Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

LLM agents autonomously run trajectory inference at expert quality, 41% faster

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

An LLM multi-agent framework (SpaCellAgent) automates end-to-end trajectory inference on single-cell and spatial transcriptomics data, achieving expert-aligned accuracy with 41.2% faster analysis time.

T0 review reviewed 2026-07-09 challenge →

load-bearing objection Useful LLM-agent automation for trajectory inference, but the SOTA claim is confounded by an asymmetric benchmark setup — baselines run untuned while the agent optimizes parameters. the 3 major comments →

arxiv 2607.07467 v1 pith:PPVD5BOQ submitted 2026-07-08 cs.AI

SpaCellAgent: A Self-Evolving LLM-Based Multi-Agent Framework for Trajectory Analysis

Songhan Wang , Haoang Chi , He Li , Zhiheng Zhang , Jiayan Yuan , Cheems Wang , Hao Peng , Xinwang Liu
show 1 more author
Wenjing Yang
This is my paper
classification cs.AI
keywords trajectory inferencemulti-agent systemslarge language modelsspatial transcriptomicssingle-cell RNA sequencingself-refinementcomputational biologypseudotime
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

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 proposes SpaCellAgent, a multi-agent system built on large language models that autonomously performs end-to-end trajectory inference (TI) analysis on single-cell and spatial transcriptomics data. The framework assigns distinct roles — a planner that decomposes the biological question into analysis steps, an executor that selects appropriate computational tools and generates runnable code, an evaluator that checks both code correctness and biological plausibility, and a reporter that synthesizes findings into a narrative. Two mechanisms distinguish it from a static pipeline: self-refinement, where the evaluator's feedback drives iterative code correction within a single analysis, and self-evolution, where validated workflows and error-fix pairs are archived in a persistent memory and retrieved for future tasks. The authors claim that this architecture achieves state-of-the-art trajectory reconstruction quality across three benchmark categories (REAL-GOLD, REAL-SILVER, SYNTHETIC) while reducing total analysis time by 41.2% compared to a panel of five domain experts, from 64.6 minutes to 38.0 minutes per analysis. The central object carrying the argument is the biological evaluator agent — an LLM that judges whether inferred trajectories are biologically coherent (e.g., flagging a terminally differentiated cell incorrectly placed at the trajectory root) and routes corrective feedback back to the planner and executor.

Core claim

The paper's central claim is that a coordinated team of LLM agents, each handling a distinct phase of trajectory inference analysis, can match or exceed specialized TI algorithms on standard benchmarks while operating autonomously from natural-language input to biological report. The improvement over individual TI methods (DPT, Slingshot, PAGA, etc.) is attributed to the system's ability to dynamically select and configure algorithms per dataset rather than relying on fixed algorithmic assumptions, and to iteratively self-correct both code-level and biology-level errors through the evaluator-driven refinement loop. The self-evolution mechanism further accelerates convergence on new datasets:

What carries the argument

Multi-agent LLM architecture with four specialized roles (planner, executor, evaluator, reporter), coupled with a dual-layer memory (local intra-task and global inter-task) that enables self-refinement within a task and self-evolution across tasks.

Load-bearing premise

The paper assumes that the LLM-based biological evaluator — which checks whether inferred trajectories are biologically plausible (e.g., whether the root cell is a progenitor, not a terminally differentiated cell) — provides reliable biological validation. This is load-bearing because the entire self-correction loop depends on the evaluator correctly diagnosing biological errors. The paper validates downstream trajectory metrics against ground truth but does not directly test

What would settle it

Benchmark the biological evaluator agent's diagnostic accuracy against expert-curated biological assessments across diverse tissue contexts. If the evaluator's false-positive or false-negative rate for biological implausibility is high in novel contexts, the self-refinement loop would be shown to be unreliable.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If the self-evolution mechanism genuinely accumulates reusable analysis templates, the system's marginal cost per new dataset should decrease over time, potentially making large-scale atlas-level trajectory analysis tractable without proportional growth in human expertise.
  • The dynamic tool discovery mechanism — where the agent can import and register new packages on the fly — suggests a path toward frameworks that stay current with the rapidly expanding bioinformatics tool ecosystem without manual registry updates.
  • The biological evaluator's PubMed-augmented fallback for unfamiliar tissue contexts raises the possibility of agent-driven literature synthesis embedded directly into the analysis loop, though its reliability as a ground-truth proxy is unbenchmarked.
  • If the 41.2% time reduction holds across labs, it could lower the barrier for non-bioinformatician biologists to run expert-quality trajectory analysis, shifting the role of human experts toward validation and hypothesis generation rather than pipeline construction.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The system's quality advantage over individual TI algorithms may partly reflect ensemble behavior — by trying multiple algorithms and selecting based on the biological evaluator's judgment, the agent effectively performs a form of model selection that no single baseline can match. If so, the fair comparison might be against a human executing the same multi-algorithm selection strategy, not against
  • The biological evaluator's reliability is the linchpin of the entire self-correction loop. If the LLM systematically misjudges trajectory plausibility in novel biological contexts — particularly tissues or organisms underrepresented in its training data — the refinement loop could converge on confidently wrong results. A direct benchmark of the evaluator's diagnostic accuracy against expert-curate
  • The cross-model sensitivity analysis (DeepSeek-V3, GPT-5.2, Claude Sonnet 4) showing consistent performance suggests the gains are driven by the agentic architecture rather than a specific LLM's capabilities, but also implies the system inherits whatever biological knowledge gaps its underlying LLM carries.
  • The 41.2% efficiency claim is measured against five experts on a standardized six-task workflow. Real-world TI analysis often involves exploratory iteration and dead ends that are not captured by a fixed workflow, so the time savings in practice may differ from the controlled benchmark.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The manuscript introduces SpaCellAgent, an LLM-driven multi-agent framework for automating end-to-end trajectory inference (TI) analysis on single-cell and spatial transcriptomics data. The system features a planner-executor-evaluator architecture with a self-refinement loop and a cross-task self-evolution mechanism (dual-layer memory). The evaluator agent performs both code validation and biological plausibility checks. The framework is evaluated on synthetic and real-world benchmarks (Saelens et al.) against standard TI baselines (DPT, Slingshot, PAGA, etc.) and in case studies on mouse midbrain and spinal cord injury datasets. The authors report state-of-the-art TI performance and a 41.2% reduction in analysis time compared to human experts.

Significance. The paper addresses a practical bottleneck in computational biology: the fragmentation of TI tools and the need for manual expertise. The multi-agent architecture with a biological evaluator and self-evolution mechanism is a reasonable and novel application of LLM agents to this domain. The provision of verbatim system prompts (Appendix A.2) and a publicly available code repository are notable strengths that aid reproducibility. The evaluation uses established external benchmarks (Saelens et al. [31]), ensuring that the trajectory metrics are not circularly defined.

major comments (3)
  1. §5.1 (Baselines): The benchmark comparison in Table 1 is asymmetric. SpaCellAgent benefits from dynamic parameter optimization via its executor and self-refinement loop, while all baselines (DPT, Slingshot, PAGA, etc.) run with default parameters ('For all baselines, we adopt the default parameter settings as provided in their original implementations'). Since SpaCellAgent's tool registry (Table 5) includes PAGA, Slingshot, and Monocle3, the framework effectively runs the same underlying algorithms but with optimized parameters. This makes it difficult to attribute the performance gap to the agent's biological reasoning rather than to parameter tuning. An ablation isolating the agent's reasoning from parameter optimization (e.g., comparing SpaCellAgent against a baseline with grid-search-tuned parameters) is needed to support the SOTA claim.
  2. §4.2 (Biological evaluator) and §4.3 (Self-refinement): The biological evaluator uses LLM reasoning to flag biologically implausible trajectories, and this assessment underpins the self-correction loop. However, the paper does not benchmark the evaluator's diagnostic accuracy against ground-truth biological assessments. If the LLM-based evaluator systematically misjudges trajectory plausibility, the self-refinement loop could converge on biologically incorrect results that pass the evaluator's checks. A quantitative evaluation of the evaluator's precision/recall on known biological errors would strengthen this load-bearing component.
  3. §5.3 and Table 2 (Efficiency analysis): The 41.2% efficiency improvement claim is based on a comparison with only 5 human experts. The standard deviations reported (e.g., Step 1: 10.0±1.6 min for humans) suggest moderate variability, but with n=5, the statistical power to establish a robust baseline is limited. The manuscript should acknowledge this limitation or provide a sensitivity analysis to show that the 41.2% figure is stable and not driven by outlier expert performances.
minor comments (6)
  1. Table 1: The HIM score for SpaCellAgent on the Real-Silver dataset is 0.322, which is lower than all baselines except RaceID/StemID. This anomaly should be discussed, as it contradicts the 'ranking first in the majority of metrics' claim.
  2. §5.1 (Model configuration): The temperature for the main agents is set to 0.0, but for self-reflection it is 0.4. The rationale for 0.4 specifically (rather than 0.3 or 0.5) is not justified. A sensitivity analysis over this parameter would be helpful.
  3. Figure 7: The axis labels and legend are rendered as unicode escape sequences (e.g., /uni00000036/uni00000057...) and are unreadable. This figure needs to be regenerated with proper text rendering.
  4. Table 3: The dataset identifiers 'S1' through 'S5' are used but not defined in the main text. A brief description of these subsets should be provided for context.
  5. Appendix B.7: The text references a 'six-task' workflow (Table 11) but then mentions a 'five-step' plan (Table 12). The inconsistency between 5 and 6 tasks/steps should be clarified.
  6. §5.4 (Case Study): The Mouse SCI dataset is described as 'unpublished' and provided by a collaborating hospital. A brief note on ethical approval or data availability for this dataset would be appropriate, even if it is only available upon request.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for a careful and constructive review. The three major comments are well-taken. We address each below and describe revisions we will make.

read point-by-point responses

  1. Referee: §5.1 (Baselines): Asymmetric comparison — SpaCellAgent benefits from dynamic parameter optimization while baselines run with default parameters. An ablation isolating agent reasoning from parameter optimization is needed.

    Authors: The referee raises a valid concern. We agree that the current comparison in Table 1 does not fully disentangle the contribution of the agent's biological reasoning from the contribution of adaptive parameter tuning. This is a genuine gap in our evaluation. In the revised manuscript, we will add an ablation experiment in which each baseline algorithm (PAGA, Slingshot, PAGA Tree) is run with grid-search-optimized parameters on the same benchmark datasets, and compared against (a) the same algorithm invoked by SpaCellAgent's executor with agent-selected parameters, and (b) the full SpaCellAgent pipeline. This will allow us to isolate the marginal contribution of the agent's reasoning (planner decomposition, biological evaluator feedback, self-refinement) above and beyond parameter optimization. We note that our existing ablation (Figure 7, Table 13) already shows that removing the evaluator or self-evolution degrades task success rate and correlation, which provides partial evidence that the agent architecture contributes beyond mere parameter selection. However, we agree that a direct comparison against grid-search-tuned baselines is the cleaner test and will include it. We will also soften the SOTA claim in the abstract and main text to clarify that the improvement reflects the combined effect of adaptive parameter selection and agent-level reasoning, pending the results of this additional ablation. revision: yes

  2. Referee: §4.2 and §4.3: The biological evaluator's diagnostic accuracy is not benchmarked against ground-truth biological assessments. A quantitative evaluation of precision/recall on known biological errors would strengthen this component.

    Authors: We agree that the biological evaluator is a load-bearing component and that its accuracy should be quantitatively assessed rather than only demonstrated qualitatively through case studies. In the revised manuscript, we will construct a benchmark of trajectory outputs containing known, deliberately injected biological errors (e.g., inverted pseudotime with terminally differentiated cells assigned as root, fragmented manifolds where continuity is expected, biologically impossible lineage transitions such as neurons differentiating into glia without a progenitor intermediate). We will then measure the evaluator's precision and recall in flagging these errors versus correctly passing valid trajectories. We will report these metrics in a new table in Section 5 or the appendix. We acknowledge that constructing a comprehensive ground-truth set of biological errors is non-trivial and inherently incomplete — there exist classes of subtle biological errors that no LLM-based evaluator may reliably catch. We will explicitly state this limitation in the revised text, noting that the evaluator serves as a necessary but not sufficient safeguard, and that final biological validation by domain experts remains essential for novel discoveries. revision: yes

  3. Referee: §5.3 and Table 2: The 41.2% efficiency improvement is based on only 5 human experts. The manuscript should acknowledge this limitation or provide a sensitivity analysis.

    Authors: The referee is correct that n=5 is a small sample and limits the statistical robustness of the 41.2% figure. We will address this in two ways in the revision. First, we will add an explicit limitation statement in Section 5.3 acknowledging that the human baseline is based on five experts and that the generalizability of the efficiency claim is therefore constrained. Second, we will provide a sensitivity analysis: we will report the efficiency improvement under leave-one-out exclusion of each expert (i.e., recomputing the baseline mean with each expert removed in turn) to demonstrate that the 41.2% figure is not driven by a single slow or fast outlier. We will present these results in a supplementary table. If any leave-one-out configuration substantially changes the improvement percentage, we will report the range transparently rather than a single point estimate. We will also revise the abstract and conclusion to qualify the claim as 'over 40% improvement in analytical efficiency in our expert benchmark (n=5)' rather than presenting it as a universally established figure. revision: yes

Circularity Check

0 steps flagged

No significant circularity: the derivation chain is self-contained against external benchmarks, with only minor non-load-bearing self-citations.

full rationale

The paper's central quantitative claim (SOTA trajectory inference performance in Table 1) is evaluated against external ground-truth benchmarks from Saelens et al. [31], an independent evaluation framework. The metrics (Corr, F1, HIM, wCor) are computed by comparing predicted trajectories to reference trajectories, not by re-using the system's own outputs. The self-refinement and self-evolution mechanisms (§4.3) archive successful workflows and error-fix pairs, but these do not define the evaluation metrics or the ground truth. The biological evaluator (§4.2) uses LLM reasoning to flag implausible trajectories, but its outputs feed into the refinement loop, not into the final metric computation. The ablation studies (§6) remove components and measure Task Success Rate against the same external benchmarks. While the paper cites several prior works by its own authors (e.g., [7, 8, 9, 10, 20, 44]) in the Additional Related Works section, these citations concern robust learning, causal reasoning, and few-shot adaptation — topics peripheral to the central trajectory inference pipeline. None of these self-citations are invoked as load-bearing premises (e.g., uniqueness theorems or ansätze) that would make the central result reduce to its own inputs by construction. The asymmetric benchmark comparison (SpaCellAgent with dynamic parameter optimization vs. baselines with default parameters) is a methodological concern about experimental fairness, not a circularity issue — the evaluation metrics remain independent of the system's internal mechanisms. No step in the derivation chain exhibits self-definitional, fitted-input, or self-citation load-bearing circularity.

Axiom & Free-Parameter Ledger

6 free parameters · 4 axioms · 3 invented entities

The framework introduces several architectural components (dual-layer memory, knowledge-augmented fallback, dynamic tool discovery) but only the dual-layer memory has independent ablation evidence. The biological evaluator's reliability is a critical unvalidated assumption. The comparison against baselines with default parameters while SpaCellAgent uses dynamic optimization creates a potential fairness confound.

free parameters (6)
  • Temperature (main agents) = 0.0
    Set deterministically for main agent outputs; not fitted to data but chosen for reproducibility.
  • Temperature (self-reflection) = 0.4
    Higher temperature for biological hypothesis exploration during refinement; chosen empirically.
  • Top-p = 1.0
    Standard sampling parameter; not data-fitted.
  • Max output tokens = 128000
    Set to accommodate long-sequence reasoning; not fitted.
  • Random seed = 42
    Fixed for reproducibility per §B.1.
  • Max refinement attempts = Not specified
    The self-refinement loop has a 'maximum attempt limit' (§4.3) but the value is not stated, making it an unstated free parameter affecting success rates.
axioms (4)
  • domain assumption DeepSeek-V3 (and GPT-5.2, claude-sonnet-4-6) can reliably generate biologically valid trajectory inference code from natural language instructions.
    The entire framework depends on the LLM's ability to generate correct Python/R code for TI tools. This is assumed, not proven independently of the benchmark results.
  • domain assumption LLM-based biological evaluation can reliably detect biologically implausible trajectories (e.g., inverted pseudotime, impossible lineage transitions).
    §4.2 Biological evaluator assumes the LLM can cross-reference findings with established biological priors. The accuracy of this evaluation is never benchmarked against expert biological assessment.
  • domain assumption The six benchmark datasets are representative of the diversity of trajectory inference tasks encountered in practice.
    Several benchmark subsets are very small (65-85 cells per Table 7), raising questions about generalization to larger, more complex datasets.
  • domain assumption Default parameters are a fair baseline for comparison.
    §5.1 states 'For all baselines, we adopt the default parameter settings as provided in their original implementations.' SpaCellAgent benefits from dynamic parameter selection while baselines do not, which may not reflect their optimal performance.
invented entities (3)
  • Dual-layer memory architecture (local + global) independent evidence
    purpose: Stores intra-task execution context and inter-task validated workflows for RAG-based retrieval.
    The ablation study (Table 13, w/o experience memory variant) provides falsifiable evidence: removing global memory reduces correlation from 0.538 to 0.526 and increases time from 38.0 to 46.0 minutes.
  • Knowledge-augmented fallback mechanism no independent evidence
    purpose: Searches PubMed for biological references when the agent encounters unfamiliar trajectory inference tasks.
    Described in §4.2 but no ablation or quantitative evaluation isolates this mechanism's contribution. The ablation in §B.8 tests 'w/o experience memory' but does not specifically test the PubMed fallback.
  • Dynamic tool discovery and registration mechanism no independent evidence
    purpose: Autonomously imports and registers new Python/R packages when the existing tool registry is insufficient.
    Described in §4.2 but no experiment demonstrates this mechanism in action. No example of a dynamically discovered tool is provided.

reviewed 2026-07-09 · how reviews work

0 comments
Cite this review

Pith. "Pith review of SpaCellAgent: A Self-Evolving LLM-Based Multi-Agent Framework for Trajectory Analysis." pith.science (2026). https://pith.science/paper/PPVD5BOQ

@misc{pith2026260707467,
  author       = {Pith},
  title        = {Pith review of: SpaCellAgent: A Self-Evolving LLM-Based Multi-Agent Framework for Trajectory Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PPVD5BOQ}},
  note         = {Machine review of arXiv:2607.07467}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Spatial and Single-cell transcriptomics are transformative in deciphering cellular dynamics. As the fundamental paradigm for reconstructing cell developmental paths, trajectory inference (TI) is critical. However, existing methods require extensive manual intervention and proficiency in heterogeneous tools, posing a significant barrier to efficient TI analysis. To bridge this gap, we propose SpaCellAgent, an autonomous large language model (LLM) multi-agent framework that automates end-to-end spatiotemporal analysis and narrative generation. SpaCellAgent utilizes a multi-agent architecture for strategic workflow planning, a dynamic tool-orchestration engine for adaptive algorithm selection, and a self-evolution module that iteratively refines performance through feedback. We evaluate SpaCellAgent on six heterogeneous datasets encompassing complex temporal developmental trajectories, diverse sequencing platforms, and spatially-resolved tissue architectures. SpaCellAgent consistently demonstrates over 40\% improvement in analytical efficiency while maintaining expert-aligned performance. By converting natural language specifications into optimized analytical workflows and fully automating the pipeline, SpaCellAgent democratizes advanced spatiotemporal modeling and establishes a scalable, agent-driven paradigm for computational biology. The code and materials are available at https://github.com/LittleXH-shw/SpaCellAgent.

Figures

Figures reproduced from arXiv: 2607.07467 by Cheems Wang, Haoang Chi, Hao Peng, He Li, Jiayan Yuan, Songhan Wang, Wenjing Yang, Xinwang Liu, Zhiheng Zhang.

Figure 1
Figure 1. Figure 1: Comparison between SpaCellAgent and the traditional workflow. Our framework utilizes an LLM-driven multi-agent [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The framework of SpaCellAgent. The planner decomposes user queries into executable steps for the executor, while [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: UMAP visualization and spatial distributions of cell types in the mouse embryonic dorsal midbrain at developmental [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: UMAP projection of cell type clusters identified by SpaCellAgent and comparison of inferred developmental trajectories [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Sankey diagram illustrating the temporal lineage [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: The results of operational effectiveness analysis. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: The visualization pseudotime results of TI analysis by SpaCellAgent compared with ground truth on REAL-GOLD [PITH_FULL_IMAGE:figures/full_fig_p017_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: The visualization pseudotime results of TI analysis by SpaCellAgent compared with ground truth on REAL-GOLD [PITH_FULL_IMAGE:figures/full_fig_p019_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: The visualization pseudotime results of TI analysis by SpaCellAgent compared with ground truth on REAL-GOLD [PITH_FULL_IMAGE:figures/full_fig_p020_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: The visualization pseudotime results of TI analysis by SpaCellAgent compared with ground truth on REAL-GOLD [PITH_FULL_IMAGE:figures/full_fig_p021_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: The visualization pseudotime results of TI analysis by SpaCellAgent compared with ground truth on REAL-GOLD [PITH_FULL_IMAGE:figures/full_fig_p022_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Spatial visualization of clusters, cell annotation by SpaCellAgent and ground truth annotation. [PITH_FULL_IMAGE:figures/full_fig_p023_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Visualization of spatial topology and neighborhood connectivity from Axolotl Neuron dataset. The left panel displays [PITH_FULL_IMAGE:figures/full_fig_p023_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: DEG heat map of Axolotl Neuron dataset. The visualization displays the top 3 marker genes identified for each [PITH_FULL_IMAGE:figures/full_fig_p024_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: (A) Pseudotime of HS and NS mouse SCI dataset. (B) HS datasets’ comparison of TI Umap plot by PAGA and Pseudotime [PITH_FULL_IMAGE:figures/full_fig_p025_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Global DEG heat map of mouse SCI dataset compared with time point and NS/HS. [PITH_FULL_IMAGE:figures/full_fig_p026_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: Gene Ontology (GO) enrichment analysis of differentially expressed genes. (A) Enriched biological processes for [PITH_FULL_IMAGE:figures/full_fig_p027_18.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Evaluating Agentic Bioinformatics through Function, Evidence, and Validation

    cs.AI 2026-07 conditional novelty 5.0

    Agentic bioinformatics systems mostly demonstrate planning and tool execution but rarely prospective empirical validation, so the paper argues evaluation should center on inspectable workflow trajectories (FEV) rather...

Reference graph

Works this paper leans on

48 extracted references · 48 canonical work pages · cited by 1 Pith paper · 4 internal anchors

  1. [1]

    Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Floren- cia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. 2023. Gpt-4 technical report.arXiv preprint arXiv:2303.08774 (2023). KDD 2026, August 9–13, 2026, Jeju Island, Republic of Korea. Songhan Wang et al

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  2. [2]

    Rhonda Bacher and Christina Kendziorski. 2016. Design and computational analysis of single-cell RNA-sequencing experiments.Genome biology17, 1 (2016), 63

    work page 2016

  3. [3]

    Daniil A Boiko, Robert MacKnight, Ben Kline, and Gabe Gomes. 2023. Au- tonomous chemical research with large language models.Nature624, 7992 (2023), 570–578

    work page 2023

  4. [4]

    Robrecht Cannoodt, Wouter Saelens, and Yvan Saeys. 2016. Computational methods for trajectory inference from single-cell transcriptomics.European journal of immunology46, 11 (2016), 2496–2506

    work page 2016

  5. [5]

    Robrecht Cannoodt, Wouter Saelens, Dorine Sichien, Simon Tavernier, Sophie Janssens, Martin Guilliams, Bart Lambrecht, Katleen De Preter, and Yvan Saeys

    work page

  6. [6]

    SCORPIUS improves trajectory inference and identifies novel modules in dendritic cell development.biorxiv(2016), 079509

    work page 2016

  7. [7]

    Ao Chen, Sha Liao, Mengnan Cheng, Kailong Ma, Liang Wu, Yiwei Lai, Xiaojie Qiu, Jin Yang, Jiangshan Xu, Shijie Hao, et al. 2022. Spatiotemporal transcriptomic atlas of mouse organogenesis using DNA nanoball-patterned arrays.Cell185, 10 (2022), 1777–1792

    work page 2022

  8. [8]

    Haoang Chi, He Li, Wenjing Yang, Feng Liu, Long Lan, Xiaoguang Ren, Tongliang Liu, and Bo Han. 2024. Unveiling causal reasoning in large language models: Reality or mirage?Advances in Neural Information Processing Systems37 (2024), 96640–96670

    work page 2024

  9. [9]

    Haoang Chi, Feng Liu, Wenjing Yang, Long Lan, Tongliang Liu, Bo Han, William Cheung, and James Kwok. 2021. TOHAN: A one-step approach towards few-shot hypothesis adaptation.Advances in neural information processing systems34 (2021), 20970–20982

    work page 2021

  10. [10]

    Haoang Chi, Feng Liu, Wenjing Yang, Long Lan, Tongliang Liu, Bo Han, Gang Niu, Mingyuan Zhou, and Masashi Sugiyama. 2022. Meta Discovery: Learning to Discover Novel Classes given Very Limited Data. InInternational Conference on Learning Representations

    work page 2022

  11. [11]

    Haoang Chi, Wenjing Yang, Feng Liu, Long Lan, Tao Qin, and Bo Han. 2024. Does Confusion Really Hurt Novel Class Discovery?International Journal of Computer Vision132, 8 (2024), 3191–3207

    work page 2024

  12. [12]

    Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, et al. 2025. Gemini 2.5: Pushing the frontier with advanced reasoning, multi- modality, long context, and next generation agentic capabilities.arXiv preprint arXiv:2507.06261(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  13. [13]

    Ling Dai, Bin Sheng, Tingli Chen, Qiang Wu, Ruhan Liu, Chun Cai, Liang Wu, Dawei Yang, Haslina Hamzah, Yuexing Liu, et al. 2024. A deep learning system for predicting time to progression of diabetic retinopathy.Nature Medicine30, 2 (2024), 584–594

    work page 2024

  14. [14]

    Dominic Grün, Mauro J Muraro, Jean-Charles Boisset, Kay Wiebrands, Anna Lyubimova, Gitanjali Dharmadhikari, Maaike van den Born, Johan Van Es, Erik Jansen, Hans Clevers, et al. 2016. De novo prediction of stem cell identity using single-cell transcriptome data.Cell stem cell19, 2 (2016), 266–277

    work page 2016

  15. [15]

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, et al . 2025. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning.arXiv preprint arXiv:2501.12948(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  16. [16]

    Laleh Haghverdi, Maren Büttner, F Alexander Wolf, Florian Buettner, and Fabian J Theis. 2016. Diffusion pseudotime robustly reconstructs lineage branching.Na- ture methods13, 10 (2016), 845–848

    work page 2016

  17. [17]

    Jian Hu, Xiangjie Li, Kyle Coleman, Amelia Schroeder, Nan Ma, David J Irwin, Edward B Lee, Russell T Shinohara, and Mingyao Li. 2021. SpaGCN: Integrating gene expression, spatial location and histology to identify spatial domains and spatially variable genes by graph convolutional network.Nature methods18, 11 (2021), 1342–1351

    work page 2021

  18. [18]

    Geert-Jan Huizing, Jules Samaran, Daniele Capocefalo, Anna Audit, Gabriel Peyré, and Laura Cantini. 2025. STORIES: learning cell fate landscapes from spatial transcriptomics using optimal transport.Nature Methods(2025), 1–10

    work page 2025

  19. [19]

    Nigel Kee, Nikolaos Volakakis, Agnete Kirkeby, Lina Dahl, Helena Storvall, Sara Nolbrant, Laura Lahti, Åsa K Björklund, Linda Gillberg, Eliza Joodmardi, et al

    work page

  20. [20]

    Single-cell analysis reveals a close relationship between differentiating dopamine and subthalamic nucleus neuronal lineages.Cell stem cell20, 1 (2017), 29–40

    work page 2017

  21. [21]

    Markus Kleinewietfeld, Arndt Manzel, Jens Titze, Heda Kvakan, Nir Yosef, Ralf A Linker, Dominik N Muller, and David A Hafler. 2013. Sodium chloride drives autoimmune disease by the induction of pathogenic TH17 cells.Nature496, 7446 (2013), 518–522

    work page 2013

  22. [22]

    He Li, Haoang Chi, Mingyu Liu, Wanrong Huang, Liyang Xu, and Wenjing Yang

    work page

  23. [23]

    InForty-second International Conference on Machine Learning

    Transformer-Based Spatial-Temporal Counterfactual Outcomes Estimation. InForty-second International Conference on Machine Learning

    work page

  24. [24]

    Bran, Sam Cox, Oliver Schilter, Carlo Baldassari, Andrew D White, and Philippe Schwaller

    Andres M. Bran, Sam Cox, Oliver Schilter, Carlo Baldassari, Andrew D White, and Philippe Schwaller. 2024. Augmenting large language models with chemistry tools.Nature Machine Intelligence6, 5 (2024), 525–535

    work page 2024

  25. [25]

    Evan Z Macosko, Anindita Basu, Rahul Satija, James Nemesh, Karthik Shekhar, Melissa Goldman, Itay Tirosh, Allison R Bialas, Nolan Kamitaki, Emily M Marter- steck, et al. 2015. Highly parallel genome-wide expression profiling of individual cells using nanoliter droplets.Cell161, 5 (2015), 1202–1214

    work page 2015

  26. [26]

    Sachin Mathur, Mathieu Beauvais, Arnau Giribet, Nicolas Aragon Barrero, Chaorui-Tom Zhang, Towsif Rahman, Seqian Wang, Jeremy Huang, Nima Nouri, Andre Kurlovs, et al. 2025. PyEvoCell: an LLM-augmented single-cell trajectory analysis dashboard.Bioinformatics41, 4 (2025), btaf158

    work page 2025

  27. [27]

    Jeffrey R Moffitt, Dhananjay Bambah-Mukku, Stephen W Eichhorn, Eric Vaughn, Karthik Shekhar, Julio D Perez, Nimrod D Rubinstein, Junjie Hao, Aviv Regev, Catherine Dulac, et al. 2018. Molecular, spatial, and functional single-cell profiling of the hypothalamic preoptic region.Science362, 6416 (2018), eaau5324

    work page 2018

  28. [28]

    Efthymia Papalexi and Rahul Satija. 2018. Single-cell RNA sequencing to explore immune cell heterogeneity.Nature Reviews Immunology18, 1 (2018), 35–45

    work page 2018

  29. [29]

    Chen Qian, Wei Liu, Hongzhang Liu, Nuo Chen, Yufan Dang, Jiahao Li, Cheng Yang, Weize Chen, Yusheng Su, Xin Cong, et al. 2024. Chatdev: Communicative agents for software development. InProceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 15174–15186

    work page 2024

  30. [30]

    Xiaojie Qiu, Qi Mao, Ying Tang, Li Wang, Raghav Chawla, Hannah A Pliner, and Cole Trapnell. 2017. Reversed graph embedding resolves complex single-cell trajectories.Nature methods14, 10 (2017), 979–982

    work page 2017

  31. [31]

    Anjali Rao, Dalia Barkley, Gustavo S França, and Itai Yanai. 2021. Exploring tissue architecture using spatial transcriptomics.Nature596, 7871 (2021), 211–220

    work page 2021

  32. [32]

    Honglei Ren, Benjamin L Walker, Zixuan Cang, and Qing Nie. 2022. Identify- ing multicellular spatiotemporal organization of cells with SpaceFlow.Nature communications13, 1 (2022), 4076

    work page 2022

  33. [33]

    Samuel G Rodriques, Robert R Stickels, Aleksandrina Goeva, Carly A Martin, Evan Murray, Charles R Vanderburg, Joshua Welch, Linlin M Chen, Fei Chen, and Evan Z Macosko. 2019. Slide-seq: A scalable technology for measuring genome-wide expression at high spatial resolution.Science363, 6434 (2019), 1463–1467

    work page 2019

  34. [34]

    Wouter Saelens, Robrecht Cannoodt, Helena Todorov, and Yvan Saeys. 2019. A comparison of single-cell trajectory inference methods.Nature biotechnology37, 5 (2019), 547–554

    work page 2019

  35. [35]

    Nicholas Schaum, Jim Karkanias, Norma F Neff, Andrew P May, Stephen R Quake, Tony Wyss-Coray, Spyros Darmanis, Joshua Batson, Olga Botvinnik, Michelle B Chen, et al. 2018. Single-cell transcriptomics of 20 mouse organs creates a Tabula Muris: The Tabula Muris Consortium.Nature562, 7727 (2018), 367

    work page 2018

  36. [36]

    Antonio Scialdone, Yosuke Tanaka, Wajid Jawaid, Victoria Moignard, Nicola K Wilson, Iain C Macaulay, John C Marioni, and Berthold Göttgens. 2016. Resolving early mesoderm diversification through single-cell expression profiling.Nature 535, 7611 (2016), 289–293

    work page 2016

  37. [37]

    Alex K Shalek, Rahul Satija, Xian Adiconis, Rona S Gertner, Jellert T Gaublomme, Raktima Raychowdhury, Schraga Schwartz, Nir Yosef, Christine Malboeuf, Diana Lu, et al. 2013. Single-cell transcriptomics reveals bimodality in expression and splicing in immune cells.Nature498, 7453 (2013), 236–240

    work page 2013

  38. [38]

    Kelly Street, Davide Risso, Russell B Fletcher, Diya Das, John Ngai, Nir Yosef, Eliz- abeth Purdom, and Sandrine Dudoit. 2018. Slingshot: cell lineage and pseudotime inference for single-cell transcriptomics.BMC genomics19, 1 (2018), 477

    work page 2018

  39. [39]

    Amos Tanay and Aviv Regev. 2017. Scaling single-cell genomics from phe- nomenology to mechanism.Nature541, 7637 (2017), 331–338

    work page 2017

  40. [40]

    Fuchou Tang, Catalin Barbacioru, Yangzhou Wang, Ellen Nordman, Clarence Lee, Nanlan Xu, Xiaohui Wang, John Bodeau, Brian B Tuch, Asim Siddiqui, et al. 2009. mRNA-Seq whole-transcriptome analysis of a single cell.Nature methods6, 5 (2009), 377–382

    work page 2009

  41. [41]

    Cole Trapnell, Davide Cacchiarelli, Jonna Grimsby, Prapti Pokharel, Shuqiang Li, Michael Morse, Niall J Lennon, Kenneth J Livak, Tarjei S Mikkelsen, and John L Rinn. 2014. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells.Nature biotechnology32, 4 (2014), 381–386

    work page 2014

  42. [42]

    Daniel E Wagner and Allon M Klein. 2020. Lineage tracing meets single-cell omics: opportunities and challenges.Nature Reviews Genetics21, 7 (2020), 410–427

    work page 2020

  43. [43]

    Hanchen Wang, Tianfan Fu, Yuanqi Du, Wenhao Gao, Kexin Huang, Ziming Liu, Payal Chandak, Shengchao Liu, Peter Van Katwyk, Andreea Deac, et al . 2023. Scientific discovery in the age of artificial intelligence.Nature620, 7972 (2023), 47–60

    work page 2023

  44. [44]

    F Alexander Wolf, Fiona K Hamey, Mireya Plass, Jordi Solana, Joakim S Dahlin, Berthold Göttgens, Nikolaus Rajewsky, Lukas Simon, and Fabian J Theis. 2019. PAGA: graph abstraction reconciles clustering with trajectory inference through a topology preserving map of single cells.Genome biology20, 1 (2019), 59

    work page 2019

  45. [45]

    Chuan Wu, Nir Yosef, Theresa Thalhamer, Chen Zhu, Sheng Xiao, Yasuhiro Kishi, Aviv Regev, and Vijay K Kuchroo. 2013. Induction of pathogenic TH17 cells by inducible salt-sensing kinase SGK1.Nature496, 7446 (2013), 513–517

    work page 2013

  46. [46]

    Yihang Xiao, Jinyi Liu, Yan Zheng, Xiaohan Xie, Jianye Hao, Mingzhi Li, Ruitao Wang, Fei Ni, Yuxiao Li, Jintian Luo, et al . 2024. Cellagent: An llm-driven multi-agent framework for automated single-cell data analysis.arXiv preprint arXiv:2407.09811(2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  47. [47]

    User Task Description

    Wenjing Yang, Haoang Chi, Yibing Zhan, Bowen Hu, Xiaoguang Ren, Dapeng Tao, and Long Lan. 2025. NT-FAN: A Simple yet Effective Noise-tolerant Few-shot Adaptation Network.Artificial Intelligence(2025), 104363. SpaCellAgent: A Self-Evolving LLM-Based Multi-Agent Framework for Trajectory Analysis KDD 2026, August 9–13, 2026, Jeju Island, Republic of Korea. P...

    work page 2025

  48. [48]

    KDD 2026, August 9–13, 2026, Jeju Island, Republic of Korea

    conducted comprehensive evaluations to unveil the underlying causal reasoning capabilities of large language models. KDD 2026, August 9–13, 2026, Jeju Island, Republic of Korea. Songhan Wang et al. Subset 2: cellbench-SC4_luyitian Subset 2: cell-cycle_leng Figure 10: The visualization pseudotime results of TI analysis by SpaCellAgent compared with ground ...

    work page 2026

This paper was first reviewed by glm-5.2 on July 9, 2026.