Pith. sign in

REVIEW 4 major objections 5 minor 36 references

GeneSUM: Large Language Model-based Gene Summary Extraction

T0 review · 4 major / 5 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 This paper claims that GeneSUM, a two-stage LLM pipeline using Gene Ontology anchors, substantially outperforms six existing gene-summarization baselines on ROUGE metrics, generating summaries closer to expert-written Entrez Gene entries.

desk verdict Useful pipeline, but the headline results are not trustworthy because the evaluation set appears to overlap with the fine-tuning set. read the letter →

arxiv 2412.18154 v1 pith:ZTEQAXRF submitted 2024-12-24 q-bio.GN cs.AIcs.CL

Zhijian Chen , Chuan Hu , Min Wu , Qingqing Long , Xuezhi Wang , Yuanchun Zhou , Meng Xiao This is my paper · ORCID
classification q-bio.GNcs.AIcs.CL
keywords genesummarylargelanguagemodelOntologytextsummarizationLoRAfine-tuningbiomedicalliteratureROUGE
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 tries to establish that gene summaries, which are normally written by expert curators and stored in databases like Entrez Gene, can be generated automatically and at scale by a two-stage large-language-model pipeline. Its central claim is that the proposed system, GeneSUM, produces summaries much closer to expert-written references than six existing extractive and generative baselines, as measured by ROUGE-1, ROUGE-2, and ROUGE-L. The authors argue that the key to this improvement is using Gene Ontology annotations as semantic anchors to select the most relevant sentences from the literature, and then fine-tuning a general LLM on those selected sentences to write the final summary. If the claim is right, automated gene summarization could keep biological knowledge bases current with the expanding literature and give researchers fast, traceable overviews of gene function.

What carries the argument

The load-bearing component is the GO-anchored sentence selection step. Gene Ontology terms for the target gene (molecular function, biological process, cellular component) are expanded into natural-language sentences by an LLM; both these descriptions and the filtered literature sentences are vectorized with BioBERT, clustered with K-means (k chosen by Calinski-Harabasz score), and within each cluster the sentence with maximum cosine similarity to a GO description is kept as a key sentence. These key sentences are then injected through a prompt into Gemma-7B fine-tuned with LoRA, which generates the final summary. Signature-term filtering (Pearson chi-square test against background literature) is the first-stage reducer that trims the candidate sentence pool.

What would settle it

Re-train GeneSUM on one half of the 8,887 genes and evaluate on the other half, or audit the training set for overlap with the evaluation genes; if ROUGE-2 falls from 0.1856 toward the best baseline's 0.0467, the central claim fails.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that a gene summary can be reconstructed by combining two cheap signals: the gene's structured Gene Ontology annotations and a filtered set of literature sentences. GO terms are rewritten by an LLM into short descriptive sentences, used to locate and select key sentences from the literature by cosine similarity in a BioBERT embedding space, and the chosen sentences are passed to a LoRA-fine-tuned Gemma-7B model to generate the final summary. The reported result is that this pipeline reaches ROUGE-1 0.3874, ROUGE-2 0.1856, and ROUGE-L 0.3681 against expert-written Entrez Gene summaries, with the GO-based selection step alone accounting for most of the gain over filtering without GO.

Load-bearing premise

The paper assumes, without describing a train/test split, that the genes used to fine-tune the generator are not the same genes whose Entrez Gene summaries serve as ROUGE references; if they overlap, the headline scores could reflect memorization rather than summarization ability.

Editorial extensions

If this is right

  • If the reported ROUGE gains hold, gene summary generation can be automated from raw PubMed literature and kept current as new papers appear, without waiting for expert curation.
  • The method produces summaries whose sentences can be traced back to specific PubMed Central articles, addressing hallucination concerns for this task.
  • The GO-anchored sentence selection finding suggests that structured ontology knowledge can materially improve LLM summarization in specialized domains.
  • Fine-tuning a relatively small 7B model with LoRA appears sufficient to outperform much larger general-purpose LLMs on this task.

Reading between the lines

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

  • A critical test the paper does not report is a train/test split by gene; if the fine-tuning set overlaps the evaluation set, the ROUGE advantage could shrink or vanish, so the method's generalization to unseen genes remains unverified.
  • The same GO-anchored selection could be applied to other ontology-based summarization targets, such as disease or drug summaries, since the mechanism is not gene-specific.
  • The GO-rewrite step could likely be replaced by direct embeddings of GO terms, avoiding reliance on an external LLM for the expansion.
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

4 major / 5 minor

Summary. The paper proposes GeneSUM, a two-stage pipeline for automatic gene-summary generation. In the first stage, gene-related literature is retrieved and filtered using chi-square-based signature terms, Gene Ontology annotations are rewritten into natural-language descriptions via a prompted LLM, and a BioBERT-vectorized K-means clustering procedure with cosine similarity selects key sentences. In the second stage, Gemma-7B is fine-tuned with LoRA on prompts built from the selected key sentences to generate final summaries. Experiments on 8,887 genes are reported against six extractive and LLM baselines using ROUGE-1, ROUGE-2, and ROUGE-L, with additional ablations, vectorization-model comparisons, and two case studies (COL4A2 and TLR9). The central claim is that GeneSUM significantly outperforms all baselines across all three ROUGE metrics and better integrates gene-specific information.

Significance. If the empirical claim were validated, GeneSUM would be a practically useful contribution to biomedical text summarization: the idea of rewriting GO annotations into full sentences and using them as semantic anchors for key-sentence selection is sensible, and the combination of literature filtering, GO-based selection, clustering, and LoRA fine-tuning is a coherent pipeline. The paper also addresses a real bottleneck in database curation, where most genes lack structured summaries. However, the paper ships no code, data, or trained models, and the evaluation methodology as written does not establish generalization: no held-out test split is described, and the fine-tuning target appears to coincide with the ROUGE reference. The reported gains, especially the fourfold ROUGE-2 improvement over the best baseline, are therefore not currently interpretable as evidence of the method's effectiveness. The underlying idea remains worth testing under a proper evaluation protocol.

major comments (4)
  1. [§II-D and §III-A] The paper never describes a held-out test split. Section II-D states that key sentences are extracted "from our training dataset" and used for LoRA fine-tuning, while Section III-A says that 8,887 genes with existing gene function description information were selected "for experimentation." If the Entrez Gene summaries used as ROUGE references are the same summaries that appear in the fine-tuning data, the model can memorize the exact bigram sequences of the reference text, which would explain the large ROUGE-2 gap in Table I (0.1856 vs. the next-best 0.0467). The authors must specify a gene-level train/validation/test split, ensure that the signature-term extraction in Section II-A uses only the training portion, and report all metrics separately on genes never seen during fine-tuning.
  2. [§III-B] The claim that "our model significantly outperforms six baselines across three ROUGE metrics" is not supported by any measure of variance or hypothesis test. Table I reports single point estimates, and no confidence intervals, standard errors, or paired tests are given. Even after a proper held-out split is introduced, the authors should report multiple runs or bootstrap over genes and use a paired significance test (e.g., Wilcoxon signed-rank or approximate randomization) to justify the word "significantly."
  3. [§III-A] The baselines are not sufficiently specified to ensure a controlled comparison. It is unclear whether the Random baseline uses the same LoRA-fine-tuned model or the base model, and the prompts given to Llama2-70B, ChatGPT-3.5, BioMistral, and Llama3-OpenBioLLM are not provided. If the baselines receive different candidate-sentence sets, different context lengths, or different generation instructions, the ROUGE differences in Table I may reflect input or prompt variation rather than the proposed method.
  4. [Figures 6 and 7] The case studies demonstrate near-verbatim reuse of expert-written reference sentences, for example "This gene encodes one of the six subunits of type IV collagen, the major structural component of basement membranes" for COL4A2 and "TLRs are highly conserved from Drosophila to humans and share structural and functional similarities" for TLR9. The authors present this as traceability, but it is also the expected signature of training-target leakage. These case studies should be repeated on held-out genes after the split is introduced, and the authors should report both ROUGE and a measure of factual overlap with the training summaries.
minor comments (5)
  1. [§II-A, Eq. (1)] The alternative hypothesis and the subsequent sentence are misstated: the null hypothesis is that term t_i is not a signature term, so rejection of the null implies that t_i is a signature term; the text as written says the opposite.
  2. [§II-C] There are typos in the K-means description: "fist step" should be "first step" and "closet center" should be "closest center."
  3. [Figures 4 and 5] Figures 4 and 5 are referenced in the text with captions, but the figures themselves are not present in the submitted manuscript. Please include the actual plots.
  4. [§III-A] Reference [24] is cited for both Llama2-70B and ChatGPT-3.5, but [24] is a paper about GPT-3 and does not support those model names. Baseline 6 (Llama3-OpenBioLLM-8B) is paired with reference [26], which is the BioMistral paper; the citation is mismatched.
  5. [§II-C] The statement that "many genes have around 10 GO annotations on average" lacks a citation or supporting statistic, and the choice of the K-means range k=3..10 is justified only informally.

Circularity Check

1 steps flagged · score 7.0 of 10

Headline ROUGE gains may reduce to memorization: no held-out gene split is described, and the fine-tuning targets are the same Entrez 'summary attribute' used as ROUGE references.

  1. fitted input called prediction [Section II-A, II-D, III-A]
    "We use the Pearson’s chi-square test [11] to extract topic signature terms from reference summaries in the training set ... The process begins with the extraction of key sentences for each gene from our training dataset using a key sentence extraction module ... We utilized gene function description information from the summary attribute of the database as a reference from NCBI sub-database Entrez Gene. ... we selected 8,887 genes that had existing gene function description information for experimentation."

    The LoRA fine-tuning targets are gene summaries derived from the Entrez Gene 'summary attribute', and Section II-A explicitly places 'reference summaries' in the training set. Section II-D then extracts key sentences from the same 'training dataset' for fine-tuning. Section III-A uses the very same Entrez 'summary attribute' as the ROUGE reference and reports results on the 8,887 genes 'for experimentation' with no described train/test/validation split, held-out gene list, or cross-validation. If the evaluation genes overlap with the fine-tuning genes, the reported ROUGE scores--especially ROUGE-2 0.1856 vs. the best baseline 0.0467--measure reproduction of training references rather than generalization.

full rationale

The central empirical claim is that GeneSUM 'significantly outperforms six baselines across three ROUGE metrics.' The only load-bearing circularity is the apparent identity between the fine-tuning data and the evaluation references. Section II-A states that signature terms are extracted from 'reference summaries in the training set'; Section II-D states that key sentences are extracted from 'our training dataset' for LoRA fine-tuning; Section III-A defines the ROUGE reference as the Entrez Gene 'summary attribute' and describes experiments on all 8,887 genes without any partition. As written, the model is fine-tuned on the same summaries against which it is scored, so the reported gains (particularly ROUGE-2, a bigram overlap metric) are consistent with memorization rather than independent prediction. The case studies of COL4A2 and TLR9 show near-verbatim reuse of expert summary sentences, which is the expected leakage signature. No code or data is released, so the reader cannot verify a hidden split. Other self-citations in the paper are not load-bearing for the main claim, and the GO-rewriting and clustering components retain independent technical content, so the circularity is partial but serious, affecting the headline performance claim.

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

The paper adds an engineering pipeline rather than a derivation; the listed free parameters are pipeline choices (thresholds, k, LoRA rank), and the axioms are domain assumptions about the gold standard and embedding semantics. No invented entities are introduced.

free parameters (4)
  • chi-square signature threshold = 10.83
    Terms with a chi-square statistic above 10.83 are kept as signature terms (Section III-A); this threshold encodes a significance level of 0.001 and is a hand-chosen pipeline parameter.
  • minimum signature terms per sentence = 3
    Sentences containing fewer than three signature terms are filtered (Section II-A); the number 3 is chosen by hand and its effect is not analyzed.
  • k-means cluster count k = selected by CH score in [3,10]
    The number of clusters is selected per gene by the Calinski-Harabasz score over a range from 3 to 10 (Section II-C); this is a data-dependent free parameter.
  • LoRA rank r = 32
    The low-rank adaptation rank is set to 32 for Gemma-7B (Section II-D); no ablation on rank is reported.
assumptions (5)
  • domain assumption The Entrez Gene summary attribute is a valid ground truth for gene summaries.
    Section III-A uses this attribute as the ROUGE reference; if it is incomplete, biased, or non-canonical, the evaluation metric is unreliable.
  • domain assumption ROUGE scores measure the quality of gene summaries.
    ROUGE measures n-gram overlap with a reference; the paper uses no human evaluation or factual-consistency check, so semantic quality and biological correctness are assumed to track ROUGE.
  • domain assumption BioBERT embedding cosine similarity captures semantic relevance between literature sentences and GO term descriptions.
    The key-sentence selection in Eq. (5) assumes that nearest neighbors in BioBERT space are thematically aligned with GO descriptions; this is a standard but unverified assumption.
  • domain assumption ChatGPT-generated GO term descriptions are faithful expansions of the GO annotations.
    Prompt#1 (Figure 2) delegates GO descriptions to ChatGPT with no verification; errors in these descriptions propagate into sentence selection.
  • domain assumption The signature terms extracted from training reference summaries generalize to unseen genes.
    Section II-A derives signature terms from the training reference summaries; if the training genes are not representative, the filtering step will remove relevant sentences for other genes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GeneSUM: Large Language Model-based Gene Summary Extraction." pith.science (2026). https://pith.science/paper/ZTEQAXRF

@misc{pith2026241218154,
  author       = {Pith},
  title        = {Pith review of: GeneSUM: Large Language Model-based Gene Summary Extraction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZTEQAXRF}},
  note         = {Machine review of arXiv:2412.18154}
}
read the original abstract

Emerging topics in biomedical research are continuously expanding, providing a wealth of information about genes and their function. This rapid proliferation of knowledge presents unprecedented opportunities for scientific discovery and formidable challenges for researchers striving to keep abreast of the latest advancements. One significant challenge is navigating the vast corpus of literature to extract vital gene-related information, a time-consuming and cumbersome task. To enhance the efficiency of this process, it is crucial to address several key challenges: (1) the overwhelming volume of literature, (2) the complexity of gene functions, and (3) the automated integration and generation. In response, we propose GeneSUM, a two-stage automated gene summary extractor utilizing a large language model (LLM). Our approach retrieves and eliminates redundancy of target gene literature and then fine-tunes the LLM to refine and streamline the summarization process. We conducted extensive experiments to validate the efficacy of our proposed framework. The results demonstrate that LLM significantly enhances the integration of gene-specific information, allowing more efficient decision-making in ongoing research.

Figures

Figures reproduced from arXiv: 2412.18154 by the authors.

Figure 1
Figure 1. An overview of our framework. II. LLM-BASED GENE SUMMARY GENERATION FRAMEWORK A. Literature Retrieving and Filtering. The Importance of Signature Filtering. Various kinds of texts often possess unique signature terms [9], [10]. For instance, the frequent occurrence of words like sweat, com￾petition, and racing in a corpus may suggest that the topic is related to sports or competitions. Likewise, gene knowledge summa… view at source ↗
Figure 2
Figure 2. Example template of Prompt#1 to instruct ChatGPT [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Example of Prompt#2 to help large language model [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (3 more)
Figure 5
Figure 5. Figure 5: The influence of different generative model. [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 4
Figure 4. Figure 4: The influence of different Vectorized models. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 6
Figure 6. Figure 6: Comparison of Generated Gene Summary for [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 21 canonical work pages

  1. [1]

    The ncbi taxonomy database,

    S. Federhen, “The ncbi taxonomy database,” Nucleic acids research , vol. 40, no. D1, pp. D136–D143, 2012

    work page 2012

  2. [2]

    Ncbi taxonomy: a comprehensive update on curation, resources and tools,

    C. L. Schoch, S. Ciufo, M. Domrachev, C. L. Hotton, S. Kannan, R. Khovanskaya, D. Leipe, R. Mcveigh, K. O’Neill, B. Robbertse et al., “Ncbi taxonomy: a comprehensive update on curation, resources and tools,” Database, vol. 2020, p. baaa062, 2020

    work page 2020

  3. [3]

    Learning to rank-based gene summary extraction,

    Y . Shang, H. Hao, J. Wu, and H. Lin, “Learning to rank-based gene summary extraction,” BMC bioinformatics , vol. 15, pp. 1–8, 2014

    work page 2014

  4. [4]

    Fast abstractive summarization with reinforce-selected sentence rewriting,

    Y .-C. Chen and M. Bansal, “Fast abstractive summarization with reinforce-selected sentence rewriting,” arXiv preprint arXiv:1805.11080, 2018

    arXiv 2018

  5. [5]

    Bottom-up abstractive summarization,

    S. Gehrmann, Y . Deng, and A. M. Rush, “Bottom-up abstractive summarization,” arXiv preprint arXiv:1808.10792 , 2018

    arXiv 2018

  6. [6]

    Expert knowledge-guided length-variant hierarchical label generation for pro- posal classification,

    M. Xiao, Z. Qiao, Y . Fu, Y . Du, P. Wang, and Y . Zhou, “Expert knowledge-guided length-variant hierarchical label generation for pro- posal classification,” in 2021 IEEE International Conference on Data Mining (ICDM) . IEEE, 2021, pp. 757–766

    work page 2021

  7. [7]

    Hierarchical interdisciplinary topic detection model for research proposal classification,

    M. Xiao, Z. Qiao, Y . Fu, H. Dong, Y . Du, P. Wang, H. Xiong, and Y . Zhou, “Hierarchical interdisciplinary topic detection model for research proposal classification,” IEEE Transactions on Knowledge and Data Engineering , vol. 35, no. 9, pp. 9685–9699, 2023

    work page 2023

  8. [8]

    In- terdisciplinary fairness in imbalanced research proposal topic inference: A hierarchical transformer-based method with selective interpolation,

    M. Xiao, M. Wu, Z. Qiao, Y . Fu, Z. Ning, Y . Du, and Y . Zhou, “In- terdisciplinary fairness in imbalanced research proposal topic inference: A hierarchical transformer-based method with selective interpolation,” ACM Transactions on Knowledge Discovery from Data

    work page

Show all 36 references
  1. [9]

    The automated acquisition of topic signatures for text summarization,

    C.-Y . Lin and E. Hovy, “The automated acquisition of topic signatures for text summarization,” in COLING 2000 volume 1: The 18th interna- tional conference on computational linguistics , 2000

    2000

  2. [10]

    Towards automatic generation of gene summary,

    F. Jin, M. Huang, Z. Lu, and X. Zhu, “Towards automatic generation of gene summary,” in Proceedings of the BioNLP 2009 Workshop , 2009, pp. 97–105

    2009

  3. [11]

    Manning and H

    C. Manning and H. Schutze, F oundations of statistical natural language processing. MIT press, 1999

    1999

  4. [12]

    Measuring gene functional similarity based on group-wise comparison of go terms,

    Z. Teng, M. Guo, X. Liu, Q. Dai, C. Wang, and P. Xuan, “Measuring gene functional similarity based on group-wise comparison of go terms,” Bioinformatics, vol. 29, no. 11, pp. 1424–1432, 2013

    2013

  5. [13]

    Gpt-ner: Named entity recognition via large language models,

    S. Wang, X. Sun, X. Li, R. Ouyang, F. Wu, T. Zhang, J. Li, and G. Wang, “Gpt-ner: Named entity recognition via large language models,” arXiv preprint arXiv:2304.10428, 2023

    2023 arXiv

  6. [14]

    Biobert based named entity recognition in electronic medical record,

    X. Yu, W. Hu, S. Lu, X. Sun, and Z. Yuan, “Biobert based named entity recognition in electronic medical record,” in 2019 10th interna- tional conference on information technology in medicine and education (ITME). IEEE, 2019, pp. 49–52

    2019

  7. [15]

    Biobert: a pre-trained biomedical language representation model for biomedical text mining,

    J. Lee, W. Yoon, S. Kim, D. Kim, S. Kim, C. H. So, and J. Kang, “Biobert: a pre-trained biomedical language representation model for biomedical text mining,” Bioinformatics, vol. 36, no. 4, pp. 1234–1240, 2020

    2020

  8. [16]

    Extracting drug-drug interactions from texts with biobert and multiple entity-aware attentions,

    Y . Zhu, L. Li, H. Lu, A. Zhou, and X. Qin, “Extracting drug-drug interactions from texts with biobert and multiple entity-aware attentions,” Journal of biomedical informatics , vol. 106, p. 103451, 2020

    2020

  9. [17]

    K-means clustering: a half-century synthesis,

    D. Steinley, “K-means clustering: a half-century synthesis,” British Journal of Mathematical and Statistical Psychology , vol. 59, no. 1, pp. 1–34, 2006

    2006

  10. [18]

    Unsupervised k-means clustering algo- rithm,

    K. P. Sinaga and M.-S. Yang, “Unsupervised k-means clustering algo- rithm,” IEEE access , vol. 8, pp. 80 716–80 727, 2020

    2020

  11. [19]

    Semantic cosine similar- ity,

    F. Rahutomo, T. Kitasuka, M. Aritsugi et al., “Semantic cosine similar- ity,” in The 7th international student conference on advanced science and technology ICAST , vol. 4, no. 1. University of Seoul South Korea, 2012, p. 1

    2012

  12. [20]

    Learning similarity with cosine similarity ensemble,

    P. Xia, L. Zhang, and F. Li, “Learning similarity with cosine similarity ensemble,” Information sciences , vol. 307, pp. 39–52, 2015

    2015

  13. [21]

    Lora: Low-rank adaptation of large language models,

    E. J. Hu, Y . Shen, P. Wallis, Z. Allen-Zhu, Y . Li, S. Wang, L. Wang, and W. Chen, “Lora: Low-rank adaptation of large language models,” arXiv preprint arXiv:2106.09685 , 2021

    2021 arXiv

  14. [22]

    Per- formance analysis of lora finetuning llama-2,

    A. Pathak, O. Shree, M. Agarwal, S. D. Sarkar, and A. Tiwary, “Per- formance analysis of lora finetuning llama-2,” in 2023 7th International Conference on Electronics, Materials Engineering & Nano-Technology (IEMENTech). IEEE, 2023, pp. 1–4

    2023

  15. [23]

    Rouge: A package for automatic evaluation of summaries,

    C.-Y . Lin, “Rouge: A package for automatic evaluation of summaries,” in Text summarization branches out , 2004, pp. 74–81

    2004

  16. [24]

    Gpt-3: Its nature, scope, limits, and consequences,

    L. Floridi and M. Chiriatti, “Gpt-3: Its nature, scope, limits, and consequences,” Minds and Machines , vol. 30, pp. 681–694, 2020

    2020

  17. [25]

    Can- cerllm: A large language model in cancer domain,

    M. Li, A. Blaes, S. Johnson, H. Liu, H. Xu, and R. Zhang, “Can- cerllm: A large language model in cancer domain,” arXiv preprint arXiv:2406.10459, 2024

    2024 arXiv

  18. [26]

    Biomistral: A collection of open-source pretrained large language models for medical domains,

    Y . Labrak, A. Bazoge, E. Morin, P.-A. Gourraud, M. Rouvier, and R. Du- four, “Biomistral: A collection of open-source pretrained large language models for medical domains,” arXiv preprint arXiv:2402.10373 , 2024

    2024 arXiv

  19. [27]

    Bert: Pre-training of deep bidirectional transformers for language understanding,

    J. Devlin, “Bert: Pre-training of deep bidirectional transformers for language understanding,” arXiv preprint arXiv:1810.04805 , 2018

    2018 arXiv

  20. [28]

    Generating gene summaries from biomedical literature: A study of semi-structured summarization,

    X. Ling, J. Jiang, X. He, Q. Mei, C. Zhai, and B. Schatz, “Generating gene summaries from biomedical literature: A study of semi-structured summarization,” Information Processing & Management , vol. 43, no. 6, pp. 1777–1791, 2007

    2007

  21. [29]

    Bottlesum: Unsupervised and self-supervised sentence summarization using the information bot- tleneck principle,

    P. West, A. Holtzman, J. Buys, and Y . Choi, “Bottlesum: Unsupervised and self-supervised sentence summarization using the information bot- tleneck principle,” arXiv preprint arXiv:1909.07405 , 2019

    1909 arXiv

  22. [30]

    Needed: Introducing hierarchical transformer to eye diseases diagnosis,

    X. Ye, M. Xiao, Z. Ning, W. Dai, W. Cui, Y . Du, and Y . Zhou, “Needed: Introducing hierarchical transformer to eye diseases diagnosis,” in Pro- ceedings of the 2023 SIAM International Conference on Data Mining (SDM). SIAM, 2023, pp. 667–675

    2023

  23. [31]

    A continued pretrained llm approach for automatic medical note generation,

    D. Yuan, E. Rastogi, G. Naik, J. Chintagunta, S. P. Rajagopal, F. Zhao, S. Goyal, and J. Ward, “A continued pretrained llm approach for automatic medical note generation,” arXiv preprint arXiv:2403.09057 , 2024

    2024 arXiv

  24. [32]

    Biorag: A rag-llm framework for biological question reasoning,

    C. Wang, Q. Long, X. Meng, X. Cai, C. Wu, Z. Meng, X. Wang, and Y . Zhou, “Biorag: A rag-llm framework for biological question reasoning,” arXiv preprint arXiv:2408.01107 , 2024

    2024 arXiv

  25. [33]

    scinterpreter: Training large language models to interpret scrna-seq data for cell type annotation,

    C. Li, M. Xiao, P. Wang, G. Feng, X. Li, and Y . Zhou, “scinterpreter: Training large language models to interpret scrna-seq data for cell type annotation,” arXiv preprint arXiv:2402.12405 , 2024

    2024 arXiv

  26. [34]

    Resolving the imbalance issue in hierarchical disciplinary topic inference via llm-based data augmentation,

    X. Cai, M. Xiao, Z. Ning, and Y . Zhou, “Resolving the imbalance issue in hierarchical disciplinary topic inference via llm-based data augmentation,” in 2023 IEEE International Conference on Data Mining (ICDM). IEEE, 2023, pp. 956–961

    2023

  27. [35]

    What makes good in-context examples for gpt- 3?

    J. Liu, D. Shen, Y . Zhang, B. Dolan, L. Carin, and W. Chen, “What makes good in-context examples for gpt- 3?” arXiv preprint arXiv:2101.06804, 2021

    2021 arXiv

  28. [36]

    Enhanced gene selection in single-cell genomics: Pre-filtering synergy and reinforced optimization,

    W. Zhang, Z. Meng, D. Wang, M. Wu, K. Liu, Y . Zhou, and M. Xiao, “Enhanced gene selection in single-cell genomics: Pre-filtering synergy and reinforced optimization,” arXiv preprint arXiv:2406.07418 , 2024

    2024 arXiv

Pith tools

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