Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat
Sign Up
Cited by 23RNA modifications and cancerViral-associated cancers and disordersRNA Research and Splicing

The DOI matches its Crossref record; 48 of the 48 references it cites were found in a registry or an index.

SUMOylation of the m6A reader YTHDF2 by PIAS1 promotes viral RNA decay to restrict EBV replication

Open access
Sugiokto, Febri Gunawan, Saiada, Farjana, Zhang, Kun, Li, Renfeng·2024·mBio
DOI 10.1128/mbio.03168-23PMC PMC10865817

Published version of a preprint

Open this paper in CoChat — free account

Abstract

m6A RNA modification pathway plays important roles in diverse cellular processes and viral life cycle. Here, we investigated the relationship between PIAS1 and the m6A reader protein YTHDF2, which is involved in regulating RNA stability by binding to m6A-modified RNA. We found that both the N-terminal and C-terminal regions of YTHDF2 interact with PIAS1. We showed that PIAS1 promotes the SUMOylation of YTHDF2 at three specific lysine residues. We also demonstrated that PIAS1 enhances the anti-EBV activity of YTHDF2. We further revealed that PIAS1 mediates the SUMOylation of other YTHDF family members, namely, YTHDF1 and YTHDF3, to limit EBV replication. These findings together illuminate an important regulatory mechanism of YTHDF proteins in controlling viral RNA decay and EBV replication through PIAS1-mediated SUMOylation.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1128/mbio.03168-23

Cites

48 of 48 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig 1 PIAS1 promotes YTHDF2 SUMOylation. HEK293T cells were transfected with Halo-V5-YTHDF2, Halo-PIAS1, and SUMO2 plasmids as indicated. (A) WB was performed for the whole-cell lysate (input) using anti-SUMO2/3, anti-PIAS1, anti-V5, and anti-β-actin antibodies. Bracket denotes SUMOylated proteins. (B) YTHDF2 was IPed with anti-V5 magnetic beads, and WB was performed using antibodies as indicated. Bracket denotes SUMOylated YTHDF2. IP, immunoprecipitation.
Fig 1 PIAS1 promotes YTHDF2 SUMOylation. HEK293T cells were transfected with Halo-V5-YTHDF2, Halo-PIAS1, and SUMO2 plasmids as indicated. (A) WB was performed for the whole-cell lysate (input) using anti-SUMO2/3, anti-PIAS1, anti-V5, and anti-β-actin antibodies. Bracket denotes SUMOylated proteins. (B) YTHDF2 was IPed with anti-V5 magnetic beads, and WB was performed using antibodies as indicated. Bracket denotes SUMOylated YTHDF2. IP, immunoprecipitation.CC BY · PubMed Central
Fig 2 PIAS1 interacts with YTHDF2. (A) Illustration of full-length YTHDF2 (1–579) or YTHDF2 truncation mutants. CB denotes CNOT1-binding domain; P/Q/N is a P/Q/N rich aggregation-prone region; YTH denotes m6A RNA-binding domain. (B) HEK293T cells were co-transfected with full-length YTHDF2 and truncated YTHDF2 plasmids as indicated. WB analyses showing that PIAS1 is Co-IPed with N-terminal and C-terminal regions of YTHDF2. β-Actin blot was included as loading control. Arrows denote the position of full-length and truncated YTHDF2. (C) The schematic representation showing PIAS1 domains, V5-tagged or Halo-V5 tagged PIAS1 truncation mutants. SAP (SAF-A/B, Acinus, and PIAS) is a DNA and protein-binding domain; PINIT denotes the nuclear localization motif; RING indicate RING finger E3 ligase domain for protein SUMOylation; ST rich denotes variable Ser/Thr-rich region. (D) HEK293T cells were co-transfected with HA-YTHDF2 and truncated versions of V5-PIAS1 or Halo-V5-PIAS1. WB analysis showing that YTHDF2 is Co-IPed with the N-terminal and middle part of PIAS1. β-Actin blot was included as loading control. (E) A proposed model showing the N- and C-terminal regions of YTHDF2 binding to the central part of PIAS1.
Fig 2 PIAS1 interacts with YTHDF2. (A) Illustration of full-length YTHDF2 (1–579) or YTHDF2 truncation mutants. CB denotes CNOT1-binding domain; P/Q/N is a P/Q/N rich aggregation-prone region; YTH denotes m6A RNA-binding domain. (B) HEK293T cells were co-transfected with full-length YTHDF2 and truncated YTHDF2 plasmids as indicated. WB analyses showing that PIAS1 is Co-IPed with N-terminal and C-terminal regions of YTHDF2. β-Actin blot was included as loading control. Arrows denote the position of full-length and truncated YTHDF2. (C) The schematic representation showing PIAS1 domains, V5-tagged or Halo-V5 tagged PIAS1 truncation mutants. SAP (SAF-A/B, Acinus, and PIAS) is a DNA and protein-binding domain; PINIT denotes the nuclear localization motif; RING indicate RING finger E3 ligase domain for protein SUMOylation; ST rich denotes variable Ser/Thr-rich region. (D) HEK293T cells were co-transfected with HA-YTHDF2 and truncated versions of V5-PIAS1 or Halo-V5-PIAS1. WB analysis showing that YTHDF2 is Co-IPed with the N-terminal and middle part of PIAS1. β-Actin blot was included as loading control. (E) A proposed model showing the N- and C-terminal regions of YTHDF2 binding to the central part of PIAS1.CC BY · PubMed Central
Fig 3 PIAS1 cooperates with YTHDF2 to suppress EBV lytic replication. HEK293 (EBV+) cells were co-transfected with plasmids encoding ZTA, YTHDF2, and truncated forms of PIAS1 (1–433, 409–651, and 1–205) as indicated. The relative EBV copy numbers were measured using the quantitative polymerase chain reaction (qPCR). The value of lane 1 was set as 1. The protein expression levels were monitored by WB using antibodies as indicated. β-Actin blot was included as loading control. Arrows denote the relative positions of PIAS1 fragments. # denotes a non-specific band. Results from three biological replicates are presented. Error bars indicate standard deviations. **P < 0.01, ***P < 0.001.
Fig 3 PIAS1 cooperates with YTHDF2 to suppress EBV lytic replication. HEK293 (EBV+) cells were co-transfected with plasmids encoding ZTA, YTHDF2, and truncated forms of PIAS1 (1–433, 409–651, and 1–205) as indicated. The relative EBV copy numbers were measured using the quantitative polymerase chain reaction (qPCR). The value of lane 1 was set as 1. The protein expression levels were monitored by WB using antibodies as indicated. β-Actin blot was included as loading control. Arrows denote the relative positions of PIAS1 fragments. # denotes a non-specific band. Results from three biological replicates are presented. Error bars indicate standard deviations. **P < 0.01, ***P < 0.001.CC BY · PubMed Central
Fig 4 PIAS1 SUMOylates YTHDF2 at K281, K571, and K572. (A) In vitro SUMOylation assay was performed with the combination of purified E1, E2, SUMO2, PIAS1, and substrate V5-YTHDF2 proteins as indicated. The reaction was terminated with sodium dodecyl sulfate (SDS) sample loading buffer, and WB was performed using anti-V5-HRP antibody. Bracket denotes SUMOylated YTHDF2. (B) The schematic representation of YTHDF2 protein and the corresponding lysine residues (red) within SUMOylation consensus motifs. In vitro SUMOylation assay was performed with the combination of E1, E2, SUMO2, PIAS1, and either wild-type (WT) YTHDF2 or individual YTHDF2 mutants (K281R, K571R, and K572R) as indicated. The reaction was terminated with SDS sample loading buffer, and WB was performed using anti-V5-HRP antibody. (C) In vitro SUMOylation assay was performed for WT YTHDF2 and RRR mutant (K281R/K571R/K572R) as indicated. (D) Localization of K281, K571, and K572 in YTHDF2. The YTHDF2 three-dimensional structure was predicted by AlphaFold. The locations of indicated SUMOylation sites are marked by arrows. (E) Sequence alignment of YTHDF2 protein sequences from 10 species using the Constraint-based Multiple Alignment Tool (COBALT). The SUMOylation sites corresponding to human YTHFD2 are highlighted in yellow.
Fig 4 PIAS1 SUMOylates YTHDF2 at K281, K571, and K572. (A) In vitro SUMOylation assay was performed with the combination of purified E1, E2, SUMO2, PIAS1, and substrate V5-YTHDF2 proteins as indicated. The reaction was terminated with sodium dodecyl sulfate (SDS) sample loading buffer, and WB was performed using anti-V5-HRP antibody. Bracket denotes SUMOylated YTHDF2. (B) The schematic representation of YTHDF2 protein and the corresponding lysine residues (red) within SUMOylation consensus motifs. In vitro SUMOylation assay was performed with the combination of E1, E2, SUMO2, PIAS1, and either wild-type (WT) YTHDF2 or individual YTHDF2 mutants (K281R, K571R, and K572R) as indicated. The reaction was terminated with SDS sample loading buffer, and WB was performed using anti-V5-HRP antibody. (C) In vitro SUMOylation assay was performed for WT YTHDF2 and RRR mutant (K281R/K571R/K572R) as indicated. (D) Localization of K281, K571, and K572 in YTHDF2. The YTHDF2 three-dimensional structure was predicted by AlphaFold. The locations of indicated SUMOylation sites are marked by arrows. (E) Sequence alignment of YTHDF2 protein sequences from 10 species using the Constraint-based Multiple Alignment Tool (COBALT). The SUMOylation sites corresponding to human YTHFD2 are highlighted in yellow.CC BY · PubMed Central
Fig 5 SUMOylation-deficient YTHDF2 impairs its ability to restrict EBV replication. (A) HEK293 (EBV+) cells were transfected with ZTA plasmid (lytic inducer), PIAS1, and WT YTHDF2 (WT) or SUMOylation-deficient YTHDF2 (RRR) as indicated. The relative EBV copy numbers were measured using the qPCR as described in the Materials and Methods. The value of lane 1 was set as 1. The expression levels of YTHDF2 and PIAS1 were monitored by WB. β-Actin blot was included as loading control. (B) Akata (EBV+) cells were used to create cell lines using pLenti-Vector, pLenti-YTHDF2-WT, and p-Lenti-YTHDF2-RRR (K281R/K571R/K572R). EBV lytic cycle was induced by anti-IgG-mediated B-cell receptor cross-linking. The relative EBV copy numbers were measured using the qPCR as described in Materials and Methods. The value of lane 1 was set as 1. The expression levels of ZTA and YTHDF2 were monitored by WB using anti-ZTA and anti-Myc antibodies, respectively. β-Actin blot was included as loading control. (C) P3HR1 (EBV+) cells were used to create cell lines using pLenti-Vector, pLenti-YTHDF2-WT, and p-Lenti-YTHDF2-RRR. EBV lytic cycle was induced by TPA and NaBu for 48 h. The relative EBV copy numbers were measured using the qPCR as described in Materials and Methods. The value of lane 1 was set as 100. The expression of YTHDF2 was monitored by WB using anti-Myc antibody. β-Actin blot was included as loading control. Results from three biological replicates are presented. Error bars indicate standard deviations. *P < 0.05, **P < 0.01, ***P < 0.001. n.s., not significant; NaBu, sodium butyrate; TPA, tetradecanoyl phorbol acetate.
Fig 5 SUMOylation-deficient YTHDF2 impairs its ability to restrict EBV replication. (A) HEK293 (EBV+) cells were transfected with ZTA plasmid (lytic inducer), PIAS1, and WT YTHDF2 (WT) or SUMOylation-deficient YTHDF2 (RRR) as indicated. The relative EBV copy numbers were measured using the qPCR as described in the Materials and Methods. The value of lane 1 was set as 1. The expression levels of YTHDF2 and PIAS1 were monitored by WB. β-Actin blot was included as loading control. (B) Akata (EBV+) cells were used to create cell lines using pLenti-Vector, pLenti-YTHDF2-WT, and p-Lenti-YTHDF2-RRR (K281R/K571R/K572R). EBV lytic cycle was induced by anti-IgG-mediated B-cell receptor cross-linking. The relative EBV copy numbers were measured using the qPCR as described in Materials and Methods. The value of lane 1 was set as 1. The expression levels of ZTA and YTHDF2 were monitored by WB using anti-ZTA and anti-Myc antibodies, respectively. β-Actin blot was included as loading control. (C) P3HR1 (EBV+) cells were used to create cell lines using pLenti-Vector, pLenti-YTHDF2-WT, and p-Lenti-YTHDF2-RRR. EBV lytic cycle was induced by TPA and NaBu for 48 h. The relative EBV copy numbers were measured using the qPCR as described in Materials and Methods. The value of lane 1 was set as 100. The expression of YTHDF2 was monitored by WB using anti-Myc antibody. β-Actin blot was included as loading control. Results from three biological replicates are presented. Error bars indicate standard deviations. *P < 0.05, **P < 0.01, ***P < 0.001. n.s., not significant; NaBu, sodium butyrate; TPA, tetradecanoyl phorbol acetate.CC BY · PubMed Central
Fig 6 PIAS1 interacts with YTHDF2 in situ. Akata (EBV+) cells were blocked with 3% bovine serum albumin (BSA) in phosphate-buffered saline (PBS) at room temperature for 1 h, then incubated with PBS control (A) or a mixture of mouse anti-YTHDF2 and rabbit anti-PIAS1 antibodies (B). Then the probes were added for ligation and amplification. Cell nuclei were stained with DAPI and visualized using Nikon AXR. The red dots represent PLA signals indicating the interaction between YTHDF2 and PIAS1 in situ. DAPI, 4′,6-diamidino-2-phenylindole.
Fig 6 PIAS1 interacts with YTHDF2 in situ. Akata (EBV+) cells were blocked with 3% bovine serum albumin (BSA) in phosphate-buffered saline (PBS) at room temperature for 1 h, then incubated with PBS control (A) or a mixture of mouse anti-YTHDF2 and rabbit anti-PIAS1 antibodies (B). Then the probes were added for ligation and amplification. Cell nuclei were stained with DAPI and visualized using Nikon AXR. The red dots represent PLA signals indicating the interaction between YTHDF2 and PIAS1 in situ. DAPI, 4′,6-diamidino-2-phenylindole.CC BY · PubMed Central
Fig 7 SUMOylation of YTHDF2 enhances its binding to and degradation of EBV lytic transcripts. Akata (EBV+) carrying WT YTHDF2 or SUMOylation-deficient mutant (RRR) were lytically induced by IgG cross-linking for 24 h. (A) Cell lysate was collected to detect YTHDF2 binding of viral RNAs by RIP-qPCR. Values are fold change over 10% input. Cellular RNA MALAT1 was included as a negative control. (B–F) After lytic induction for 12 h, the cells were then treated with actinomycin D. The immediate early (ZTA and RTA), early (BALF5 and BGLF4), and late (BLLF1) gene levels were analyzed by quantitative reverse transcription-PCR. The relative mRNA level at 0 h after actinomycin D treatment was set as 1. Results from three biological replicates are presented. Error bars indicate standard deviations. *P < 0.05, **P < 0.01, ***P < 0.001. n.s., not significant.
Fig 7 SUMOylation of YTHDF2 enhances its binding to and degradation of EBV lytic transcripts. Akata (EBV+) carrying WT YTHDF2 or SUMOylation-deficient mutant (RRR) were lytically induced by IgG cross-linking for 24 h. (A) Cell lysate was collected to detect YTHDF2 binding of viral RNAs by RIP-qPCR. Values are fold change over 10% input. Cellular RNA MALAT1 was included as a negative control. (B–F) After lytic induction for 12 h, the cells were then treated with actinomycin D. The immediate early (ZTA and RTA), early (BALF5 and BGLF4), and late (BLLF1) gene levels were analyzed by quantitative reverse transcription-PCR. The relative mRNA level at 0 h after actinomycin D treatment was set as 1. Results from three biological replicates are presented. Error bars indicate standard deviations. *P < 0.05, **P < 0.01, ***P < 0.001. n.s., not significant.CC BY · PubMed Central
Fig 8 YTHDF1 and YTHDF3 are also SUMOylated by PIAS1. (A) Sequence alignment of human YTHDF2, YTHDF1 and YTHDF3 using COBALT. The SUMOylation sites corresponding to human YTHDF2 are highlighted in yellow. (B) PIAS1 promotes YTHDF1 and YTHDF3 SUMOylation. In vitro SUMOylation of YTHDF1 and YTHDF3 was performed with E1, E2, SUMO2 and PIAS1 as indicated. The reaction was terminated with SDS sample loading buffer, and WB was perfromed using anti-V5-HRP antibody. *, non-specific band. (C) PIAS1 interacts with YTHDF1 and YTHDF3. HEK293T cells were co-transfected with HA-PIAS1 and V5-YTHDF1, V5-YTHDF3, as well as V5-YTHDF2 as indicated. WB analysis showing that PIAS1 is Co-IPed with YTHDF1 and YTHDF3. Whole-cell lysates were blotted for PIAS1 and V5-YTHDFs as input. β-Actin blot was included as loading control. (D) Sequence alignment of YTHDF1 and YTHDF3 protein sequences from 10 species using COBALT. The positions of lysine corresponding to human YTHDF1 and YTHDF3 were labeled as indicated. The conserved lysines are highlighted in yellow. (E) In vitro SUMOylation assay was performed with the combination of E1, E2, SUMO2, PIAS1, and either WT or K277R/K551R mutant (RR) YTHDF1. The reaction was terminated with SDS sample loading buffer, and WB was performed using anti-V5-HRP antibody. (F) In vitro SUMOylation assay was performed with the combination of E1, E2, SUMO2, PIAS1, and either WT or K282R mutant YTHDF3. The reaction was terminated with SDS sample loading buffer and WB was performed using anti-V5-HRP antibody. COBALT, Constraint-based Multiple Alignment Tool.
Fig 8 YTHDF1 and YTHDF3 are also SUMOylated by PIAS1. (A) Sequence alignment of human YTHDF2, YTHDF1 and YTHDF3 using COBALT. The SUMOylation sites corresponding to human YTHDF2 are highlighted in yellow. (B) PIAS1 promotes YTHDF1 and YTHDF3 SUMOylation. In vitro SUMOylation of YTHDF1 and YTHDF3 was performed with E1, E2, SUMO2 and PIAS1 as indicated. The reaction was terminated with SDS sample loading buffer, and WB was perfromed using anti-V5-HRP antibody. *, non-specific band. (C) PIAS1 interacts with YTHDF1 and YTHDF3. HEK293T cells were co-transfected with HA-PIAS1 and V5-YTHDF1, V5-YTHDF3, as well as V5-YTHDF2 as indicated. WB analysis showing that PIAS1 is Co-IPed with YTHDF1 and YTHDF3. Whole-cell lysates were blotted for PIAS1 and V5-YTHDFs as input. β-Actin blot was included as loading control. (D) Sequence alignment of YTHDF1 and YTHDF3 protein sequences from 10 species using COBALT. The positions of lysine corresponding to human YTHDF1 and YTHDF3 were labeled as indicated. The conserved lysines are highlighted in yellow. (E) In vitro SUMOylation assay was performed with the combination of E1, E2, SUMO2, PIAS1, and either WT or K277R/K551R mutant (RR) YTHDF1. The reaction was terminated with SDS sample loading buffer, and WB was performed using anti-V5-HRP antibody. (F) In vitro SUMOylation assay was performed with the combination of E1, E2, SUMO2, PIAS1, and either WT or K282R mutant YTHDF3. The reaction was terminated with SDS sample loading buffer and WB was performed using anti-V5-HRP antibody. COBALT, Constraint-based Multiple Alignment Tool.CC BY · PubMed Central
Fig 9 PIAS1-mediated SUMOylation of YTHDF1 and YTHDF3 restrict EBV lytic replication. (A) Akata (EBV+) cells were used to create cell lines using pLenti-Vector, pLenti-YTHDF1-WT, and p-Lenti-YTHDF1-RR (K277R/K551R). (B) Akata (EBV+) cells were used to create cell lines using pLenti-Vector, pLenti-YTHDF3-WT, and p-Lenti-YTHDF3-R (K282R). EBV lytic cycle was induced by anti-IgG-mediated B-cell receptor cross-linking. The relative EBV copy numbers were measured using the qPCR as described in Material and Methods. The value of lane 1 was set as 1. The expression levels of ZTA and YTHDF proteins were monitored by WB using anti-ZTA and anti-Myc antibodies, respectively. β-Actin blot was included as loading control. Results from three biological replicates are presented. Error bars indicate standard deviations. **P < 0.01, ***P < 0.001.
Fig 9 PIAS1-mediated SUMOylation of YTHDF1 and YTHDF3 restrict EBV lytic replication. (A) Akata (EBV+) cells were used to create cell lines using pLenti-Vector, pLenti-YTHDF1-WT, and p-Lenti-YTHDF1-RR (K277R/K551R). (B) Akata (EBV+) cells were used to create cell lines using pLenti-Vector, pLenti-YTHDF3-WT, and p-Lenti-YTHDF3-R (K282R). EBV lytic cycle was induced by anti-IgG-mediated B-cell receptor cross-linking. The relative EBV copy numbers were measured using the qPCR as described in Material and Methods. The value of lane 1 was set as 1. The expression levels of ZTA and YTHDF proteins were monitored by WB using anti-ZTA and anti-Myc antibodies, respectively. β-Actin blot was included as loading control. Results from three biological replicates are presented. Error bars indicate standard deviations. **P < 0.01, ***P < 0.001.CC BY · PubMed Central
Fig 10 Model summarizing PIAS1-mediated SUMOylation of YTHDF family proteins in control of EBV latency and reactivation.
Fig 10 Model summarizing PIAS1-mediated SUMOylation of YTHDF family proteins in control of EBV latency and reactivation.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Research works printed but not linked: 2

  1. 10.1101/2021.10.04.463034:2021.10.04.463034 · DOI
  2. 10.1145/3093338.3093390 · DOI

Metadata

Authors
Sugiokto, F. G., Saiada, F., Zhang, K., & Li, R.
Affiliations
  • University of Pittsburgh (US)
  • Virginia Commonwealth University (US)
Corresponding author
Li, Renfeng
ORCID
Venue
mBio
Publisher
American Society for Microbiology
Published
2024-02-14
Volume
15
Issue
2
Pages
e0316823-e0316823
ISSN
2150-7511 (ISSN-L)
DOI
10.1128/mbio.03168-23
PMC
PMC10865817
OpenAlex
W4390973646
Cited by
23
References
48
Citations by year
2026: 7 · 2025: 7 · 2024: 9
FWCI
3.12
Type
journal article
Language
en
Open access
Licence
Locations
  • mBio (publishedVersion, cc-by)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nc-nd)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (submittedVersion, cc-by-nc-nd)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • HHS | NIH | National Institute of Allergy and Infectious Diseases (AI104828, AI141410)
  • American Cancer Society (134703-RSG-20-054-01-MPC)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › RNA modifications and cancer
  • Health Sciences › Medicine › Oncology › Viral-associated cancers and disorders
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › RNA Research and Splicing
Keywords
YTHDF2, PIAS1, SUMOylation, Epstein-Barr virus, m6A RNA methylation, YTHDF1, YTHDF3, viral restriction factors, RNA-binding proteins

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.

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.