The DOI matches its Crossref record; 108 of the 109 references it cites were found in a registry or an index; 32 of the 99 sources its claims cite were checked for support.
The origins and genetic interactions of KRAS mutations are allele- and tissue-specific
Open accessAbstract
Mutational activation of KRAS promotes the initiation and progression of cancers, especially in the colorectum, pancreas, lung, and blood plasma, with varying prevalence of specific activating missense mutations. Although epidemiological studies connect specific alleles to clinical outcomes, the mechanisms underlying the distinct clinical characteristics of mutant KRAS alleles are unclear. Here, we analyze 13,492 samples from these four tumor types to examine allele- and tissue-specific genetic properties associated with oncogenic KRAS mutations. The prevalence of known mutagenic mechanisms partially explains the observed spectrum of KRAS activating mutations. However, there are substantial differences between the observed and predicted frequencies for many alleles, suggesting that biological selection underlies the tissue-specific frequencies of mutant alleles. Consistent with experimental studies that have identified distinct signaling properties associated with each mutant form of KRAS, our genetic analysis reveals that each KRAS allele is associated with a distinct tissue-specific comutation network. Moreover, we identify tissue-specific genetic dependencies associated with specific mutant KRAS alleles. Overall, this analysis demonstrates that the genetic interactions of oncogenic KRAS mutations are allele- and tissue-specific, underscoring the complexity that drives their clinical consequences.
Overview
CoChat Summary
TL;DR
This study analyzes 13,492 COAD, LUAD, MM, and PAAD samples to show that oncogenic KRAS allele frequencies are shaped partly by tissue-specific mutational processes but also by biological selection. It finds that each KRAS allele has distinct tissue-specific comutation networks and genetic dependencies, with some dependencies explained by comutating partner genes.
Key findings
- KRAS mutation prevalence varied by cancer: PAAD 86%, COAD 41%, LUAD 35%, and MM 22%. Across hotspots, codon 12 accounted for 72.2% of mutations, codon 13 for 9.8%, codon 61 for 14.8%, and codon 146 for 3.2%; incidence-adjusted distribution was 76.8%, 11.4%, 8.1%, and 3.7%, respectively.
- Mutational signatures only partially explained allele frequencies. Examples: COAD G13D was predicted at 27% but observed at 20%; MM Q61H was predicted at 15.0% but observed at 35.7%; PAAD G12R was expected at 5.2% but observed at 16.7%. Pearson correlations between observed and predicted frequencies were 0.4–0.6 overall and 0.7–0.9 when restricted to G12 alleles.
- Each KRAS allele had a distinct tissue-specific comutation network. In MM, NRAS had reduced comutation with KRAS G12D, Q61L, and Q61R, but 18.5% comutation with Q61H, just below the 23.6% NRAS mutation rate in KRAS WT tumors. In COAD, TP53 comutation was reduced with KRAS G12D and G13D.
- Allele-specific genetic dependencies were identified in cell lines: 62 genes in COAD and 130 genes in PAAD. Examples include stronger LIN7C dependency in KRAS G13D COAD lines, greater TFPT dependency in G12D lines, and reduced STARD9 dependency in G12D lines. Integrated models showed some dependencies could be attributed to comutation partners, such as STARD9 with TP53 and EEF1E1, ABI1, and MYBL2 with SMAD4.
Methods
- Public WGS, WES, and targeted-sequencing data were collected for COAD, LUAD, MM, and PAAD from cBioPortal, ICGC, MMRF-CoMMpass, and AACR GENIE v5, totaling 13,492 samples. Hypermutated COAD samples were identified with a Gaussian mixed model and excluded from comutation analysis.
- Mutational signatures were discovered with SigProfiler NMF and quantified by nonnegative least squares. Predicted KRAS allele frequencies were calculated from same-trinucleotide-context mutation frequencies genome-wide and compared with observed frequencies using χ-squared tests with Benjamini–Hochberg FDR. Comutation was tested with one-sided Fisher’s exact tests for increased comutation and the Leiserson row-column exclusivity test for reduced comutation.
- Genetic dependencies were analyzed using DepMap 2020Q1 CERES CRISPR scores, restricted to KRAS alleles present in at least three cell lines. Differential dependency was tested by ANOVA (p < 0.01) and t tests (FDR-adjusted p < 0.05), with GSEA and elastic net linear models incorporating comutation covariates and RNA expression.
Limitations
- The paper states that MM comutation analysis was hampered by frequent multiclonality, so it was limited to genes known to be recurrently mutated in MM to reduce false positives. It also notes that reduced-comutation analysis required WGS/WES data and could not use targeted-sequencing data.
- The paper states that hypermutated COAD samples were excluded from comutation analysis because passenger mutations would add noise, and that genetic dependency analysis was restricted to COAD and PAAD cell lines and to KRAS alleles with at least three cell lines; PAAD lacked enough WT KRAS cell lines for inclusion.
- Evident from the text: comutation and dependency analyses are observational and correlational, and the authors note that some allele-specific dependency signals could instead be attributed to comutating partner genes. The cell-line dependency models also had limited numbers of cell lines relative to covariates and used elastic net regularization.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures





Verification
Claims the paper makes
15 Valid · 0 Not valid · 28 Not checked · 125 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“SBS17, also of uncertain etiology though linked to oxidative stress in other cancers33, was likely the primary cause for Q61H mutations in PAAD”
§ The origins and genetic interactions of KRAS mutations are allele- and tissue-specific > Results > The KRAS alleles have different mutagenic origins
What the sources say
- Supportedcites [33] Tomkova, M., Tomek, J., Kriaucionis, S. & Schuster-Böckler, B. Mutational signature distribution varies with DNA replication timing and strand asymmetry. Genome Biol. 19, 129 (2018). source
The citation is attached to the description of SBS17 as having uncertain etiology but a proposed link to oxidative stress in other cancers, not to the citing authors’ conclusion about Q61H mutations in PAAD. In Results and discussion, under “Signatures linked to environmental mutagens” (page unavailable), the paper says the process causing signature 17 is unclear and discusses oxidative damage as a possible cause, particularly in esophageal adenocarcinoma. (LLM, conf=0.90)
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“forced expression of KRAS A146T in mouse pancreas does not induce pancreatic intraepithelial neoplasia4”
§ The origins and genetic interactions of KRAS mutations are allele- and tissue-specific > Results > The frequency of most KRAS alleles cannot be solely attributed to the prevalence of detected mutagens
What the sources say
- Supportedcites [4] Poulin, E. J. et al. Tissue-specific oncogenic activity of KRASA146T. Cancer Discov. 9, 738–755 (2019). source
The Results section (‘Tissue-specific phenotypes of K-Ras alleles’) reports that pancreases expressing K-RasA146T showed no histologic evidence of PanIN at eight weeks, and none was observed in the followed cohort, including four mice older than 300 days. This supports the claim for the mouse model studied. (LLM, conf=0.94)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“KRAS G12V had an increased rate of comutation with TCF7L2, which encodes TCF4, a regulator of Wnt signaling often dysregulated in COAD49–51, specifically the R488C mutation.”
§ The origins and genetic interactions of KRAS mutations are allele- and tissue-specific > Results > The KRAS alleles have distinct comutation networks
What the sources say
- Partly supportedcites [49] Cancer Genome Atlas Network. Comprehensive molecular characterization of human colon and rectal cancer. Nature 487, 330–337 (2012). sourceExperimental — accuracy not yet measured
The Results sections on gene mutations and altered pathways identify KRAS and TCF7L2 mutations in colorectal cancer and place TCF7L2 alterations in the WNT pathway. The Translocations section identifies TCF7L2 as encoding TCF4. Our copy does not report an increased co-mutation rate between KRAS G12V and TCF7L2 R488C; searches for both variant names found no such passage. Page numbers were not available in the indexed copy. (LLM, conf=0.85)
- Not checked yetcites [50] Roose, J. & Clevers, H. TCF transcription factors: molecular switches in carcinogenesis. Biochim. Biophys. Acta 1424, M23–M37 (1999).
- Could not verifycites [51] van de Wetering, M. et al. The beta-catenin/TCF-4 complex imposes a crypt progenitor phenotype on colorectal cancer cells. Cell 111, 241–250 (2002). sourceExperimental — accuracy not yet measured
The Introduction (p. 1) describes TCF proteins as Wnt effectors, and the Summary (p. 1) discusses Wnt-pathway mutations in colorectal cancer. Neither passage addresses KRAS G12V co-mutation with TCF7L2 or the R488C variant; a further search for those terms found no supporting passage in our copy. (LLM, conf=0.77)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on 2 assumptions not shown here.
- Partly supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“knocking out LIN7C, a gene that maintains the asymmetric distribution of membrane proteins in polarized epithelial cells67, had a more severe reduction on growth in KRAS G13D cell lines compared to the others”
§ The origins and genetic interactions of KRAS mutations are allele- and tissue-specific > Results > KRAS allele-specific genetic dependencies reveal potential synthetic lethal vulnerabilities
What the sources say
- Supportedcites [67] Monastyrskaya, K. et al. miR-199a-5p regulates urothelial permeability and may play a role in bladder pain syndrome. Am. J. Pathol. 182, 431–448 (2013). source
Citation 67 is attached to the description of LIN7C, not the subsequent claim about growth in KRAS G13D cell lines. The section on LIN7C expression in human bladder identifies LIN7C as important for maintaining epithelial membrane polarity, supporting that description. The copy does not address the KRAS growth comparison. (LLM, conf=0.90)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“the kinetochore-associated protein (KNTC1), a regulator of the mitotic checkpoint70,71, which demonstrated moderate to strong lethal effects when knocked out in almost every cell line except for those with a KRAS G12V allele”
§ The origins and genetic interactions of KRAS mutations are allele- and tissue-specific > Results > KRAS allele-specific genetic dependencies reveal potential synthetic lethal vulnerabilities
What the sources say
- Not checked yetcites [70] Chan, G. K. T., Jablonski, S. A., Starr, D. A., Goldberg, M. L. & Yen, T. J. Human Zw10 and ROD are mitotic checkpoint proteins that bind to kinetochores. Nat. Cell Biol. 2, 944–947 (2000). source
- Partly supportedcites [71] Kops, G. J. P. L. et al. ZW10 links mitotic checkpoint signaling to the structural kinetochore. J. Cell Biol. 169, 49–60 (2005). sourceExperimental — accuracy not yet measured
Our copy of Kops et al. (2005) supports only that loss of ZW10 (the human ortholog of KNTC1) is lethal/toxic: the Results section 'ZW10 recruits Mad1–Mad2 to unattached human kinetochores' states that ZW10-deficient cells 'underwent aberrant mitoses, which resulted in cell death after several divisions, as indicated by markedly diminished colony formation' (Fig. 6 e) plus chromatin bridges and micronuclei, and the Methods 'Colony outgrowth assay' frames this as measuring general toxicity. However, this was measured by siRNA/plasmid-based depletion in HeLa cells (and Xenopus CSF extracts, Fig. 4) — the paper reports no knockout screen across a panel of cancer cell lines, no 'moderate to strong' effect-size classification, and no mention of KRAS or KRAS G12V (searched: KRAS G12V cell line knockout lethal; ZW10 depletion viability across cell lines). The pan-cell-line lethality with a KRAS G12V exception is therefore not attributable to this source; only the general identity of KNTC1/ZW10 as a mitotic-checkpoint regulator whose depletion kills cells is supported. (LLM, conf=0.70)
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
110 more claims · 2 with their cited sources checked
Create a free account to see all 168 claims and to verify the 67 references still open.
- Probably not a substantial claim. 21
- It may state two separate claims as one. 34
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“G12D and G12V mutations are the most common in PDAC, followed by G12R.”
From KRAS: Biology, Inhibition, and Mechanisms of Inhibitor Resistance (2024) · cites this paper as [30]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“KRAS G12R, G13D, Q61H, Q61L, and A146T mutations have increased intrinsic exchange relative to wildtype, but G12C and G12V do not display large changes [4,29,30,31,32].”
From KRAS: Biology, Inhibition, and Mechanisms of Inhibitor Resistance (2024) · cites this paper as [30]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“KRAS A146 mutations are primarily found in colorectal cancer and are rarely reported in other types.”
From KRAS: Biology, Inhibition, and Mechanisms of Inhibitor Resistance (2024) · cites this paper as [30]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“G12C and G12V mutations are predominant in lung adenocarcinomas with smoking-associated mutational signatures [30,32].”
From KRAS: Biology, Inhibition, and Mechanisms of Inhibitor Resistance (2024) · cites this paper as [30]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“genetic studies have revealed that specific mutant KRAS alleles are associated with distinct tissue-specific genetic dependencies [46]”
From Feedback activation of EGFR/wild-type RAS signaling axis limits KRASG12D inhibitor efficacy in KRASG12D-mutated colorectal cancer (2023) · cites this paper as [46]
Experimental — accuracy not yet measured
2 more claims about this paper
Create a free account to see every claim other papers make about this one.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ The origins and genetic interactions of KRAS mutations are allele- and tissue-specific > Results > An integrated analysis of allele-specific comutation and genetic dependencies
“A similar effect was found between KRAS G12D and SMAD4 in PAAD cell lines for the dependency on EEF1E1, ABI1, and MYBL2 (Fig. 5b–d).”
§ The origins and genetic interactions of KRAS mutations are allele- and tissue-specific > Results > An integrated analysis of allele-specific comutation and genetic dependencies
“b–d The nonzero coefficients for the models of b EEF1E1, c ABL1, and d MYBL2 dependency in PAAD cell lines”
Citations
Cites
Subramanian, A., Tamayo, P., Mootha, V. K., Mukherjee, S., Ebert, B. L., Gillette, M. A., Paulovich, A., Pomeroy, S. L., Golub, T. R., Lander, E. S., et al. (2005) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.0506580102 · Cited by 53,625
Regularization and Variable Selection Via the Elastic Net
Zou, H., & Hastie, T. (2005) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.1467-9868.2005.00503.x · Cited by 20,957
Siegel, R. L., Miller, K. D., & Jemal, A. (2020) · CA: A Cancer Journal for Clinicians · doi:10.3322/caac.21590 · Cited by 20,444
Cerami, E., Gao, J., Dogrusoz, U., Gross, B. E., Sumer, S. O., Aksoy, B. A., Jacobsen, A., Byrne, C. J., Heuer, M. L., Larsson, E., et al. (2012) · Cancer Discovery · doi:10.1158/2159-8290.cd-12-0095 · Cited by 17,455
Integrative Analysis of Complex Cancer Genomics and Clinical Profiles Using the cBioPortal
Gao, J., Aksoy, B. A., Dogrusoz, U., Dresdner, G., Gross, B., Sumer, S. O., Sun, Y., Jacobsen, A., Sinha, R., Larsson, E., et al. (2013) · Science Signaling · doi:10.1126/scisignal.2004088 · Cited by 15,160
A method and server for predicting damaging missense mutations.
Adzhubei, I. A., Schmidt, S., Peshkin, L., Ramensky, V. E., Gerasimova, A., Bork, P., Kondrashov, A. S., & Sunyaev, S. R. (2010) · Nature Methods · doi:10.1038/nmeth0410-248 · Cited by 13,194
Enrichr: a comprehensive gene set enrichment analysis web server 2016 update
Kuleshov, M. V., Jones, M. R., Rouillard, A. D., Fernandez, N. F., Duan, Q., Wang, Z., Koplev, S., Jenkins, S. L., Jagodnik, K. M., Lachmann, A., et al. (2016) · Nucleic Acids Research · doi:10.1093/nar/gkw377 · Cited by 10,799
Signatures of mutational processes in human cancer
Initiative, A. P. C. G., Alexandrov, L. B., ICGC Breast Cancer Consortium, ICGC MMML-Seq Consortium, PedBrain, I., Nik-Zainal, S., Wedge, D. C., Aparicio, S. A. J. R., Behjati, S., Biankin, A. V., et al. (2013) · Nature · doi:10.1038/nature12477 · Cited by 9,660
KEGG: new perspectives on genomes, pathways, diseases and drugs
Kanehisa, M., Furumichi, M., Tanabe, M., Sato, Y., & Morishima, K. (2017) · Nucleic Acids Research · doi:10.1093/nar/gkw1092 · Cited by 8,865
Comprehensive molecular characterization of human colon and rectal cancer
The Cancer Genome Atlas Network (2012) · Nature · doi:10.1038/nature11252 · Cited by 8,499
Cited by
Exploring treatment options in cancer: tumor treatment strategies
Liu, B., Zhou, H., Tan, L., Siu, K. T. H., & Guan, X.-Y. (2024) · Signal Transduction and Targeted Therapy · doi:10.1038/s41392-024-01856-7 · Cited by 1,201
The current state of the art and future trends in RAS-targeted cancer therapies
Punekar, S. R., Velcheti, V., Neel, B. G., & Wong, K.-K. (2022) · Nature Reviews Clinical Oncology · doi:10.1038/s41571-022-00671-9 · Cited by 549
Concurrent inhibition of oncogenic and wild-type RAS-GTP for cancer therapy
Holderfield, M., Lee, B. J., Jiang, J., Tomlinson, A., Seamon, K. J., Mira, A., Patrucco, E., Goodhart, G., Dilly, J., Gindin, Y., et al. (2024) · Nature · doi:10.1038/s41586-024-07205-6 · Cited by 369
Translational and therapeutic evaluation of RAS-GTP inhibition by RMC-6236 in RAS-driven cancers
Jiang, J., Jiang, L., Maldonato, B. J., Wang, Y., Holderfield, M., Aronchik, I., Winters, I. P., Salman, Z., Blaj, C., Menard, M., et al. (2024) · Cancer Discovery · doi:10.1158/2159-8290.cd-24-0027 · Cited by 343
“The relative representation of oncogenic RAS variants is highly variable across cancer types, likely due to an interplay between allele-specific biochemical, structural, and signaling distinctions and tissue (indication)-specific properties (3, 5).”
Translational advances in pancreatic ductal adenocarcinoma therapy
Hosein, A. N., Dougan, S. K., Aguirre, A. J., & Maitra, A. (2022) · Nature Cancer · doi:10.1038/s43018-022-00349-2 · Cited by 250
Metadata
- Authors
- Cook, J. H., Melloni, G. E. M., Gulhan, D. C., Park, P. J., & Haigis, K. M.
- Affiliations
- Brigham and Women's Hospital (US)
- Harvard University (US)
- Broad Institute (US)
- Corresponding author
- Park, Peter J.
- Haigis, Kevin M.
- Venue
- Nature Communications
- Published
- 2021-03-22
- Volume
- 12
- Issue
- 1
- Pages
- 1808-1808
- ISSN
- 2041-1723 (ISSN-L)
- PMC
- PMC7985210
- OpenAlex
- W3138323180
- Cited by
- 208
- References
- 109
- Citations by year
- 2026: 30 · 2025: 45 · 2024: 48 · 2023: 39 · 2022: 35 · 2021: 11
- FWCI
- 13.33
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- Nature Communications (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- American Association for Cancer Research
- Multiple Myeloma Research Foundation
- Cancer Research UK (27144)
- National Institutes of Health
- National Cancer Institute (R01CA232372, U01CA199252)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Cancer Research › Cancer Genomics and Diagnostics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Kinase Regulation and GTPase Signaling
- Health Sciences › Medicine › Oncology › Pancreatic and Hepatic Oncology Research
- Keywords
- KRAS mutation, colorectal cancer, pancreatic cancer, lung cancer, missense mutations, gene dependency, mutagenic mechanisms
- MeSH terms
- Alleles, Gene Frequency, Humans, Mutation, Neoplasms, Organ Specificity, Gene Expression Regulation, Neoplastic, Proto-Oncogene Proteins p21(ras), Gene Regulatory Networks
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-29, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-09-29 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: The origins and genetic interactions of KRAS mutations are allele- and tissue-specific. https://app.cochat.ai/papers/10.1038/s41467-021-22125-z
- The paper
- Cook, J. H., Melloni, G. E. M., Gulhan, D. C., Park, P. J., & Haigis, K. M. (2021). The origins and genetic interactions of KRAS mutations are allele- and tissue-specific. Nature Communications. https://doi.org/10.1038/s41467-021-22125-z
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.