CoChat

Assessing Target Specificity of the Small Molecule Inhibitor MARIMASTAT to Snake Venom Toxins: A Novel Application of Thermal Proteome Profiling

Open access

Published version of a preprint

The DOI matches its Crossref record; 139 of the 139 references it cites were found in a registry or an index; none of the 107 sources its claims cite is checked for support yet.

Abstract

New treatments that circumvent the pitfalls of traditional antivenom therapies are critical to address the problem of snakebite globally. Numerous snake venom toxin inhibitors have shown promising cross-species neutralization of medically significant venom toxins in vivo and in vitro. The development of high-throughput approaches for the screening of such inhibitors could accelerate their identification, testing, and implementation and thus holds exciting potential for improving the treatments and outcomes of snakebite envenomation worldwide. Energetics-based proteomic approaches, including thermal proteome profiling and proteome integral solubility alteration (PISA) assays, represent "deep proteomics" methods for high throughput, proteome-wide identification of drug targets and ligands. In the following study, we apply thermal proteome profiling and PISA methods to characterize the interactions between venom toxin proteoforms in Crotalus atrox (Western Diamondback Rattlesnake) and the snake venom metalloprotease (SVMP) inhibitor marimastat. We investigate its venom proteome-wide effects and characterize its interactions with specific SVMP proteoforms, as well as its potential targeting of non-SVMP venom toxin families. We also compare the performance of PISA thermal window and soluble supernatant with insoluble precipitate using two inhibitor concentrations, providing the first demonstration of the utility of a sensitive high-throughput PISA-based approach to assess the direct targets of small molecule inhibitors for snake venom.

Overview

CoChat Summary

TL;DR

This study applies thermal proteome profiling and PISA to *Crotalus atrox* venom to map thermal stability and identify direct targets of marimastat. It finds SVMPs are stabilized by marimastat, with a narrow 56–60 °C PISA window and pellet fraction giving the most sensitive target identification, while also detecting possible off-target interactions with PLA2s, CTLs, and other families.

Key findings

  • The *C. atrox* venom proteome contained 46 unique proteoforms; SVMPs and SVSPs represented nearly 70% of the venom proteome. Meltome analysis gave toxin Tm values from 47.8 to 74.3 °C, with most toxins melting between 50–60 °C (n = 22) or 60–70 °C (n = 28); 11 proteoforms had no calculable Tm at 75 °C. SVSPs had the highest average Tm (63.9 °C, stdev = 5.5 °C), while four CTLs had an average Tm of 49.3 °C (stdev = 1.1 °C) and hyaluronidase melted at 50.6 °C.
  • In the broad 40–70 °C PISA window, 15 μM marimastat caused five of 21 SVMP proteoforms to show a stabilizing shift in treated supernatant; 150 μM marimastat additionally stabilized PII 23556 and PII 27392. In pellets, 15 μM marimastat significantly reduced four SVMPs (Atro B, Atro-D, SVMPIII 27520, and SVMPIII 28348), and 150 μM also reduced PII 23541.
  • A narrower 56–60 °C PISA window improved target detection: 15 μM marimastat stabilized three SVMPs in supernatant, 150 μM stabilized seven, and pellet analysis identified 11 of 21 SVMP proteoforms at 15 μM and 12 at 150 μM. When requiring significance in both supernatant and pellet, the broad window identified only two SVMPs, whereas the narrow window identified six (VAP1, PII 23556, PIII 28348, PIII 28771, PII 23541, PIII 27392).
  • Validation by RP-HPLC showed heat-treated SVMPs lost 44% of original peak area, but marimastat treatment resulted in only a 7% decrease; PLA2 at 27 min decreased by 75% with heat alone versus 26% with marimastat, and PLA2 at 30 min was virtually absent in heated control but lost only 52% abundance with marimastat. PISA also detected possible off-target stabilization of PLA2s, CTLs, hyaluronidase, VEGF, and SVSPs.

Methods

  • *C. atrox* venom was profiled by TPP across 37–75 °C and by PISA after 30 min incubation with marimastat (15 μM or 150 μM) or vehicle, using both a broad 40–70 °C and a narrow 56–60 °C thermal window; soluble supernatant and insoluble pellet fractions were analyzed separately.
  • A custom *C. atrox* venom gland transcriptome database (570 entries) was used for LC-MS/MS data interpretation; PISA assays were performed in triplicate, and label-free quantification used sum-normalized spectral intensity.
  • TPP melting curves were generated in ProSAP; PISA targets were identified using soluble abundance ratios (SAR), two-tailed Student’s t tests (p < 0.05), and significance thresholds of log2SAR ≥ 0.5 or ≤ −0.5, with validation by SDS-PAGE and RP-HPLC.

Limitations

  • Stated: PISA may suffer reduced screening sensitivity because it pools temperature points, and thermal-shift assays are less ideal for thermostable proteoforms such as SVSPs and CRISPs, which may have high thermal tolerance or partially refold after thermal stress.
  • Stated: pellet data showed more noise, with non-target proteins reaching significance, whereas soluble fraction analysis was considered more reliable for quantification; no reduction in PLA2 activity was detected in marimastat-treated samples (data not shown).
  • Evident from the text: the study is in vitro, uses a single venom species (*C. atrox*) and a single inhibitor (marimastat), and does not test in vivo efficacy or cross-species neutralization.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1016/j.mcpro.2024.100779

Cites

139 of 139 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Graphical abstract
Graphical abstract CC BY-NC-ND · PubMed Central
Fig. 1 TPP and PISA workflows.A, in TPP experiments, samples are heated between 40–70 °C and centrifuged to pellet denatured proteins. Samples are reduced, alkylated, and trypsin digested and analyzed with LC-MS/MS for protein identification. Melting curves are generated in ProSAP using unique intensity for each protein identified. Tm = melting temperature of 50% of population. B, in PISA, venom is incubated for 30 min at 37 °C with an inhibitor or alone. Samples are heated from 40 to 70 °C and pooled before centrifuging to pellet insoluble material. Samples are prepared as mentioned above and analyzed via LC-MS/MS for protein identification. To identify inhibitor targets, unique intensity is used to calculate SAR values for each protein followed by identification of significant outliers. PISA, proteome integral solubility alteration; TPP, thermal proteome profiling.
Fig. 1 TPP and PISA workflows.A, in TPP experiments, samples are heated between 40–70 °C and centrifuged to pellet denatured proteins. Samples are reduced, alkylated, and trypsin digested and analyzed with LC-MS/MS for protein identification. Melting curves are generated in ProSAP using unique intensity for each protein identified. Tm = melting temperature of 50% of population. B, in PISA, venom is incubated for 30 min at 37 °C with an inhibitor or alone. Samples are heated from 40 to 70 °C and pooled before centrifuging to pellet insoluble material. Samples are prepared as mentioned above and analyzed via LC-MS/MS for protein identification. To identify inhibitor targets, unique intensity is used to calculate SAR values for each protein followed by identification of significant outliers. PISA, proteome integral solubility alteration; TPP, thermal proteome profiling.CC BY-NC-ND · PubMed Central
Fig. 2 Crotalus atrox venom proteome characterization.A, toxin family abundances in C. atrox venom modified from Calvete et al., 2009. B, the number of proteoforms identified in C. atrox venom in the present study organized by family. C, RP-HPLC separated C. atrox venom. For peak identification, fractions were analyzed with mass spectrometry and SDS-PAGE and compared to known masses from Calvete et al., 2009. PP, bradykinin potentiating peptide; CRISP, cysteine rich secretory protein; CTL, C-type lectin; Dis, disintegrin; GPC, glutaminyl-peptide cyclotransferase; HPLC, high-performance liquid chromatography; HYAL, hyaluronidase; L-AAO, L-amino acid oxidase; NGF, nerve growth factor; PLA2, phospholipase A2; PLB, phospholipase B; RP-HPLC, reverse-phase high-performance liquid chromatography; SVMP, snake venom metalloprotease; SVMPi, SVMP tripeptide inhibitor; SVSP, snake venom serine protease; VEGF, vascular endothelial growth factor.
Fig. 2 Crotalus atrox venom proteome characterization.A, toxin family abundances in C. atrox venom modified from Calvete et al., 2009. B, the number of proteoforms identified in C. atrox venom in the present study organized by family. C, RP-HPLC separated C. atrox venom. For peak identification, fractions were analyzed with mass spectrometry and SDS-PAGE and compared to known masses from Calvete et al., 2009. PP, bradykinin potentiating peptide; CRISP, cysteine rich secretory protein; CTL, C-type lectin; Dis, disintegrin; GPC, glutaminyl-peptide cyclotransferase; HPLC, high-performance liquid chromatography; HYAL, hyaluronidase; L-AAO, L-amino acid oxidase; NGF, nerve growth factor; PLA2, phospholipase A2; PLB, phospholipase B; RP-HPLC, reverse-phase high-performance liquid chromatography; SVMP, snake venom metalloprotease; SVMPi, SVMP tripeptide inhibitor; SVSP, snake venom serine protease; VEGF, vascular endothelial growth factor.CC BY-NC-ND · PubMed Central
Fig. 3 Crotalus atrox meltome characterization.A, SDS-PAGE of C. atrox venom heated at temperatures between 37–70 °C for 3 min. B, heatmap showing the soluble fraction of most toxin proteoforms relative to 37 °C in C. atrox venom at 40, 45, 49, 52, 57, 62, 66, 69, and 75 °C. Heatmap colors represent the number of SDs away from the mean of the protein’s intensity in each row of the heatmap. C, distribution of melting temperatures organized by family. Dotted lines represent median and quartile ranges. D, distribution of melting temperatures for all toxins identified. Nonmelters (NM) are classified as proteoforms for which Tm could not be calculated when heated to a maximum temperature of 75 °C. E, representative melting curve of a CTL (Crotocetin). Abundance is normalized to 37 °C. F, representative melting curve of a PLA2 (Cvv-N6). Abundance is normalized to 37 °C. G, representative melting curve of an SVMP (PIII 28325). Abundance is normalized to 37 °C. CTL, c-type lectin; SVMP, snake venom metalloprotease.
Fig. 3 Crotalus atrox meltome characterization.A, SDS-PAGE of C. atrox venom heated at temperatures between 37–70 °C for 3 min. B, heatmap showing the soluble fraction of most toxin proteoforms relative to 37 °C in C. atrox venom at 40, 45, 49, 52, 57, 62, 66, 69, and 75 °C. Heatmap colors represent the number of SDs away from the mean of the protein’s intensity in each row of the heatmap. C, distribution of melting temperatures organized by family. Dotted lines represent median and quartile ranges. D, distribution of melting temperatures for all toxins identified. Nonmelters (NM) are classified as proteoforms for which Tm could not be calculated when heated to a maximum temperature of 75 °C. E, representative melting curve of a CTL (Crotocetin). Abundance is normalized to 37 °C. F, representative melting curve of a PLA2 (Cvv-N6). Abundance is normalized to 37 °C. G, representative melting curve of an SVMP (PIII 28325). Abundance is normalized to 37 °C. CTL, c-type lectin; SVMP, snake venom metalloprotease.CC BY-NC-ND · PubMed Central
Fig. 4 Crotalus atrox venom-wide interactions with two concentrations of marimastat with temperature window from 40 to 70 °C.A, volcano plot comparing soluble supernatant of heat-treated venom + marimastat (15 μM) to heat-treated venom alone. X indicates SVMP proteoforms, red = positive outliers, blue = negative outliers, gray = not significant. B, volcano plot comparing soluble supernatant of heat-treated venom + marimastat (150 μM) to heat-treated venom alone. C, volcano plot comparing insoluble precipitate of heat-treated venom + marimastat (15 μM) to heat-treated venom alone. D, volcano plot comparing insoluble precipitate of heat-treated venom + marimastat (150 μM) to heat-treated venom alone. SVMP, snake venom metalloprotease.
Fig. 4 Crotalus atrox venom-wide interactions with two concentrations of marimastat with temperature window from 40 to 70 °C.A, volcano plot comparing soluble supernatant of heat-treated venom + marimastat (15 μM) to heat-treated venom alone. X indicates SVMP proteoforms, red = positive outliers, blue = negative outliers, gray = not significant. B, volcano plot comparing soluble supernatant of heat-treated venom + marimastat (150 μM) to heat-treated venom alone. C, volcano plot comparing insoluble precipitate of heat-treated venom + marimastat (15 μM) to heat-treated venom alone. D, volcano plot comparing insoluble precipitate of heat-treated venom + marimastat (150 μM) to heat-treated venom alone. SVMP, snake venom metalloprotease.CC BY-NC-ND · PubMed Central
Fig. 5 Validation assays of inhibitor interactions.A, SDS-PAGE comparison of heat-treated venom to heat-treated venoms incubated with 150 μM of marimastat with a thermal window of 40–70 °C. Note the recovery in band size and intensity of SVMP and PLA2 bands in marimastat-treated samples. The left side indicates molecular mass standards in kDa. B, enlarged HPLC-separated SVMP peak overlay comparing abundance of non-heat-treated venom (black), heat-treated venom (red), and venom heat-treated after incubation with 150 μM of marimastat (blue). Note the recovery of peak area in the inhibitor-treated sample. C, enlarged HPLC-separated PLA2 peak overlay comparing abundance of non-heat-treated venom (black), heat-treated venom (red), and venom heat-treated after incubation with 150 μM of marimastat (blue). Note the recovery of PLA2 peak area in the inhibitor-treated sample. HPLC, high-performance liquid chromatography; SVMP, snake venom metalloprotease.
Fig. 5 Validation assays of inhibitor interactions.A, SDS-PAGE comparison of heat-treated venom to heat-treated venoms incubated with 150 μM of marimastat with a thermal window of 40–70 °C. Note the recovery in band size and intensity of SVMP and PLA2 bands in marimastat-treated samples. The left side indicates molecular mass standards in kDa. B, enlarged HPLC-separated SVMP peak overlay comparing abundance of non-heat-treated venom (black), heat-treated venom (red), and venom heat-treated after incubation with 150 μM of marimastat (blue). Note the recovery of peak area in the inhibitor-treated sample. C, enlarged HPLC-separated PLA2 peak overlay comparing abundance of non-heat-treated venom (black), heat-treated venom (red), and venom heat-treated after incubation with 150 μM of marimastat (blue). Note the recovery of PLA2 peak area in the inhibitor-treated sample. HPLC, high-performance liquid chromatography; SVMP, snake venom metalloprotease.CC BY-NC-ND · PubMed Central
Fig. 6 Crotalus atrox venom-wide interactions with two concentrations of marimastat with temperature window from 56 to 60 °C.A, volcano plot comparing soluble supernatant of heat-treated venom + marimastat (15 μM) to heat-treated venom alone. X indicates SVMP proteoforms, red = positive outliers, blue = negative outliers, gray = not significant. B, volcano plot comparing soluble supernatant of heat-treated venom + marimastat (150 μM) to heat-treated venom alone. C, volcano plot comparing insoluble precipitate of heat-treated venom + marimastat (15 μM) to heat-treated venom alone. D, volcano plot comparing insoluble precipitate of heat-treated venom + marimastat (150 μM) to heat-treated venom alone. SVMP, snake venom metalloprotease.
Fig. 6 Crotalus atrox venom-wide interactions with two concentrations of marimastat with temperature window from 56 to 60 °C.A, volcano plot comparing soluble supernatant of heat-treated venom + marimastat (15 μM) to heat-treated venom alone. X indicates SVMP proteoforms, red = positive outliers, blue = negative outliers, gray = not significant. B, volcano plot comparing soluble supernatant of heat-treated venom + marimastat (150 μM) to heat-treated venom alone. C, volcano plot comparing insoluble precipitate of heat-treated venom + marimastat (15 μM) to heat-treated venom alone. D, volcano plot comparing insoluble precipitate of heat-treated venom + marimastat (150 μM) to heat-treated venom alone. SVMP, snake venom metalloprotease.CC BY-NC-ND · PubMed Central
Fig. 7 Comparison of a broad (40–70 °C) to a narrow (56–60 °C) PISA thermal window. PCA plot comparing replicates of soluble supernatant of (A) heat-treated venom + 15 μM marimastat to heat-treated venom alone and (B) heat-treated venom +150 μM marimastat to heat-treated venom alone. PCA plot with 95% confidence intervals comparing replicates of insoluble precipitate of (C) heat-treated venom +15 μM marimastat to heat-treated venom alone and (D) heat-treated venom +150 μM marimastat to heat-treated venom alone. E, number of SVMP and non-SVMP proteins identified with significant thermal shifts toward stabilization (p-value < 0.05 log2SAR> 0.5) after 15 μM or 150 μM marimastat treatment in supernatants and pellets heat-treated at a broad (40–70 °C) or a narrow (56–60 °C) thermal window. Venn diagrams of SVMP proteins identified with significant thermal shifts toward stabilization (p-value < 0.05, log2SAR > 0.5) in supernatants and pellets heat-treated at a broad (40–70 °C) or a narrow (56–60 °C) thermal window after (F) 15 μM or (G) 150 μM marimastat treatment. Scatter plot showing log2SAR values calculated from 150 μM marimastat-treated venom versus vehicle treatment heated at from (H) 40–70 °C or (I) 56 to 60 °C that meet significance criteria in both supernatant (SN) and precipitate (Pellet) group. Largest outliers of SVMP and non-SVMP proteoforms are labeled. ACLD, PIII SVMP ACLD; black, SVMP proteoforms; Con, control; CTL, C-type lectin; gray, non-SVMP toxins; Hyal, hyaluronidase; PCA, principal component analysis; Pell, pellet; PISA, proteome integral solubility alteration; SN, supernatant; SVMP, snake venom metalloprotease.
Fig. 7 Comparison of a broad (40–70 °C) to a narrow (56–60 °C) PISA thermal window. PCA plot comparing replicates of soluble supernatant of (A) heat-treated venom + 15 μM marimastat to heat-treated venom alone and (B) heat-treated venom +150 μM marimastat to heat-treated venom alone. PCA plot with 95% confidence intervals comparing replicates of insoluble precipitate of (C) heat-treated venom +15 μM marimastat to heat-treated venom alone and (D) heat-treated venom +150 μM marimastat to heat-treated venom alone. E, number of SVMP and non-SVMP proteins identified with significant thermal shifts toward stabilization (p-value < 0.05 log2SAR> 0.5) after 15 μM or 150 μM marimastat treatment in supernatants and pellets heat-treated at a broad (40–70 °C) or a narrow (56–60 °C) thermal window. Venn diagrams of SVMP proteins identified with significant thermal shifts toward stabilization (p-value < 0.05, log2SAR > 0.5) in supernatants and pellets heat-treated at a broad (40–70 °C) or a narrow (56–60 °C) thermal window after (F) 15 μM or (G) 150 μM marimastat treatment. Scatter plot showing log2SAR values calculated from 150 μM marimastat-treated venom versus vehicle treatment heated at from (H) 40–70 °C or (I) 56 to 60 °C that meet significance criteria in both supernatant (SN) and precipitate (Pellet) group. Largest outliers of SVMP and non-SVMP proteoforms are labeled. ACLD, PIII SVMP ACLD; black, SVMP proteoforms; Con, control; CTL, C-type lectin; gray, non-SVMP toxins; Hyal, hyaluronidase; PCA, principal component analysis; Pell, pellet; PISA, proteome integral solubility alteration; SN, supernatant; SVMP, snake venom metalloprotease.CC BY-NC-ND · PubMed Central
Fig. 8 Effects of marimastat treatment and a narrow thermal window on SVMP proteoforms only.A, number of proteins identified in supernatant (SN) and pellet of narrow thermal window that meet significance criteria (p-value < 0.01, log2SAR> 0.5); (B) heatmap of sum-normalized intensity values in supernatant or precipitate of SVMPs from 15 μM or 150 μM marimastat-treated venom. Heatmap colors are scaled by row to better visualize variation in sum-normalized intensity between classes. C, heatmap of sum-normalized intensity values in supernatant or precipitate of SVMPs from 15 μM or 150 μM marimastat-treated venom or vehicle control showing concentration-dependent shifts in abundance. Heatmap colors are scaled by row to better visualize variation in sum-normalized intensity between classes. D, correlation plot of SAR values showing strongest marimastat targets based on the effects of marimastat treatment on SVMP proteoform intensity. Comparison of concentration-dependent intensity shifts between precipitate and supernatant of (E) the most abundant SVMP proteoform VAP2B and (F) a less abundant SVMP proteoform PIII 28348 at both concentrations of marimastat at the narrow thermal window. SVMP, snake venom metalloprotease.
Fig. 8 Effects of marimastat treatment and a narrow thermal window on SVMP proteoforms only.A, number of proteins identified in supernatant (SN) and pellet of narrow thermal window that meet significance criteria (p-value < 0.01, log2SAR> 0.5); (B) heatmap of sum-normalized intensity values in supernatant or precipitate of SVMPs from 15 μM or 150 μM marimastat-treated venom. Heatmap colors are scaled by row to better visualize variation in sum-normalized intensity between classes. C, heatmap of sum-normalized intensity values in supernatant or precipitate of SVMPs from 15 μM or 150 μM marimastat-treated venom or vehicle control showing concentration-dependent shifts in abundance. Heatmap colors are scaled by row to better visualize variation in sum-normalized intensity between classes. D, correlation plot of SAR values showing strongest marimastat targets based on the effects of marimastat treatment on SVMP proteoform intensity. Comparison of concentration-dependent intensity shifts between precipitate and supernatant of (E) the most abundant SVMP proteoform VAP2B and (F) a less abundant SVMP proteoform PIII 28348 at both concentrations of marimastat at the narrow thermal window. SVMP, snake venom metalloprotease.CC BY-NC-ND · PubMed Central

Verification

Claims the paper makes

0 Valid · 0 Not valid · 62 Not checked · 107 Own results and interpretations

79 more claims

Create a free account to see all 169 claims and to verify the 106 references still open.

See all claims — free account
  • It may state two separate claims as one. 52
  • Probably not a substantial claim. 41

Create a free account to see these claims.

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.

3 of 3

Citations

Cites

Research works printed but not linked: 1

  1. Mackessy S.P. Venom composition in rattlesnakes: trends and biological significance The Biology of Rattlesnakes 2008 Loma Linda University Press Loma Linda

Cited by

See all 11

Metadata

Authors
Smith, C. F., Modahl, C. M., Ceja Galindo, D., Larson, K. Y., Maroney, S. P., Bahrabadi, L., Brandehoff, N. P., Perry, B. W., McCabe, M. C., Petras, D., Lomonte, B., Calvete, J. J., Castoe, T. A., Mackessy, S. P., Hansen, K. C., & Saviola, A. J.
Affiliations
University of California, Riverside (US)
Corresponding author
Saviola, Anthony J.
Venue
Molecular & Cellular Proteomics
Publisher
Elsevier BV
Published
2024-06
Volume
23
Issue
6
Pages
100779
ISSN
1535-9476 (ISSN-L)
OpenAlex
W4395694226
Cited by
11
References
139
Citations by year
2026: 7 · 2025: 3 · 2024: 1
FWCI
2.33
Type
journal article
Language
en
Open access
Licence
Locations
  • Molecular & Cellular Proteomics (publishedVersion, cc-by-nc-nd)
  • bioRxiv (Cold Spring Harbor Laboratory) (submittedVersion, cc-by-nd)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
  • LSTM Online Archive (Liverpool School of Tropical Medicine) (acceptedVersion)
  • Investigative News in Education (Universidad de Costa Rica) (submittedVersion, cc-by-nc-nd)
  • DIGITAL.CSIC (Spanish National Research Council (CSIC)) (submittedVersion, cc-by-nc-nd)
Funders
US Department of Defense (ID07200010-301-35)
Topics
  • Life Sciences › Pharmacology, Toxicology and Pharmaceutics › Toxicology › Venomous Animal Envenomation and Studies
  • Physical Sciences › Chemistry › Spectroscopy › Advanced Proteomics Techniques and Applications
  • Health Sciences › Medicine › Pharmacology (medical) › Healthcare and Venom Research
Keywords
marimastat, snake venom toxins, Crotalus atrox, snakebite, thermal proteome profiling, target specificity, small molecule inhibitors, antivenom alternatives, high-throughput inhibitor screening
MeSH terms
Animals, Crotalid Venoms, Hydroxamic Acids, Snake Venoms, Crotalus, Proteome, Proteomics, Metalloproteases

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.

A language model drew the claims from the paper on 2026-11-22, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-22 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

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

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.