Pith. sign in

REVIEW 3 major objections 6 minor 47 references

Timestamp calibration for time-series single cell RNA-seq expression data

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read ScPace introduces a latent-variable classifier to detect and correct noisy collection-time labels in time-series single-cell RNA-seq, improving timestamp annotation and supervised pseudotime analysis.

desk verdict ScPace is a plausible tool for a real problem, but the paper's central claim—that its loss threshold identifies mislabeled timestamps—is under-supported, and the pseudotime evaluation is circular. read the letter →

arxiv 2412.03027 v2 pith:T3764ZH7 submitted 2024-12-04 q-bio.GN

Xiran Chen , Sha Lin , Xiaofeng Chen , Weikai Li , Yifei Li This is my paper · ORCID
classification q-bio.GN
keywords TimestampcalibrationTime-seriessingle-cellRNA-seqSelf-pacedlearningLabelnoiseLatentvariableindicatorSupportvectormachineSupervisedpseudotimeanalysisScPace
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

The paper seeks to establish that noisy timestamps—cells assigned to the wrong collection time—are a correctable bottleneck in time-series single-cell RNA-seq analysis, and that the most effective correction comes from a latent-variable indicator that lets a classifier flag its own suspicious training samples instead of sampling candidates probabilistically. It introduces ScPace, an SVM-based calibration method that iteratively trains only on cells it currently considers clean, computes a multiclass hinge loss for every cell, and then deletes or reclassifies cells whose loss exceeds a threshold. On simulated and real datasets with artificially injected swap and random mislabeling up to 40%, ScPace reports the highest accuracy among the compared classifiers, including 97.33% on the linear simulated dataset Sim1 under swap mislabeling, and it improves the correlation of supervised pseudotime analysis on real datasets. If the claim holds, timestamp calibration becomes a practical preprocessing step that makes downstream trajectory inference more reliable without requiring users to discard whole time points.

What carries the argument

The load-bearing object is a per-cell latent variable v_i ∈ {0,1} (Eq. 16) that is optimized together with an SVM. After each training round, ScPace computes the Crammer-Singer multiclass hinge loss (Eq. 19) for every cell from one-vs-one RBF SVM decision values converted to one-vs-rest scores by Algorithm 1 and Eqs. (17)-(18); cells with loss above the threshold λ are assigned v_i=0 and excluded from the next round, while cells below threshold are kept. The regularization parameter C grows by a rate p each iteration to avoid underfitting after sample removal, and class-specific penalties handle imbalanced timestamps. This mechanism replaces probability-based sample selection with deterministic margin-based selection, which is the step that carries the argument.

What would settle it

Run ScPace on a simulated four-timestamp dataset with 40% swap mislabeling and compare its flagged cells against the known mislabeled set; if the flagged cells do not substantially overlap the true mislabeled cells (low recall) or if reclassifying them does not beat a baseline SVM trained on clean labels, the central claim would be falsified.

Watch

Extended reading notes

Core claim

The central claim is that a binary latent variable v_i, updated by thresholding a multiclass hinge loss, separates correctly timestamped cells from mislabeled cells in time-series scRNA-seq data, and that acting on those flags improves both timestamp automatic annotation and supervised pseudotime analysis. The paper argues that prior calibration approaches select candidate noisy samples by probability sampling, which can pick unimportant cells, whereas ScPace selects deterministically by loss magnitude after training a weighted one-vs-one SVM. In cross-validation with swap and random mislabeling, ScPace stays above roughly 90% accuracy even at a 40% mislabeling rate and outperforms the existing semi-supervised calibration method and confident-learning baselines on most simulated and real datasets. The authors also report that reclassifying or deleting the flagged cells before running supervised pseudotime analysis raises the resulting trajectory correlations, and they position ScPace as the first method specifically designed for timestamp calibration.

Load-bearing premise

The load-bearing assumption is that a single loss threshold separates genuinely mislabeled cells from cells that are correctly labeled but biologically unusual, so that removing or relabeling high-loss cells removes noise rather than real developmental states.

Editorial extensions

If this is right

  • Time-series scRNA-seq pipelines can treat ScPace as a preprocessing step that cleans timestamps before timestamp annotation or supervised pseudotime, improving trajectory correlations on noisy data.
  • Because ScPace maintains accuracy above about 90% even at 40% swap mislabeling on simulated data, downstream analyses can tolerate higher label noise than previously assumed.
  • The deletion mode suits technical noise, while the reclassification mode is recommended when high-loss cells reflect overlapping biological states, as the GSE90047 case shows.
  • ScPace can run with or without PCA or kernel PCA, so it remains applicable to datasets where dimension reduction hides informative signal.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same latent-variable thresholding scheme could be transferred to other ordinal label-noise problems, such as batch-effect labels or progression stages, whenever a base classifier produces per-sample margins.
  • Because the paper's sensitivity analysis identifies the regularization parameter C as the most performance-sensitive hyperparameter, an adaptive C schedule may matter more than threshold tuning in real deployments.
  • A diagnostic that separates technical from biological high-loss cells—for example, stability of v=0 flags across bootstrap resamples—could decide automatically between deletion and reclassification, which the paper currently leaves as a user choice.
  • The reported failure above 50% mislabeling suggests a natural stress test: at 60% noise the clean subset may no longer represent the true distribution, and combining ScPace with robust ordinal regression could push the usable noise range higher.
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 proposes ScPace, a timestamp calibration method for time-series scRNA-seq data. The method augments a one-vs-one support vector machine with a latent-variable indicator v_i (Eq. 16) that, after each training iteration, flags cells whose multi-class hinge loss (Eq. 19) exceeds a threshold lambda; flagged cells are either deleted or reclassified. The SVM is re-weighted and the regularization parameter C is increased over iterations (Algorithm 2). The authors evaluate ScPace in stratified cross-validation experiments with artificially injected swap and random mislabeling on three simulated datasets (Splatter linear and bifurcation paths) and four real time-series datasets, reporting accuracy and a correction score (Eq. 25) against ScReclassify, CleanLab, and several baseline classifiers. They also apply the calibrated timestamps to Psupertime supervised pseudotime analysis and report that ScPace enhances pseudotime performance more than the alternatives. The paper claims that ScPace is the first method designed specifically for timestamp calibration and that its latent-variable selection mechanism improves over probability-sampling approaches.

Significance. If the stated results hold, ScPace would be a practically useful contribution: timestamp noise is a real concern in time-series scRNA-seq, and the paper provides a concrete algorithmic proposal, a public implementation, and a fairly broad experimental setup that includes simulated and real data, two noise models, several mislabeling rates, and sensitivity analyses. The simulation-based cross-validation against artificially mislabeled ground truth is a sensible evaluation strategy, and the authors should be credited for making the code available and for reporting computational timings. However, the central comparative claims are currently not established because of three load-bearing methodological issues: potential information leakage in the dimensionality-reduction step, a circular metric for pseudotime enhancement, and the absence of direct validation that the latent-variable flag actually identifies mislabeled cells rather than hard or boundary cells. These issues affect the interpretation of Figures 4-7, Tables 4-5, and the core claim of superiority over ScReclassify and CleanLab.

major comments (3)
  1. [Section 3.1.1, Algorithm 2, Section 4.2.1] The manuscript does not state whether PCA/KernelPCA is fitted on the training folds only in the cross-validation experiments. Algorithm 2 explicitly says 'Conduct PCA or KernelPCA on Original Datasets', and Section 3.1.1 says dimensional reduction is conducted on 'both training and testing sets'. If PCA or KernelPCA is fitted on the full dataset before the k-fold split, information from the holdout test set leaks into the training procedure, which would inflate the reported accuracy and correction scores in Figures 4-7. The authors correctly specify for ScReclassify that dimensionality reduction on the test set should use parameters determined from the training set, but no such statement is given for ScPace. This needs to be fixed by fitting the reduction inside each training fold, and the paper should report whether the conclusions change.
  2. [Section 4.3.1, Tables 4 and 5] The pseudotime 'enhancement' results are measured as the Spearman and Kendall correlation between the Psupertime pseudotime values and the calibrated timestamps — that is, the timestamps produced by the very calibration method being evaluated. On the original real datasets, this measures self-consistency rather than agreement with any independent ground truth. On simulated data, the paper should instead correlate the pseudotime with the known true simulation time, and on real data it should compare against the original uncalibrated timestamps or an external biological ordering. As written, the larger correlations in Tables 4 and 5 may partly reflect the calibration procedure making the labels more self-consistent, not making them more correct, so the enhancement claim is not supported.
  3. [Section 3.1.3, Eq. (16), Eq. (25), Section 5.1] The paper never directly validates that cells flagged as noisy (v_i = 0) coincide with the artificially injected mislabels. The correction score in Eq. (25) measures the fraction of mislabeled cells whose final reclassified label matches the original true label; it does not report detection precision or recall of the v-flag. The GSE90047 case study in Section 5.1 explicitly finds that overlapping developmental states between E14.5 and E15.5 are flagged as noisy and that the apparent noise 'may reflect biological variability', which the authors themselves use to argue against deletion for that dataset. This concession shows that the threshold in Eqs. (16) and (21) does not separate label errors from genuine boundary cells. Consequently, the deletion-based pseudotime improvements in Tables 4 and 5 could arise from removing hard-to-order cells rather than from correcting timestamp labels. The authors should report precision/recall of the v-flag against known mislabels in the simulations and examine the stability of deletion-based gains when the threshold is varied.
minor comments (6)
  1. [Algorithm 2] The instruction 'Conduct PCA or KernelPCA on Original Datasets' conflicts with the need for leakage-free cross-validation; please clarify the intended usage for both the calibration setting and the cross-validation setting.
  2. [Eq. (5)] In Eq. (5), the symbol m is used both for the original feature dimension (Eq. 1) and for the number of classes in the class-penalty vector; please use distinct notation.
  3. [Section 4.2.1] Please clarify whether artificial mislabeling is applied only to the training folds while the test folds retain their original clean labels, and how the correction score in Eq. (25) is aggregated across folds.
  4. [Figure 16] The caption of Figure 16 repeats the caption of Figure 1; it should describe the cardiomyocyte prediction result shown in the figure.
  5. [Section 6] In Section 6, '125Gi' should be '125 GiB'.
  6. [Abstract and Introduction] The claim that ScPace is 'the first method specifically designed' for timestamp calibration is strong and should be supported by a more systematic comparison with existing label-noise and timestamp-calibration approaches.

Circularity Check

1 steps flagged · score 6.0 of 10

ScPace's supervised-pseudotime 'enhancement' is measured against ScPace's own calibrated timestamps, so that headline claim is self-referential rather than independently validated; the core classification benchmark is not circular.

  1. self definitional [Section 4.3.1 (Experimental Setup) and Section 4.3.2 / Tables 4-5 (Enhancement of supervised pseudotime analysis)]
    "We applied Psupertime to the reclassified timestamps from both methods independently and then computed the average training accuracy, testing accuracy and pseudotime value to each cells. To assess the performance of the computed pseudotime, we calculated the Spearman and Kendall correlation between the computed pseudotime value and the calibrated timestamps."

    The 'calibrated timestamps' used as the reference are the outputs of the calibration method being evaluated. Psupertime is trained on those same reclassified (or deletion-filtered) timestamps, so the reported Spearman/Kendall improvement measures how well the pseudotime fits ScPace's own labels, not how close the calibrated labels are to true time. An ordinal-label model can appear 'enhanced' whenever the calibration makes the labels more self-consistent with the classifier, even if the reclassification moved cells away from their true timestamps. The claimed enhancement in Tables 4-5 therefore reduces, by construction, to a self-consistency check rather than an independent validation of timestamp accuracy.

full rationale

The cross-validation experiments in Section 4.2 are genuinely informative: artificial mislabels are injected with known identity, training labels are corrupted, holdout labels are clean, and the correction score in Eq. (25) compares reclassifications to the original true labels. That part of the paper is self-contained and supports ScPace's classification robustness. The circularity lies in the second headline claim. Section 4.3.1 defines the pseudotime enhancement metric as the correlation between Psupertime's output and the 'calibrated timestamps' produced by the method under evaluation; Psupertime is itself trained on those timestamps. Therefore the reported Spearman/Kendall gains in Tables 4-5 and Figures 12-13 can be inflated simply by making the ordinal labels easier to fit, without establishing that the calibrated labels are closer to true developmental time. The GSE90047 case study reinforces this concern by acknowledging that high-loss cells flagged as 'noisy' may instead reflect biological variability. No load-bearing self-citation chain is present; the core classification result has external ground truth. Overall, one central evaluation claim reduces to self-consistency by construction, while the classification contribution remains independently supported.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central method relies on no physical or biological entities beyond the data. Its assumptions are about how label noise manifests: that a hard loss threshold can separate mislabeled samples from clean ones, that the two synthetic mislabeling schemes represent real timestamp errors, and that Splatter simulations capture the structure of real time-series scRNA-seq data. These are modeling assumptions rather than standard mathematical facts.

free parameters (5)
  • lambda (loss threshold) = grid-searched over {0.1, 0.5, 1, 2}; final value not reported
    Controls the trade-off between retaining and discarding samples in Eqs. (16) and (21); the central selection mechanism.
  • C (SVM regularization) = grid-searched over {0.1, 1, 10}; final value not reported
    Controls SVM penalty strength; sensitivity analysis shows it is the most influential hyperparameter.
  • p (C-growth rate) = grid-searched over {0.1, 0.5, 1}; final value not reported
    Increases C each iteration to avoid excessive sample removal.
  • iter (number of calibration iterations) = grid-searched over {5, 10}; final value not reported
    Number of self-paced retraining rounds.
  • m' (reduced dimension after PCA/KernelPCA) = 10 or 20 set by rule in Eq. (1)
    Optional preprocessing dimensionality, not data-fitted but chosen by a fixed rule.
assumptions (4)
  • ad hoc to paper Cells with multi-class hinge loss above the threshold lambda are mislabeled and can be removed or reclassified.
    Eqs. (16) and (21) implement this hard cutoff; no independent evidence is given that loss separates noise from biological heterogeneity, and the GSE90047 case study shows detected high-loss cells include overlapping but biologically meaningful cholangiocytes.
  • ad hoc to paper Swap mislabeling (neighbor time points) and random mislabeling faithfully model real timestamp errors.
    Section 4.2.1 defines these two artificial noise models; the claim of real-world robustness depends on them resembling real label noise.
  • domain assumption Splatter-simulated trajectories with dropout faithfully reproduce the noise structure of real time-series scRNA-seq data.
    Section 4.1 uses Splatter to generate Sim1-Sim3; conclusions about robustness on real data are partly extrapolated from these simulations.
  • domain assumption Increasing the correlation between Psupertime's pseudotime and the calibrated timestamps indicates better temporal accuracy.
    Section 4.3 evaluates enhancement against labels produced by ScPace itself, so the correlation measures consistency with the calibrated labels rather than with true developmental time.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Timestamp calibration for time-series single cell RNA-seq expression data." pith.science (2026). https://pith.science/paper/T3764ZH7

@misc{pith2026241203027,
  author       = {Pith},
  title        = {Pith review of: Timestamp calibration for time-series single cell RNA-seq expression data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/T3764ZH7}},
  note         = {Machine review of arXiv:2412.03027}
}
read the original abstract

Timestamp automatic annotation (TAA) is a crucial procedure for analyzing time-series ScRNA-seq data, as they unveil dynamic biological developments and cell regeneration process. However, current TAA methods heavily rely on manual timestamps, often overlooking their reliability. This oversight can significantly degrade the performance of timestamp automatic annotation due to noisy timestamps. Nevertheless, the current approach for addressing this issue tends to select less critical cleaned samples for timestamp calibration. To tackle this challenge, we have developed a novel timestamp calibration model called ScPace for handling noisy labeled time-series ScRNA-seq data. This approach incorporates a latent variable indicator within a base classifier instead of probability sampling to detect noisy samples effectively. To validate our proposed method, we conducted experiments on both simulated and real time-series ScRNA-seq datasets. Cross-validation experiments with different artificial mislabeling rates demonstrate that ScPace outperforms previous approaches. Furthermore, after calibrating the timestamps of the original time-series ScRNA-seq data using our method, we performed supervised pseudotime analysis, revealing that ScPace enhances its performance significantly. These findings suggest that ScPace is an effective tool for timestamp calibration by enabling reclassification and deletion of detected noisy labeled samples while maintaining robustness across diverse ranges of time-series ScRNA-seq datasets. The source code is available at https://github.com/OPUS-Lightphenexx/ScPace.

Figures

Figures reproduced from arXiv: 2412.03027 by the authors.

Figure 1
Figure 1. Overall procedure of ScPace including data preprocessing, training ScPace and conduct two types of timestamp calibration [PITH_FULL_IMAGE:figures/full_fig_p019_1.png] view at source ↗
Figure 2
Figure 2. Visualization of the simulated datasets simulated by Splatter, Sim1 and Sim2 are class balanced while Sim3 is imbalanced. Chen et al.: Preprint submitted to Elsevier Page 19 of 19 [PITH_FULL_IMAGE:figures/full_fig_p019_2.png] view at source ↗
Figure 3
Figure 3. Demonstration of Swap Mislabeling and Random Mislabeling [PITH_FULL_IMAGE:figures/full_fig_p020_3.png] view at source ↗
Figures from the paper (13 more)
Figure 4
Figure 4. Figure 4: Cross validation results with different mislabeling rate on simulated datasets (A-C) cross validation results with swap mislabeling (D-F) cross validation results with random mislabeling Chen et al.: Preprint submitted to Elsevier Page 20 of 19 [PITH_FULL_IMAGE:figure…
Figure 5
Figure 5. Figure 5: Cross validation results with different mislabeling rate on real time-series ScRNA-seq datasets (A-D) cross validation results with swap mislabeling (E-H) cross validation results with random mislabeling [PITH_FULL_IMAGE:figures/full_fig_p021_5.png]
Figure 6
Figure 6. Figure 6: Correction score of cross validation with different mislabeling rate on simulated datasets (A-C) correction score results with swap mislabeling (D-F) correction score results with random mislabeling Chen et al.: Preprint submitted to Elsevier Page 21 of 19 [PITH_FULL_…
Figure 7
Figure 7. Figure 7: Correction score of cross validation with different mislabeling rate on real-time-series ScRNA-seq data (A-D) correction score results with swap mislabeling (E-H) correction score results with random mislabeling Datasets Methods Training time(Sec) GSE90047 CL 115.07 Sc…
Figure 8
Figure 8. Figure 8: Sensitive analysis of four hyperparameters(C, 𝜆, iter, p) on simulated datasets. Each datasets contains two rows, first row represents the cross validation sensitive analysis on swap mislabeling, second row represents the cross validation sensitive analysis on random m…
Figure 9
Figure 9. Figure 9: Sensitive analysis of four hyperparameters(C, 𝜆, iter, p) on real time-series ScRNA-seq data. Each datasets contains two rows, first row represents the cross validation sensitive analysis on swap mislabeling, second row represents the cross validation sensitive analysi…
Figure 10
Figure 10. Figure 10: Sensitive analysis results using different dimensional reduction technique on simulated datasets (A-C) Cross validation results with different dimensional reduction technique on swap mislabeling (D-F) Cross validation results with different dimensional reduction techn…
Figure 11
Figure 11. Figure 11: Sensitive analysis results using different dimensional reduction technique on real time-series ScRNA-seq datasets (A-D) Cross validation results with different dimensional reduction technique on swap mislabeling(E-H) Cross validation results with different dimensional…
Figure 12
Figure 12. Figure 12: Enhancement results of supervised pseudotime analysis on reclassification with two types of mislabeling and two types of metrics (A-C) Enhancement results using Kendall correlation of supervised pseudotime analysis on deletion with swap mislabeling (D-F) Enhancement r…
Figure 13
Figure 13. Figure 13: Enhancement results of supervised pseudotime analysis on deletion with two types of mislabeling and two types of metrics (A-C) Enhancement results using Kendall correlation of supervised pseudotime analysis with swap mislabeling (D-F) Enhancement results using Spearma…
Figure 14
Figure 14. Figure 14: (A) Dimensional reduction visualization using TSNE on GSE90047 (B) Potential Mislabeled cells(v=0) detected by ScPace in GSE90047 [PITH_FULL_IMAGE:figures/full_fig_p028_14.png]
Figure 15
Figure 15. Figure 15: (A) Confusion matrix results generated by the validation sets (B) PCA visualization of the cardiomyocyte collected at 6 embryonic days [PITH_FULL_IMAGE:figures/full_fig_p028_15.png]
Figure 16
Figure 16. Figure 16: Overall procedure of ScPace including data preprocessing, training ScPace and conduct two types of timestamp calibration Chen et al.: Preprint submitted to Elsevier Page 28 of 19 [PITH_FULL_IMAGE:figures/full_fig_p028_16.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 33 canonical work pages

  1. [1]

    Karlsson, T

    J. Karlsson, T. Kroneis, E. Jonasson, E. Larsson, A. Ståhlberg, Transcriptomic characterization of the human cell cycle in individual unsynchronizedcells,Journalofmolecularbiology429(24)(2017)3909–3924. doi:https://doi.org/10.1016/j.jmb.2017.10.011

    work page doi:10.1016/j.jmb.2017.10.011 2017

  2. [2]

    D. S. Fischer, A. K. Fiedler, E. M. Kernfeld, R. M. Genga, A. Bastidas-Ponce, M. Bakhti, H. Lickert, J. Hasenauer, R. Maehr, F. J. Theis, Inferring population dynamics from single-cell rna-sequencing time series data, Nature biotechnology 37 (4) (2019) 461–468.doi:https: //doi.org/10.1038/s41587-019-0088-0

    work page doi:10.1038/s41587-019-0088-0 2019

  3. [3]

    S. U. Khan, Y. Huang, H. Ali, I. Ali, S. Ahmad, S. U. Khan, T. Hussain, M. Ullah, K. Lu, Single-cell rna sequencing (scrna-seq): Advances and challenges for cardiovascular diseases (cvds), Current Problems in Cardiology (2023) 102202doi:https://doi.org/10.1016/j. cpcardiol.2023.102202

    arXiv 2023

  4. [4]

    Pasquini, J

    G. Pasquini, J. E. R. Arias, P. Schäfer, V. Busskamp, Automated methods for cell type annotation on scrna-seq data, Computational and Structural Biotechnology Journal 19 (2021) 961–969.doi:https://doi.org/10.1016/j.csbj.2021.01.015

    work page doi:10.1016/j.csbj.2021.01.015 2021

  5. [5]

    T.Tian,J.Wan,Q.Song,Z.Wei,Clusteringsingle-cellrna-seqdatawithamodel-baseddeeplearningapproach,NatureMachineIntelligence 1 (4) (2019) 191–198.doi:https://doi.org/10.1038/s42256-019-0037-0

    work page doi:10.1038/s42256-019-0037-0 2019

  6. [6]

    F. Yang, W. Wang, F. Wang, Y. Fang, D. Tang, J. Huang, H. Lu, J. Yao, scbert as a large-scale pretrained deep language model for cell type annotation of single-cell rna-seq data, Nature Machine Intelligence 4 (10) (2022) 852–866.doi:https://doi.org/10.1038/ s42256-022-00534-z

    work page 2022

  7. [7]

    F. X. Galdos, S. Xu, W. R. Goodyer, L. Duan, Y. V. Huang, S. Lee, H. Zhu, C. Lee, N. Wei, D. Lee, et al., devcellpy is a machine learning- enabled pipeline for automated annotation of complex multilayered single-cell transcriptomic data, Nature Communications 13 (1) (2022)

    work page 2022

  8. [8]

    G. Li, Y. Yang, E. Van Buren, Y. Li, Dropout imputation and batch effect correction for single-cell rna sequencing data, Journal of Bio-X Research 2 (04) (2019) 169–177.doi:https://doi.org/10.1097/JBR.0000000000000053

    work page doi:10.1097/jbr.0000000000000053 2019

Show all 47 references
  1. [9]

    Yang, W.-H

    L. Yang, W.-H. Wang, W.-L. Qiu, Z. Guo, E. Bi, C.-R. Xu, A single-cell transcriptomic analysis reveals precise pathways and regulatory mechanisms underlying hepatoblast differentiation, Hepatology 66 (5) (2017) 1387–1401.doi:https://doi.org/10.1002/hep.29353. Chen et al.: Prep...

    2017 doi

  2. [10]

    T.Kim,K.Lo,T.A.Geddes,H.J.Kim,J.Y.H.Yang,P.Yang,screclassify:posthoccelltypeclassificationofsingle-cellrna-seqdata,BMC genomics 20 (2019) 1–10.doi:https://doi.org/10.1186/s12864-019-6305-x

    2019 doi

  3. [11]

    doi:https: //doi.org/10.1038/s41592-021-01171-x

    P.V.Kharchenko,Thetriumphsandlimitationsofcomputationalmethodsforscrna-seq,Naturemethods18(7)(2021)723–732. doi:https: //doi.org/10.1038/s41592-021-01171-x

    2021 doi

  4. [12]

    M.Kumar,B.Packer,D.Koller,Self-pacedlearningforlatentvariablemodels,Advancesinneuralinformationprocessingsystems23(2010)

    2010

  5. [13]

    W. Xu, W. Liu, X. Huang, J. Yang, S. Qiu, Multi-modal self-paced learning for image classification, Neurocomputing 309 (2018) 134–144. doi:https://doi.org/10.1016/j.neucom.2018.04.075

    2018 doi

  6. [14]

    T. N. Ngo, L. V. Ngo, T. H. Nguyen, Unsupervised domain adaptation for text classification via meta self-paced learning, in: Proceedings of the 29th International Conference on Computational Linguistics (COLING), 2022

    2022

  7. [15]

    Q.Wang,Y.Zhou,W.Zhang,Z.Tang,X.Chen,Adaptivesamplingusingself-pacedlearningforimbalancedcancerdatapre-diagnosis,Expert Systems with Applications 152 (2020) 113334.doi:https://doi.org/10.1016/j.eswa.2020.113334

    2020

  8. [16]

    W.Li,L.Qiao,L.Zhang,Z.Wang,D.Shen,Functionalbrainnetworkestimationwithtimeseriesself-scrubbing,IEEEjournalofbiomedical and health informatics 23 (6) (2019) 2494–2504.doi:https://doi.org/10.1109/JBHI.2019.2893880

    2019

  9. [17]

    S.Zhang,X.Li,J.Lin,Q.Lin,K.-C.Wong,Reviewofsingle-cellrna-seqdataclusteringforcell-typeidentificationandcharacterization,Rna 29 (5) (2023) 517–530.doi:https://doi.org/10.1261/rna.078965.121

    2023 doi

  10. [18]

    doi:https://doi.org/10.1186/s13059-017-1188-0

    P.Lin,M.Troup,J.W.Ho,Cidr:Ultrafastandaccurateclusteringthroughimputationforsingle-cellrna-seqdata,Genomebiology18(2017) 1–11. doi:https://doi.org/10.1186/s13059-017-1188-0

    2017 doi

  11. [19]

    L. Yang, J. Liu, Q. Lu, A. D. Riggs, X. Wu, Saic: an iterative clustering approach for analysis of single cell rna-seq data, BMC genomics 18 (2017) 9–17. doi:https://doi.org/10.1186/s12864-017-4019-5

    2017 doi

  12. [20]

    Jiang, H

    L. Jiang, H. Chen, L. Pinello, G.-C. Yuan, Giniclust: detecting rare cell types from single-cell gene expression data with gini index, Genome biology 17 (2016) 1–13.doi:https://doi.org/10.1186/s13059-016-1010-4

    2016 doi

  13. [21]

    Alquicira-Hernandez, A

    J. Alquicira-Hernandez, A. Sathe, H. P. Ji, Q. Nguyen, J. E. Powell, scpred: accurate supervised method for cell-type classification from single-cell rna-seq data, Genome biology 20 (2019) 1–17.doi:https://doi.org/10.1186/s13059-019-1862-5

    2019 doi

  14. [22]

    doi:https://doi.org/10.1016/j.cels.2019.06.004

    Y.Tan,P.Cahan,Singlecellnet:acomputationaltooltoclassifysinglecellrna-seqdataacrossplatformsandacrossspecies,Cellsystems9(2) (2019) 207–213. doi:https://doi.org/10.1016/j.cels.2019.06.004

    2019 doi

  15. [23]

    F. Ma, M. Pellegrini, Actinn: automated identification of cell types in single cell rna sequencing, Bioinformatics 36 (2) (2020) 533–538. doi:https://doi.org/10.1093/bioinformatics/btz592

    2020 doi

  16. [24]

    X.Qiu,Q.Mao,Y.Tang,L.Wang,R.Chawla,H.A.Pliner,C.Trapnell,Reversedgraphembeddingresolvescomplexsingle-celltrajectories, Nature methods 14 (10) (2017) 979–982.doi:https://doi.org/10.1038/nmeth.4402

    2017 doi

  17. [25]

    Satija, J

    R. Satija, J. A. Farrell, D. Gennert, A. F. Schier, A. Regev, Spatial reconstruction of single-cell gene expression data, Nature biotechnology 33 (5) (2015) 495–502.doi:https://doi.org/10.1038/nbt.3192

    2015 doi

  18. [26]

    T. N. Tran, G. D. Bader, Tempora: cell trajectory inference using time-series single-cell rna sequencing data, PLoS computational biology 16 (9) (2020) e1008205.doi:https://doi.org/10.1371/journal.pcbi.1008205

    2020 doi

  19. [27]

    Y.Hong,H.Li,C.Long,P.Liang,J.Zhou,Y.Zuo,Anincrementofdiversitymethodforcellstatetrajectoryinferenceoftime-seriesscrna-seq data, Fundamental Research (2024).doi:https://doi.org/10.1016/j.fmre.2024.01.020

    2024 doi

  20. [28]

    P. K. Mondal, U. S. Saha, I. Mukhopadhyay, Pseudoga: cell pseudotime reconstruction based on genetic algorithm, Nucleic Acids Research 49 (14) (2021) 7909–7924.doi:https://doi.org/10.1093/nar/gkab457

    2021 doi

  21. [29]

    doi:https://doi.org/10.1093/nar/gkw430

    Z.Ji,H.Ji,Tscan:Pseudo-timereconstructionandevaluationinsingle-cellrna-seqanalysis,Nucleicacidsresearch44(13)(2016)e117–e117. doi:https://doi.org/10.1093/nar/gkw430

    2016 doi

  22. [30]

    Macnair, R

    W. Macnair, R. Gupta, M. Claassen, psupertime: supervised pseudotime analysis for time-series single-cell rna-seq data, Bioinformatics 38 (Supplement_1) (2022) i290–i298.doi:https://doi.org/10.1093/bioinformatics/btac227

    2022 doi

  23. [31]

    P. Yang, J. T. Ormerod, W. Liu, C. Ma, A. Y. Zomaya, J. Y. Yang, Adasampling for positive-unlabeled and label noise learning with bioinformatics applications, IEEE transactions on cybernetics 49 (5) (2018) 1932–1943.doi:https://doi.org/10.1109/TCYB.2018. 2816984

    2018 doi

  24. [32]

    Schölkopf, A

    B. Schölkopf, A. Smola, K.-R. Müller, Nonlinear component analysis as a kernel eigenvalue problem, Neural computation 10 (5) (1998) 1299–1319. doi:https://doi.org/10.1162/089976698300017467

    1998 doi

  25. [33]

    E. E. Osuna, Support vector machines: Training and applications, Ph.D. thesis, Massachusetts Institute of Technology (1998)

    1998

  26. [34]

    K.Crammer,Y.Singer,Onthealgorithmicimplementationofmulticlasskernel-basedvectormachines,Journalofmachinelearningresearch 2 (Dec) (2001) 265–292

    2001

  27. [35]

    Zappia, B

    L. Zappia, B. Phipson, A. Oshlack, Splatter: simulation of single-cell rna sequencing data, Genome biology 18 (1) (2017) 174.doi:https: //doi.org/10.1186/s13059-017-1305-0

    2017 doi

  28. [36]

    doi:https://doi.org/10.1038/s41587-019-0071-9

    W.Saelens,R.Cannoodt,H.Todorov,Y.Saeys,Acomparisonofsingle-celltrajectoryinferencemethods,Naturebiotechnology37(5)(2019) 547–554. doi:https://doi.org/10.1038/s41587-019-0071-9

    2019 doi

  29. [37]

    Hayashi, H

    T. Hayashi, H. Ozaki, Y. Sasagawa, M. Umeda, H. Danno, I. Nikaido, Single-cell full-length total rna sequencing uncovers dynamics of recursivesplicingandenhancerrnas,Naturecommunications9(1)(2018)619. doi:https://doi.org/10.1038/s41467-018-02866-0

    2018 doi

  30. [38]

    doi:https: //doi.org/10.1186/s13059-017-1382-0

    F.A.Wolf,P.Angerer,F.J.Theis,Scanpy:large-scalesingle-cellgeneexpressiondataanalysis,Genomebiology19(2018)1–5. doi:https: //doi.org/10.1186/s13059-017-1382-0

    2018 doi

  31. [39]

    Treutlein, Q

    B. Treutlein, Q. Y. Lee, J. G. Camp, M. Mall, W. Koh, S. A. M. Shariati, S. Sim, N. F. Neff, J. M. Skotheim, M. Wernig, et al., Dissecting direct reprogramming from fibroblast to neuron using single-cell rna-seq, Nature 534 (7607) (2016) 391–395.doi:https://doi.org/10. 1038/na...

    2016

  32. [40]

    Northcutt, L

    C. Northcutt, L. Jiang, I. Chuang, Confident learning: Estimating uncertainty in dataset labels, Journal of Artificial Intelligence Research 70 (2021) 1373–1411. doi:https://doi.org/10.1613/jair.1.12125

    2021 doi

  33. [41]

    Opitz, S

    J. Opitz, S. Burst, Macro f1 and macro f1, arXiv preprint arXiv:1911.03347 (2019). Chen et al.: Preprint submitted to Elsevier Page 17 of 19

    2019 arXiv

  34. [42]

    Frénay, M

    B. Frénay, M. Verleysen, Classification in the presence of label noise: a survey, IEEE transactions on neural networks and learning systems 25 (5) (2013) 845–869.doi:https://doi.org/10.1109/TNNLS.2013.2292894

    2013

  35. [43]

    T. Y. de Soysa, S. S. Ranade, S. Okawa, S. Ravichandran, Y. Huang, H. T. Salunga, A. Schricker, A. Del Sol, C. A. Gifford, D. Srivastava, Single-cell analysis of cardiogenesis reveals basis for organ-level developmental defects, Nature 572 (7767) (2019) 120–124.doi:https: //do...

    2019 doi

  36. [44]

    M. C. Hill, Z. A. Kadow, L. Li, T. T. Tran, J. D. Wythe, J. F. Martin, A cellular atlas of pitx2-dependent cardiac development, Development 146 (12) (2019) dev180398.doi:https://doi.org/10.1242/dev.180398

    2019 doi

  37. [45]

    W. R. Goodyer, B. M. Beyersdorf, D. T. Paik, L. Tian, G. Li, J. W. Buikema, O. Chirikian, S. Choi, S. Venkatraman, E. L. Adams, et al., Transcriptomic profiling of the developing cardiac conduction system at single-cell resolution, Circulation research 125 (4) (2019) 379–397. ...

    2019 doi

  38. [46]

    J.Feng,Y.Li,Y.Li,Q.Yin,H.Li,J.Li,B.Zhou,J.Meng,H.Lian,M.Wu,etal.,Versicanpromotescardiomyocyteproliferationandcardiac repair, Circulation 149 (13) (2024) 1004–1015.doi:https://doi.org/10.1161/CIRCULATIONAHA.123.066298. Chen et al.: Preprint submitted to Elsevier Page 18 of 19 ...

    2024 doi

  39. [5271]

    doi:https://doi.org/10.1038/s41467-022-33045-x

    doi

Pith tools

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