785–794, 2016

    2016

  32. [40]

    Pytorch: An imperative style, high-performance deep learning library.Advances in neural information processing systems, 32, 2019

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. Pytorch: An imperative style, high-performance deep learning library.Advances in neural information processing systems, 32, 2019

    2019

  33. [41]

    Cvxpy: A python-embedded modeling language for convex optimiza- tion.Journal of Machine Learning Research, 17(83):1–5, 2016

    Steven Diamond and Stephen Boyd. Cvxpy: A python-embedded modeling language for convex optimiza- tion.Journal of Machine Learning Research, 17(83):1–5, 2016

    2016

  34. [42]

    Efficient projections onto the l1-ball for learning in high dimensions.Proceedings of the 25th International Conference on Machine Learning, pages 272–279, 07 2008

    John Duchi, Shai Shalev-Shwartz, Yoram Singer, and Tushar Chandra. Efficient projections onto the l1-ball for learning in high dimensions.Proceedings of the 25th International Conference on Machine Learning, pages 272–279, 07 2008. doi: 10.1145/1390156.1390191

    2008 doi

  35. [43]

    Megakaryocyte-and erythroblast-specific cell-free dna patterns in plasma and platelets reflect thrombopoiesis and erythropoiesis levels.Nature communications, 14(1):7542, 2023

    Joshua Moss, Roni Ben-Ami, Ela Shai, Ofer Gal-Rosenberg, Yosef Kalish, Agnes Klochendler, Gordon Cann, Benjamin Glaser, Ariela Arad, Ruth Shemer, et al. Megakaryocyte-and erythroblast-specific cell-free dna patterns in plasma and platelets reflect thrombopoiesis and erythropoi...

    2023

  36. [44]

    Tabula sapiens reveals transcription factor expression, senescence effects, and sex-specific features in cell types from 28 human organs and tissues.bioRxiv, pages 2024–12, 2025

    Stephen R Quake, Tabula Sapiens Consortium, et al. Tabula sapiens reveals transcription factor expression, senescence effects, and sex-specific features in cell types from 28 human organs and tissues.bioRxiv, pages 2024–12, 2025

    2024

  37. [45]

    The tabula sapiens: A multiple-organ, single-cell transcriptomic atlas of humans.Science, 376(6594):eabl4896, 2022

    The Tabula Sapiens Consortium*, Robert C Jones, Jim Karkanias, Mark A Krasnow, Angela Oliveira Pisco, Stephen R Quake, Julia Salzman, Nir Yosef, Bryan Bulthaup, Phillip Brown, et al. The tabula sapiens: A multiple-organ, single-cell transcriptomic atlas of humans.Science, 376(...

    2022

  38. [46]

    Better bootstrap confidence intervals.Journal of the American Statistical Association, 82 (397):171–185, 1987

    Bradley Efron. Better bootstrap confidence intervals.Journal of the American Statistical Association, 82 (397):171–185, 1987. ISSN 01621459, 1537274X. URLhttp://www.jstor.org/stable/2289144

    1987

  39. [47]

    Benchmarking of methods for dna methylome deconvolution.Nature Communications, 15(1):4134, 2024

    Kobe De Ridder, Huiwen Che, Kaat Leroy, and Bernard Thienpont. Benchmarking of methods for dna methylome deconvolution.Nature Communications, 15(1):4134, 2024

    2024

  40. [48]

    Systematic evaluation of methylation-based cell type deconvolution methods for plasma cell-free dna.Genome biology, 25(1):318, 2024

    Tongyue Sun, Jinqi Yuan, Yacheng Zhu, Jingqi Li, Shen Yang, Junpeng Zhou, Xinzhou Ge, Susu Qu, Wei Li, Jingyi Jessica Li, et al. Systematic evaluation of methylation-based cell type deconvolution methods for plasma cell-free dna.Genome biology, 25(1):318, 2024

    2024

  41. [49]

    A benchmark of dna methylation deconvolution methods for tumoral fraction estimation using deconflow.bioRxiv, pages 2025–11, 2025

    Edoardo Giuili, Sofie Van de Velde, Sam Kint, Maísa R Ferro dos Santos, Lotte Cornelli, Sofie Roelandt, Kathleen Schoofs, Renske Imschoot, Ruben Van Paemel, Leander Meuris, et al. A benchmark of dna methylation deconvolution methods for tumoral fraction estimation using deconf...

    2025

  42. [50]

    Guidelines on optimizing dna methylation reference panels for cell-type deconvolution.Communications Biology, 2026

    Xiaolong Guo and Andrew E Teschendorff. Guidelines on optimizing dna methylation reference panels for cell-type deconvolution.Communications Biology, 2026

    2026

  43. [51]

    Systematic evaluation of cell-type deconvolution pipelines for sequencing-based bulk dna methylomes.Briefings in bioinformatics, 23(4):bbac248, 2022

    Yunhee Jeong, Lisa Barros de Andrade e Sousa, Dominik Thalmeier, Reka Toth, Marlene Ganslmeier, Kersten Breuer, Christoph Plass, and Pavlo Lutsik. Systematic evaluation of cell-type deconvolution pipelines for sequencing-based bulk dna methylomes.Briefings in bioinformatics, 2...

    2022

  44. [52]

    wgbstools: a computational suite for dna methylation sequencing data analysis.Life Science Alliance, 9(4), 2026

    Netanel Loyfer, Jonathan Rosenski, and Tommy Kaplan. wgbstools: a computational suite for dna methylation sequencing data analysis.Life Science Alliance, 9(4), 2026. 12

    2026

  45. [53]

    removal of ‘confounding’ markers

    James McLaughlin, Josh Lagrimas, Haider Iqbal, Helen Parkinson, and Henriette Harmse. Ols4: a new ontology lookup service for a growing interdisciplinary knowledge ecosystem.Bioinformatics, 41(5): btaf279, 2025. 13 A Supplementary Material A.1 Broader impact Read-level classif...

    2025

  46. [54]

    Filtering the reads of the bulk sample to deconvolve by keeping only the reads in GR groups that were seen in the training data, obtaining the following dataset:D ′ :={(r i, gi, si)}N ′ i=1

  47. [55]

    Computing, for all (ri, gi, si)∈ D ′, the likelihood of the read to belong to each class c∈ C according to the beta distribution parameters of GR groupg i and classc: ℓ(ri|c) = LiY j=1 P(s i,j |Beta(η c gi , ρc gi)),(8) = LiY j=1 Z 1 0 psi,j(1−p) 1−si,j pηc gi −1(1−p) ρc gi −1...

  48. [56]

    In our experiments, we use as class prior both the distribution of classes in the training data (which is the mathematically correct choice) and the uniform distribution

    Computing, for all (ri, gi, si)∈ D ′, the posterior probability of the read to belong to each classc∈ Caccording to Bayes’ rule: P(c|r i, gi, si) = ℓ(ri|c)P(c)P c′∈C ℓ(ri|c′)P(c ′) ,(12) whereP(c)is the prior probability of classc. In our experiments, we use as class prior bot...

  49. [57]

    top-25 unmethylated markers atlases

    with 10,000 resamples. Calibrator R2 LoA LoA worst-class MAE MSE KL (×10−2) (×10 −2) (×10 −2) (×10 −3) (×10 −4) (×10 −2) No calibration 96.39[96.34, 96.43][-3.39, 3.39] [-6.96, 9.68] 5.59[5.57, 5.61]3.00[2.96, 3.04]14.54[14.42, 14.66] Linear cal. (clip-norm) 96.88[96.85, 96.92...

  50. [58]

    We selected U25 Best Based on Proxy GSS (right graph) by subsetting the Top-25 regions per cell type from the U250 atlas, ordered by Proxy GSS

    Colon-Fibro U25 Original Atlas Mean: 0.709 0.2 0.0 0.2 0.4 0.6 0.8 1.0 Proxy GSS (Higher is better) U25 Best Based on Proxy GSS Mean: 0.753 Figure S3: Proxy Gap Specificity Scores or difference between proportion of on-target unmethylated reads and maximal proportion of unmeth...

  51. [59]

    If a target cell type is covered by a single sample, then the data from this sample is divided across all three splits

  52. [60]

    If a target cell type is covered by exactly two samples, one is allocated to train, and the other is split between validation and test sets

  53. [61]

    Tissue Concordance Score

    If a target cell type is covered by three or more biological samples, those samples must not be individually broken down into separate splits, and the validation and test splits must receive at least one full sample. Applying the described procedure results in the final train/...

  54. [62]

    *”, “**” and “***

    As Figure S8 reveals, linear calibration with clip-normalize, linear calibration with simplex projection and vector scaling slightly improve upon the original baseline scores (+1.24, p≤0.001; +3.21, p≤0.001 and +3.05, p≤0.001 respectively). The best results are achieved with t...

Pith tools

Reviewed July 11, 2026 · model on record in the stance chip above.