Cited by 175Single-cell and spatial transcriptomicsNeuroscience and Neuropharmacology ResearchNeurotransmitter Receptor Influence on Behavior

The DOI matches its Crossref record; 79 of the 79 references it cites were found in a registry or an index; 2 of the 72 sources its claims cite were checked for support.

Cell type-specific transcriptional programs in mouse prefrontal cortex during adolescence and addiction

Open access

Abstract

Coordinated activity-induced transcriptional changes across multiple neuron subtypes of the prefrontal cortex (PFC) play a pivotal role in encoding and regulating major cognitive behaviors. Yet, the specific transcriptional programs in each neuron subtype remain unknown. Using single-cell RNA sequencing (scRNA-seq), here we comprehensively classify all unique cell subtypes in the PFC. We analyze transcriptional dynamics of each cell subtype under a naturally adaptive and an induced condition. Adaptive changes during adolescence (between P21 and P60), a highly dynamic phase of postnatal neuroplasticity, profoundly impacted transcription in each neuron subtype, including cell type-specific regulation of genes implicated in major neuropsychiatric disorders. On the other hand, an induced plasticity evoked by chronic cocaine addiction resulted in progressive transcriptional changes in multiple neuron subtypes and became most pronounced upon prolonged drug withdrawal. Our findings lay a foundation for understanding cell type-specific postnatal transcriptional dynamics under normal PFC function and in neuropsychiatric disease states.

Overview

AI snapshot

Written from the paper’s full text.

TL;DR

Using single-cell RNA sequencing, this study classifies cell types and subtypes in mouse prefrontal cortex and maps their transcriptional dynamics during adolescence (P21–P60) and chronic cocaine self-administration/withdrawal. It reports cell-type-specific changes, including regulation of neuropsychiatric disease-relevant GWAS genes and distinct pathways for adaptive versus cocaine-induced plasticity.

Key findings

  • The PFC contained 8 major cell clusters; excitatory neurons made up 52.3% and inhibitory neurons 4.3% of total cells. Excitatory neurons resolved into 13 broad subtypes (26 at higher resolution) and inhibitory neurons into 12 subtypes; PFC excitatory subtypes were transcriptionally distinct from VISp and ALM, with unique subtypes and 273–555 differentially expressed genes even in similar clusters (FC > 2, q < 0.05).
  • During adolescence (P21→P60), expression of at least 338 genes (Exc-1) to 1080 genes (Exc-8) changed across excitatory clusters; inhibitory changes ranged from 6 genes (Inhib-3) to 117 genes (Inhib-7), and non-neuronal endothelial cells had 879 genes affected.
  • Chronic cocaine IVSA produced little transcriptional change during maintenance (≤10 genes per excitatory subtype, except Exc-6 with >40), but effects increased after withdrawal; at 15-day withdrawal all excitatory subtypes were impacted, with Exc-2, 6, 7, and 8 showing the greatest effects. Oxidative phosphorylation, mitochondrial function, and sirtuin signaling were the most significantly affected pathways.
  • GWAS candidate genes for 12 PFC-related disorders showed cell type/subtype-enriched expression; 26–36% of GWAS candidates were dynamically regulated during adolescence for each disorder. For schizophrenia, Exc-13 had the most altered neuronal GWAS candidates; for bipolar disorder, Exc-8 had the most, followed by Exc-13; endothelial cells were the most affected non-neuronal type.

Methods

  • Single-cell RNA-seq (10X Chromium) of mouse PFC: P60 for cell-type classification, P21 vs P60 for adolescence, and cocaine IVSA (saline vs cocaine) with PFC collected at maintenance, 48 h withdrawal, and 15-day withdrawal. All mice were male; 12 independent biological replicates were used, each pooling PFC from two mice; P21 data used 4 mice, and cocaine IVSA used 12 mice per group.
  • Analyses included Seurat clustering, CCA/t-SNE alignment, differential expression (negative binomial GLM, likelihood ratio test, SC2P), MetaNeighbor for cross-region comparison, IPA pathway enrichment, and GWAS catalog enrichment; findings were validated with RNA in situ hybridization and smFISH.

Limitations

  • Stated: low inhibitory neuron abundance limits statistical power and makes cocaine effects in inhibitory subtypes difficult to interpret; the age difference between adolescence (P21) and cocaine starting at P60 may contribute to differences; many factors modulate drug self-administration, and further studies are needed.
  • Stated: although 26 high-resolution excitatory clusters were generated, differential expression used the 13 lower-resolution clusters for better power and accuracy; a marginal role of DNA methylation in early P21→P60 changes cannot be fully ruled out.
  • Evident from the text: all mice were male, so sex differences are not addressed; the study is transcriptomic and correlational in mouse and does not report causal manipulations of candidate genes or subtypes.

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

Integrity signals

Registry

Crossref · 10.1038/s41467-019-12054-3

Cites

79 of 79 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1 Transcriptome-based cell classification in mouse PFC. a Schematic of the experimental design. PFC sections (red outline) were isolated for scRNA-Seq. Created with BioRender.com. b t-SNE plot showing the broad clustering of PFC cell types based on transcriptome. c Expression of cell type-specific markers in each broad cell cluster color-highlighted on t-SNE plots. d Single cell heat map showing cell type-specific gene marker expression in the different PFC cell clusters
Fig. 1 Transcriptome-based cell classification in mouse PFC. a Schematic of the experimental design. PFC sections (red outline) were isolated for scRNA-Seq. Created with BioRender.com. b t-SNE plot showing the broad clustering of PFC cell types based on transcriptome. c Expression of cell type-specific markers in each broad cell cluster color-highlighted on t-SNE plots. d Single cell heat map showing cell type-specific gene marker expression in the different PFC cell clustersCC BY · PubMed Central
Fig. 2 PFC contains distinct excitatory and inhibitory neuron subtypes. a t-SNE plot showing that excitatory neurons of PFC can be broadly classified into 13 unique subtypes based on their transcriptome. b Violin plot showing expression of specific markers for each of the 13 excitatory neuron clusters or subtypes. c Multi-channel FISH detecting distinct excitatory neuron types within the same cortical layer (Pou3f1 and Tshz2 in L5) or across different layers (Foxp2 in L6). d Assigning excitatory neurons to respective cortical layers by projecting expression of layer-specific markers onto the t-SNE plot. e Identification of two distinct cell types (Pou3f1+ and Tshz2+) in L5 that commonly express L5 marker Etv1. f t-SNE plot showing classification of PFC inhibitory neurons into 12 distinct subtypes based on their transcriptome. g Violin plot showing distinct markers for each of the 12 different subtypes of inhibitory neurons. h Multi-channel FISH detecting subtypes within a known inhibitory neuron subpopulation (Cartpt+, arrow, and Cartpt−, arrowhead within Sst neurons: enlarged view of single cells in box area shown in side panel) and a rare neuron subtypes (Pbx3+: inset is an enlarged view of single cell)
Fig. 2 PFC contains distinct excitatory and inhibitory neuron subtypes. a t-SNE plot showing that excitatory neurons of PFC can be broadly classified into 13 unique subtypes based on their transcriptome. b Violin plot showing expression of specific markers for each of the 13 excitatory neuron clusters or subtypes. c Multi-channel FISH detecting distinct excitatory neuron types within the same cortical layer (Pou3f1 and Tshz2 in L5) or across different layers (Foxp2 in L6). d Assigning excitatory neurons to respective cortical layers by projecting expression of layer-specific markers onto the t-SNE plot. e Identification of two distinct cell types (Pou3f1+ and Tshz2+) in L5 that commonly express L5 marker Etv1. f t-SNE plot showing classification of PFC inhibitory neurons into 12 distinct subtypes based on their transcriptome. g Violin plot showing distinct markers for each of the 12 different subtypes of inhibitory neurons. h Multi-channel FISH detecting subtypes within a known inhibitory neuron subpopulation (Cartpt+, arrow, and Cartpt−, arrowhead within Sst neurons: enlarged view of single cells in box area shown in side panel) and a rare neuron subtypes (Pbx3+: inset is an enlarged view of single cell)CC BY · PubMed Central
Fig. 3 Distinct transcriptional features of PFC excitatory neurons relative to those of VISp and ALM. a t-SNE plot (left) showing high resolution clustering of PFC excitatory neurons and the dendrogram of hierarchical clustering (right). b Violin plot showing the distinct markers for each of the clusters in a. c Heat map showing the degree of similarity [0–100] between the excitatory neuron subtypes in PFC, primary visual cortex (VISp), and anterolateral motor (ALM) based on similarity of transcriptional profiles. Subtypes showing a similarity ≥ 90% were clustered together into blocks. Subpopulations showing <90% similarity to any of the subpopulation from the other dataset were considered as region specific (shown with bold font and *). VISp and ALM cluster names were abbreviated, the original corresponding names are available in Supplementary Data 3
Fig. 3 Distinct transcriptional features of PFC excitatory neurons relative to those of VISp and ALM. a t-SNE plot (left) showing high resolution clustering of PFC excitatory neurons and the dendrogram of hierarchical clustering (right). b Violin plot showing the distinct markers for each of the clusters in a. c Heat map showing the degree of similarity [0–100] between the excitatory neuron subtypes in PFC, primary visual cortex (VISp), and anterolateral motor (ALM) based on similarity of transcriptional profiles. Subtypes showing a similarity ≥ 90% were clustered together into blocks. Subpopulations showing <90% similarity to any of the subpopulation from the other dataset were considered as region specific (shown with bold font and *). VISp and ALM cluster names were abbreviated, the original corresponding names are available in Supplementary Data 3CC BY · PubMed Central
Fig. 4 Widespread transcriptional changes in PFC cell types between P21 and P60 mice. a t-SNE plot showing the distribution of the merged cells from P60 (purple) and P21 (light-blue) PFC cells after alignment using correlation analysis (CCA). b t-SNE plot showing the cell-type assignment and proportion of the cell-types in the P60 (left) and P21 (right) PFC cells. c Number of genes dynamically changed in excitatory neurons, inhibitory neurons and non-neuronal cells. Red: upregulated genes; Blue: downregulated genes. d Representative volcano plots showing altered gene expression in the indicated clusters (cutoff: FC > 1.5 and q-value < 0.05). e Heatmap showing the -log10(p-value) (right-tailed Fisher exact test) of the functional pathways impacted in excitatory neuron subtypes based on gene expression changes between P21 and P60. f Gene names and functional categories of epigenetic modifiers up-regulated (red) and down-regulated (blue) in each PFC excitatory neuron subtypes between P21 and P60 (cutoff: FC > 1.5 and q-value < 0.05; numbers 1–13 on left axis indicate the clusters). The differential gene expression p-values were calculated using the negative binomial generalized linear model and the q-values were derived using the Bonferroni method
Fig. 4 Widespread transcriptional changes in PFC cell types between P21 and P60 mice. a t-SNE plot showing the distribution of the merged cells from P60 (purple) and P21 (light-blue) PFC cells after alignment using correlation analysis (CCA). b t-SNE plot showing the cell-type assignment and proportion of the cell-types in the P60 (left) and P21 (right) PFC cells. c Number of genes dynamically changed in excitatory neurons, inhibitory neurons and non-neuronal cells. Red: upregulated genes; Blue: downregulated genes. d Representative volcano plots showing altered gene expression in the indicated clusters (cutoff: FC > 1.5 and q-value < 0.05). e Heatmap showing the -log10(p-value) (right-tailed Fisher exact test) of the functional pathways impacted in excitatory neuron subtypes based on gene expression changes between P21 and P60. f Gene names and functional categories of epigenetic modifiers up-regulated (red) and down-regulated (blue) in each PFC excitatory neuron subtypes between P21 and P60 (cutoff: FC > 1.5 and q-value < 0.05; numbers 1–13 on left axis indicate the clusters). The differential gene expression p-values were calculated using the negative binomial generalized linear model and the q-values were derived using the Bonferroni methodCC BY · PubMed Central
Fig. 5 Expression of many GWAS candidates is enriched in specific clusters and dynamically changed during adolescence. a t-SNE plot showing examples of some cell type- and subtype-specific expression of GWAS candidate genes in schizophrenia. Top-panel shows the gene expression enrichment of four representative candidates on the global t-SNE plot. The bottom panels show zoomed t-SNE of excitatory and inhibitory clusters, respectively, and Allen Brain ISH images showing the specific gene expression. b Similar representations for bipolar disorder as shown for schizophrenia in a. c Heatmap showing GWAS candidate disease-relevant genes with cell subtype-specific expression. The blue color indicates no enrichment, the orange color indicates gene enrichment. Schizophrenia and Bipolar disorders are shown as examples. d Circular heatmap showing the percent of enriched subtype- and cell type-specific GWAS candidate genes in each of the 12 PFC-relevant diseases. Each disease is scaled individually, bright red: high enrichment, dark blue: depletion. e t-SNE-plot indicating the number of differentially expressed schizophrenia and bipolar GWAS candidates per cluster (shown in distinct colors) between P21 and P60 in each cell subtypes
Fig. 5 Expression of many GWAS candidates is enriched in specific clusters and dynamically changed during adolescence. a t-SNE plot showing examples of some cell type- and subtype-specific expression of GWAS candidate genes in schizophrenia. Top-panel shows the gene expression enrichment of four representative candidates on the global t-SNE plot. The bottom panels show zoomed t-SNE of excitatory and inhibitory clusters, respectively, and Allen Brain ISH images showing the specific gene expression. b Similar representations for bipolar disorder as shown for schizophrenia in a. c Heatmap showing GWAS candidate disease-relevant genes with cell subtype-specific expression. The blue color indicates no enrichment, the orange color indicates gene enrichment. Schizophrenia and Bipolar disorders are shown as examples. d Circular heatmap showing the percent of enriched subtype- and cell type-specific GWAS candidate genes in each of the 12 PFC-relevant diseases. Each disease is scaled individually, bright red: high enrichment, dark blue: depletion. e t-SNE-plot indicating the number of differentially expressed schizophrenia and bipolar GWAS candidates per cluster (shown in distinct colors) between P21 and P60 in each cell subtypesCC BY · PubMed Central
Fig. 6 Chronic cocaine IVSA induces transcriptional changes in multiple PFC cell types. a Schematic diagram showing the IVSA experimental setting and workflow. Created with BioRender.com. b Reward earnt by lever pressing in IVSA mice (saline and cocaine groups) through acquisition and maintenance of a 15-day period. c t-SNE-plot showing the uniform distribution of cocaine and saline samples in the different cell-types. d Dot-plot showing the number of differentially expressed genes between cocaine and saline in each excitatory neuron subtypes at each time point (cutoff: FC > 1.5 and SC2P model phase2 FDR < 0.05, empirical Bayes statistic, limma package). e Heatmap showing the percent of commonly differentially expressed genes between the different clusters. Clusters tend to show cluster-specific differentially expressed genes with the highest overlap observed around 30% during the 15-day withdrawal. f Heatmap showing the -log10(p-value) of the pathways enriched for differentially expressed genes in the different clusters at the 15-day withdrawal
Fig. 6 Chronic cocaine IVSA induces transcriptional changes in multiple PFC cell types. a Schematic diagram showing the IVSA experimental setting and workflow. Created with BioRender.com. b Reward earnt by lever pressing in IVSA mice (saline and cocaine groups) through acquisition and maintenance of a 15-day period. c t-SNE-plot showing the uniform distribution of cocaine and saline samples in the different cell-types. d Dot-plot showing the number of differentially expressed genes between cocaine and saline in each excitatory neuron subtypes at each time point (cutoff: FC > 1.5 and SC2P model phase2 FDR < 0.05, empirical Bayes statistic, limma package). e Heatmap showing the percent of commonly differentially expressed genes between the different clusters. Clusters tend to show cluster-specific differentially expressed genes with the highest overlap observed around 30% during the 15-day withdrawal. f Heatmap showing the -log10(p-value) of the pathways enriched for differentially expressed genes in the different clusters at the 15-day withdrawalCC BY · PubMed Central

Verification

Claims the paper makes

136 Own result · 54 Not checked yet · 43 Own interpretation · 17 Own method · 7 No basis given · 1 Supported · 1 Partly supported · accuracy not yet measured

Verify all 70 references — free account

Show all 127 claims

  • Probably not a substantial claim. 56
  • It may state two separate claims as one. 92
132 of 132
  • The paper relies on this without support.

    Coordinated activity-induced transcriptional changes across multiple neuron subtypes of the PFC play a pivotal role in encoding and regulating major cognitive behaviors.

    “Coordinated activity-induced transcriptional changes across multiple neuron subtypes of the prefrontal cortex (PFC) play a pivotal role in encoding and regulating major cognitive behaviors.”
  • The paper relies on this without support.

    The cellular mechanisms underlying the dysfunctions caused by early and later life events are largely unknown.

    “However, the cellular mechanisms underlying these dysfunctions are largely unknown.”
  • The paper relies on this without support.

    The profound cellular heterogeneity of the PFC has hindered the study of cell-type-specific transcriptional dynamics.

    “However, the profound cellular heterogeneity of the PFC has hindered the study of cell-type-specific transcriptional dynamics.”
  • The paper relies on this without support.

    The PFC is comprised of 4 layers (L1, L2/3, L5, and L6) and does not contain a L4.

    “The PFC is comprised of 4 layers (L) L1, L2/3, L5, and L6 (but does not contain a L4).”
  • The paper relies on this without support.

    Mice are weaned from maternal care at P21 and begin exploring the environment on their own, and are considered as adult at P60.

    “Mice are weaned from maternal care at P21 and begin exploring the environment on their own, and are considered as adult at P60.”
  • The paper relies on this without support.

    Widespread change in transcription is often associated with broad chromatin and epigenetic regulations.

    “Widespread change in transcription is often associated with broad chromatin and epigenetic regulations.”
  • The paper relies on this without support.

    Altered histone modifiers can regulate gene expression either positively or negatively.

    “The altered histone modifiers can regulate gene expression either positively or negatively.”
  • The paper relies on this without support.

    Transcriptional regulation by Chd8 has been associated with neuronal function.

    “Transcriptional regulation by Chd8 has been associated with neuronal function.”
  • The paper relies on this without support.

    The PFC plays a central role in the pathophysiology of several neuropsychiatric disorders, including schizophrenia, bipolar disorder, psychosis, mania, depression or suicidal tendencies.

    “The PFC is believed to play a central role in the pathophysiology of several neuropsychiatric disorders, including schizophrenia, bipolar disorder, psychosis, mania, depression or suicidal tendencies.”
  • The paper relies on this without support.

    The cellular and molecular mechanisms underlying involvement of GWAS candidate genes in neuropsychiatric disorders remain largely unknown.

    “However, the cellular and molecular mechanisms underlying their involvement remain largely unknown.”
  • The paper relies on this without support.

    A causal relationship between a gene function and a disorder, as well as the underlying mechanism, is difficult to establish.

    “However, a causal relationship between a gene function and a disorder, as well as the underlying mechanism is difficult to establish.”
  • The paper relies on this without support.

    Many psychiatric disorders may emerge from dysfunctions of non-neuronal cells, which cause disruptions of major support systems like myelination or blood-brain barrier.

    “It is now appreciated that many psychiatric disorders may emerge from dysfunctions of non-neuronal cells, which cause disruptions of major support systems like myelination or blood-brain barrier.”
  • The paper relies on this without support.

    Adolescence is a major landmark of postnatal PFC maturation.

    “Adolescence is known to be a major landmark of postnatal PFC maturation.”
  • The paper relies on this without support.

    Which type of cells in PFC exhibit transcriptional dynamics during adolescence, and by what means, have remained unknown.

    “Yet which type of cells in PFC exhibit transcriptional dynamics during this period, and by what means, have remained unknown.”
  • The paper relies on this without support.

    Less is known about cell type-specific regulation of the neuronal epigenome in later postnatal life.

    “Less is known about cell type-specific regulation of the neuronal epigenome in later postnatal life.”
  • The paper relies on this without support.

    Induced plasticity should mainly impact neurons within the circuits that respond to the specific stimulus.

    “Unlike global adaptations that may occur in all cells of a neuron subtype during development, induced plasticity should mainly impact neurons within the circuits that respond to the specific stimulus.”
  • The paper relies on this without support.

    Abnormally high mitochondrial mRNA is an indication of dead or dying cells.

    “Abnormally high mitochondrial mRNA is an indication of dead or dying cells.”
  • The paper relies on this without support.

    Incorporation of cells with abnormally high mitochondrial mRNA into the analysis can skew the results and introduce errors in detection of differentially expressed genes.

    “Incorporation of these cells into the analysis can skew the results and introduce errors in detection of differentially expressed genes.”
  • The paper relies on this without support.

    Conventional correlation analysis can lead to inconsistent results when two datasets are sequenced at different depths.

    “Since the two datasets were sequenced at different depths, conventional correlation analysis can lead to inconsistent results.”
  • The paper relies on this without support.

    Canonical correlation analysis (CCA) can only present cells into another projection space that brings similar cells into near proximity; it does not correct the gene expression between two datasets.

    “As the canonical correlation analysis (CCA) can only present cells into another projection space that brings similar cells into near proximity, it does not correct the gene expression between two datasets.”
  • The paper relies on this without support.

    The second criterion (cells expressing a gene concentrated in a subset of clusters) enables detection of potential false-positive genes that are broadly expressed by showing variability in their expression along cell-types.

    “The second criteria enables us to detect potential false-positive genes that broadly expressed by showing a variability in their expression along the cell-types.”
  • The authors state this as an assumption.

    The two-phase transcription model assumes that cells in a homogeneous population respond non-homogeneously to a stimulus.

    “To detect the genes affected by cocaine in each cluster, we used a “Two-phase transcription” model (SC2P package43), which assumes that cells in a homogeneous population response non-homogeneously to a stimulus.”
22 of 22

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?

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

Show all 79 references

Research works printed but not linked: 1

  1. Crow, M., Paul, A., Ballouz, S., Huang, Z. J. & Gillis, J. Characterizing the replicability of cell types defined by single cell RNA-sequencing data using MetaNeighbor. Nature. Communications 9, 884 (2018).

Cited by

Sorted by citations

See all 175

Metadata

Authors
Bhattacherjee, A., Djekidel, M. N., Chen, R., Chen, W., Tuesta, L. M., & Zhang, Y.
Affiliations
Boston Children's Hospital (US)
Corresponding author
Zhang, Yi
Venue
Nature Communications
Published
2019-09-13
Volume
10
Issue
1
Pages
4169-4169
ISSN
2041-1723 (ISSN-L)
OpenAlex
W2972625781
Cited by
174
References
79
Citations by year
2026: 21 · 2025: 32 · 2024: 33 · 2023: 32 · 2022: 25 · 2021: 20 · 2020: 9 · 2019: 1
FWCI
4.55
Type
journal article
Language
en
Open access
Locations
  • Nature Communications (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
Howard Hughes Medical Institute
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Single-cell and spatial transcriptomics
  • Life Sciences › Neuroscience › Cellular and Molecular Neuroscience › Neuroscience and Neuropharmacology Research
  • Life Sciences › Neuroscience › Cellular and Molecular Neuroscience › Neurotransmitter Receptor Influence on Behavior
Keywords
prefrontal cortex, neuronal cell types, cell-type-specific transcription, adolescence, cocaine dependence, drug withdrawal, transcriptional dynamics, neuropsychiatric disorders, single-cell transcriptomics
MeSH terms
Animals, Male, Neurosciences, Behavior, Addictive, Prefrontal Cortex, Sequence Analysis, RNA, Mice

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 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

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