CoChat
Sign Up
Cited by 18Genetic Associations and EpidemiologyCognitive Abilities and TestingGenetic Mapping and Diversity in Plants and Animals

The DOI matches its Crossref record; 69 of the 69 references it cites were found in a registry or an index; 54 of the 59 sources its claims cite were checked for support.

Shared genetic architectures of educational attainment in East Asian and European populations

Open access
Chen, Tzu-Ting, Kim, Jaeyoung, Lam, Max, Chuang, Yi-Fang, Chiu, Yen-Ling, Lin, Shu-Chin, Jung, Sang-Hyuk, Kim, Beomsu, Kim, Soyeon, Cho, Chamlee, Shim, Injeong, Park, Sanghyeon, Ahn, Yeeun, Okbay, Aysu, Jang, Hyemin, Kim, Hee Jin, Seo, Sang Won, Park, Woong-Yang, Ge, Tian, Huang, Hailiang, Feng, Yen-Chen Anne, Lin, Yen-Feng, Myung, Woojae, Chen, Chia-Yen, Won, Hong-Hee·2024·Nature Human Behaviour
DOI 10.1038/s41562-023-01781-9PMC PMC10963262
Open this paper in CoChat — free accountOpen PDF

Abstract

Educational attainment (EduYears), a heritable trait often used as a proxy for cognitive ability, is associated with various health and social outcomes. Previous genome-wide association studies (GWASs) on EduYears have been focused on samples of European (EUR) genetic ancestries. Here we present the first large-scale GWAS of EduYears in people of East Asian (EAS) ancestry (n = 176,400) and conduct a cross-ancestry meta-analysis with EduYears GWAS in people of EUR ancestry (n = 766,345). EduYears showed a high genetic correlation and power-adjusted transferability ratio between EAS and EUR. We also found similar functional enrichment, gene expression enrichment and cross-trait genetic correlations between two populations. Cross-ancestry fine-mapping identified refined credible sets with a higher posterior inclusion probability than single population fine-mapping. Polygenic prediction analysis in four independent EAS and EUR cohorts demonstrated transferability between populations. Our study supports the need for further research on diverse ancestries to increase our understanding of the genetic basis of educational attainment.

Overview

AI snapshot

Written from the paper’s full text.

TL;DR

This study presents the first large-scale educational attainment (EduYears) GWAS in East Asian ancestry (n = 176,400) and a cross-ancestry meta-analysis with European ancestry (n = 766,345). It finds high genetic correlation and transferability between EAS and EUR, shared functional and pathway enrichments, and improved fine-mapping and polygenic prediction from cross-population analysis.

Key findings

  • The EAS EduYears GWAS meta-analysis identified 7 genome-wide significant loci with 11 independent SNPs, all previously reported; SNP-based heritability was 9.7% in TWB, 8.7% in KoGES and 9.0% in EAS, with within-EAS genetic correlation 0.871 (s.e. 0.073).
  • Cross-ancestry meta-analysis (942,745 samples; 12,232,310 variants) identified 315 lead SNPs at 102 loci; MAMA EAS-specific analysis found 94 independent genome-wide significant SNPs, 2 previously unreported (rs2881903 and rs16930687). EAS–EUR genetic correlation was 0.873 (s.e. 0.042), and the EUR-to-EAS power-adjusted transferability ratio was 0.62 (95/153).
  • Functional analyses showed similar enrichment: 13 genes mapped via cis-eQTL, 16 EAS and 27 EUR pathways significantly associated with EduYears, with 14 shared; cross-population fine-mapping increased maximum PIP, for example at the chromosome 17 locus from 0.201 in EAS to 0.708 cross-population.
  • Cross-population PGS from PRS-CSx explained up to 4.0% of EduYears phenotypic variance in EAS testing cohorts and 6.1% in the EUR NIA-LOAD cohort, outperforming single-population PGS in EAS cohorts.

Methods

  • GWAS was performed in Taiwan Biobank (104,722 samples) and Korean Genome and Epidemiology Study (71,678 samples) after stringent QC and imputation, then meta-analysed in EAS (n = 176,400) using Regenie and METAL fixed-effect models.
  • Cross-ancestry meta-analysis combined EAS with publicly available EUR summary statistics from Lee et al. (n = 766,345) using METAL and MAMA; downstream analyses included LDSC/S-LDXR, PAT ratio, SuSiEx fine-mapping, FUMA/MAGMA, GSA-SNP2 and PRS-CS/PRS-CSx.
  • Polygenic prediction used independent testing cohorts: EMCIT (n = 395), Korean-based cohort (n = 2,622), Chinese samples in UK Biobank (n = 1,747) and NIA-LOAD (n = 1,241).

Limitations

  • The paper states that self-reported educational categories may not equal actual education years, and differences in compulsory education between Taiwan and South Korea may limit phenotypic variation.
  • The paper states that the EAS sample is much smaller than the EUR GWAS, yielding only 7 loci versus 3,952 in EUR and all 7 previously reported, reflecting lower power for gene discovery.
  • The paper notes that cross-ancestry meta-analysis used Lee et al. EUR data (n = 766,345) rather than the larger Okbay et al. sample due to access limitations; fine-mapping used external 1KG reference panels whose LD may not perfectly match discovery GWAS, though LD consistency was assessed (s values 0.007–0.027). The paper also states that results do not support immediate medical or clinical applications.

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

Integrity signals

Registry

Crossref · 10.1038/s41562-023-01781-9

Cites

69 of 69 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

54 of 59 cited sources checked · 23 supported · 0 flagged · accuracy not yet measured

  • Findingown data

    Using the Cahoy dataset, EduYears-associated SNPs were more enriched in neurons than astrocytes or oligodendrocytes in the EAS population.

    “By using the Cahoy dataset27 to examine the enrichment in three brain cell types (neurons, astrocytes and oligodendrocytes), we identified that EduYears-associated SNPs were more enriched in neurons than astrocytes or oligodendrocytes in the EAS population (Supplementary Table 12).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [27] Cahoy, J. D. et al. A transcriptome database for astrocytes, neurons, and oligodendrocytes: a new resource for understanding brain development and function. J. Neurosci. 28, 264–278 (2008). source
  • Interpretationown data

    PGS analyses indicated the advantages of ancestral diversity over a single population in PGS construction.

    “Furthermore, our PGS analyses also indicated the advantages of ancestral diversity over a single population in PGS construction36.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [36] Ruan, Y. et al. Improving polygenic prediction in ancestrally diverse populations. Nat. Genet. 54, 573–580 (2022). source
  • Findingown data

    In the EAS population, SNP heritability was significantly enriched in the central nervous system, including the foetal brain, dorsolateral prefrontal cortex and inferior temporal lobe after FDR correction, consistent with previous EUR results.

    “In the EAS population, SNP heritability was significantly enriched in the central nervous system, including the foetal brain, dorsolateral prefrontal cortex and inferior temporal lobe after FDR correction (Supplementary Fig. 9b and Supplementary Table 11), which is also consistent with previous results in EUR14.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [14] Lee, J. J. et al. Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals. Nat. Genet. 50, 1112–1121 (2018). source
  • Findingcitation

    The result that cross-population PGS outperformed ancestry-matched PGSs replicated previous studies showing that multi-ancestry PGS demonstrated improved prediction performance relative to ancestry-matched PGSs.

    “This result replicated previous studies showing that multi-ancestry PGS demonstrated an improved prediction performance relative to ancestry-matched PGSs41,42.”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [41] Graham, S. E. et al. The power of genetic diversity in genome-wide association studies of lipids. Nature 600, 675–679 (2021). source

      The ‘Trans-ancestry PRS are most predictive’ section reports that trans-ancestry scores had the best or near-best performance across the groups tested. It describes prediction relative to ancestry-matched scores as ‘improved or comparable,’ so the claim overstates the finding by presenting improvement without that qualification. (LLM, conf=0.84)

      Experimental — accuracy not yet measured
    • cites [42] Kim, S. et al. Shared genetic architectures of subjective well-being in East Asian and European ancestry populations. Nat. Hum. Behav. 6, 1014–1026 (2022). source
  • Premisecitation

    Lack of diversity in genetic studies on EduYears could lead to social and health disparities due to inadequate comprehension of EduYears and its impact on socioeconomic and health outcomes in understudied populations.

    “The lack of diversity in genetic studies on EduYears could lead to social and health disparities due to inadequate comprehension of EduYears and its impact on socioeconomic and health outcomes in understudied populations15.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [15] Fatumo, S. et al. A roadmap to increase diversity in genomic studies. Nat. Med. 28, 243–250 (2022). source

      The ‘Clinical unmet needs and genetic diversity imbalances in genomics’ section says Eurocentric study biases limit understanding of population risk factors and warns that genomic translation could exacerbate health disparities. It does not address EduYears, its effects on socioeconomic or health outcomes, or a disparity arising specifically from inadequate understanding of that trait (searched: EduYears, educational attainment, socioeconomic outcomes). (LLM, conf=0.88)

      Experimental — accuracy not yet measured
  • It may state two separate claims as one. 28
  • Probably not a substantial claim. 22

  • It may state two separate claims as one.

    Study design and conductown data

    LD consistency between discovery and reference samples in the EAS population was high; s values from SuSiE-RSS ranged from 0.007 to 0.027 across fine-mapped loci.

    “The s values from SuSiE-RSS ranged from 0.007 to 0.027 across fine-mapped loci, and the diagnostic plots demonstrated high consistency of LD between the summary statistics and reference panel (Supplementary Fig. 19a–g).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [34] Zou, Y., Carbonetto, P., Wang, G. & Stephens, M. Fine-mapping from summary data with the “Sum of Single Effects” model. PLoS Genet. 18, e1010299 (2022). source

      The Description of the method section explains how SuSiE-RSS assesses consistency between summary statistics and a reference-panel LD matrix, including diagnostic plots. But our copy does not report EAS fine-mapped loci, s values of 0.007–0.027, or the claimed high LD consistency for those samples; those specific results are not supported by this source. (LLM, conf=0.82)

      Experimental — accuracy not yet measured

  • It may state two separate claims as one.

    Findingcitation

    SNP-based heritability was similar within the EAS population (9.7 ± 0.8% for TWB and 8.7 ± 0.9% for KoGES) and between EAS (9.0 ± 0.6%) and EUR (10.7 ± 0.3%) populations.

    “The SNP-based heritability was similar within the EAS population (9.7 ± 0.8%) for TWB and (8.7 ± 0.9%) for KoGES and between EAS (9.0 ± 0.6%) and EUR (10.7 ± 0.3%) populations14.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [14] Lee, J. J. et al. Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals. Nat. Genet. 50, 1112–1121 (2018). source

      The Results section (Primary GWAS of EduYears; page unavailable) says the association analyses were restricted to European-descent individuals. Searches for TWB, KoGES, and the quoted heritability estimates found no passage reporting the claimed within-East-Asian or East-Asian–European comparisons. (LLM, conf=0.76)

      Experimental — accuracy not yet measured

  • It may state two separate claims as one.

    Findingcitation

    The MAF values of rs2881903 and rs16930687 were 5.2% versus 8.8% and 2.3% versus 0.7% in EAS and EUR populations, respectively.

    “The MAF values of rs2881903 and rs16930687 were 5.2% versus 8.8% and 2.3% versus 0.7% in EAS and EUR14, respectively.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [14] Lee, J. J. et al. Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals. Nat. Genet. 50, 1112–1121 (2018). source

      The Results section, “Primary GWAS of EduYears” (page unavailable), says the association analyses were restricted to European-descent individuals. Searches for rs2881903, rs16930687, and the stated percentages found no passage in our copy reporting either SNP’s minor allele frequency in EAS or EUR populations. (LLM, conf=0.75)

      Experimental — accuracy not yet measured

  • It may state two separate claims as one.

    Findingown data

    GSA-SNP2 identified 16 pathways significantly associated with EduYears in the EAS population and 27 in the EUR population.

    “In total, 16 and 27 pathways were identified as significantly associated with EduYears in EAS and EUR populations, respectively (Fig. 2c).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [28] Yoon, S. et al. Efficient pathway enrichment and network analysis of GWAS summary data using GSA-SNP2. Nucleic Acids Res. 46, e60–e60 (2018). source

      The Results and Discussion section on real GWAS data describes analyses of type 2 diabetes and height datasets, not EduYears in EAS and EUR populations. Searches for “EduYears,” “educational attainment,” and the EAS/EUR pathway counts found no passage reporting 16 and 27 significant pathways. (LLM, conf=0.78)

      Experimental — accuracy not yet measured

  • It may state two separate claims as one.

    Findingown data

    SNP-based heritability for EduYears was estimated to be 9.7% in TWB, 8.7% in KoGES and 9.0% in the EAS GWAS meta-analysis.

    “At the genome-wide level, we first showed that the SNP-based heritability for EduYears was estimated to be 9.7% in TWB, 8.7% in KoGES and 9.0% in the EAS GWAS meta-analysis”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [18] Bulik-Sullivan, B. K. et al. LD score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat. Genet. 47, 291–295 (2015). source

      The Results (Overview of Methods; page unavailable) explain how an LD Score regression slope can be used to estimate SNP heritability under strong assumptions. The passages found do not report EduYears heritability estimates for TWB, KoGES, or an EAS GWAS meta-analysis; searches for those cohorts and the three percentages found no matching result. (LLM, conf=0.76)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim.

    Premisecitation

    Previous GWAS meta-analyses and replication studies have identified genetic variants associated with EduYears.

    “Previous GWAS meta-analyses and replication studies have identified genetic variants associated with EduYears11–14.”
    Quote in the paperSubstantial claim4 of 4 sources checked
    • cites [11] Rietveld, C. A. et al. GWAS of 126,559 individuals identifies genetic variants associated with educational attainment. Science 340, 1467–1471 (2013). source
    • cites [12] Rietveld, C. A. et al. Replicability and robustness of genome-wide-association studies for behavioral traits. Psychol. Sci. 25, 1975–1986 (2014). source
    • cites [13] Davies, G. et al. Genome-wide association study of cognitive functions and educational attainment in UK Biobank (N = 112 151). Mol. Psychiatry 21, 758–767 (2016). source
    • cites [14] Lee, J. J. et al. Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals. Nat. Genet. 50, 1112–1121 (2018). source

  • It may state two separate claims as one.

    Findingown data

    MAMA identified 94 independent genome-wide significant SNPs with EAS-specific meta-analysis, 2 of which were previously unreported for EduYears (rs2881903 and rs16930687).

    “Using MAMA, we identified 94 independent genome-wide significant SNPs with EAS-specific meta-analysis (Fig. 1c and Supplementary Table 13), 2 of which were previously unreported for EduYears (rs2881903 and rs16930687)”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [30] Turley, P. et al. Multi-ancestry meta-analysis yields novel genetic discoveries and ancestry-specific associations. Preprint at bioRxiv https://doi.org/10.1101/2021.04.23.441003 (2021). source

  • It may state two separate claims as one.

    Findingown data

    MAMA EUR-specific meta-analysis found 357 independent genome-wide significant SNPs, all of which had been previously reported to be associated with EduYears.

    “The MAMA EUR-specific meta-analysis found 357 independent genome-wide significant SNPs; however, all of them had been previously reported to be associated with EduYears10”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [10] Okbay, A. et al. Polygenic prediction of educational attainment within and between families from genome-wide association analyses in 3 million individuals. Nat. Genet. 54, 437–449 (2022). source

      The Results section (page unavailable) reports 3,952 lead SNPs associated with EduYears, but does not identify the 357 SNPs from the citing paper’s MAMA EUR-specific analysis. Searches for “357,” “MAMA,” and prior reporting found no passage establishing that all 357 had previously been reported as associated with EduYears. (LLM, conf=0.76)

      Experimental — accuracy not yet measured

  • It may state two separate claims as one.

    Premisecitation

    This disparity is problematic because genomic discoveries may not be transferable across populations, even though it is assumed that they share the underlying biological mechanisms.

    “This disparity is problematic because genomic discoveries may not be transferable across populations, even though it is assumed that they share the underlying biological mechanisms37”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [37] Atkinson, E. G. et al. Cross-ancestry genomic research: time to close the gap. Neuropsychopharmacology 47, 1737–1738 (2022). source
    • The authors state this as an assumption: EAS and EUR populations share underlying biological mechanisms for EduYears.

  • It may state two separate claims as one.

    Study design and conductown data

    Genome-wide associations in TWB, KoGES and their meta-analysis were consistent with a highly polygenic architecture and did not indicate inflation due to potential population stratification, with LDSC intercepts from 1.029 to 1.046.

    “Genome-wide associations in the TWB, KoGES and their meta-analysis were consistent with the highly polygenic architecture of educational attainment and did not indicate inflation due to potential population stratification (linkage disequilibrium score regression (LDSC)18 intercept ranged from 1.029 to 1.046).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [18] Bulik-Sullivan, B. K. et al. LD score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat. Genet. 47, 291–295 (2015). source

  • Probably not a substantial claim.

    Premisecitation

    Previous studies found that population diversity may improve fine-mapping resolution.

    “Previous studies have found that population diversity may improve fine-mapping resolution38,39”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [38] Lam, M. et al. Comparative genetic architectures of schizophrenia in East Asian and European populations. Nat. Genet. 51, 1670–1678 (2019). source
    • cites [39] Peterson, R. E. et al. Genome-wide association studies in ancestrally diverse populations: opportunities, methods, pitfalls, and recommendations. Cell 179, 589–603 (2019). source

  • It may state two separate claims as one.

    Findingown data

    From the seven associated loci, fine-mapping identified 8 credible sets in the EAS GWAS and 13 credible sets in the cross-population GWAS.

    “From the seven associated loci, we identified 8 credible sets in the EAS GWAS fine-mapping and 13 credible sets in the cross-population GWAS fine-mapping”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [33] Yuan, K. et al. Fine-mapping across diverse ancestries drives the discovery of putative causal variants underlying human complex traits and diseases. Preprint at medRxiv https://doi.org/10.1101/2023.01.07.23284293 (2023). source

      The Methods section describes EAS and cross-population fine-mapping, but does not report results for seven associated loci. The Results sections report different aggregate counts; searches for “seven associated loci,” “8 credible sets,” and “13 credible sets” found no passage supporting the citing authors’ specific counts. No page numbers are available in our copy. (LLM, conf=0.78)

      Experimental — accuracy not yet measured

  • It may state two separate claims as one.

    Findingcitation

    All 11 independent SNPs identified in the EAS GWAS meta-analysis were previously reported, located within ±500 kb of the 3,952 lead SNPs reported by Okbay et al.

    “All these 11 independent SNPs were previously reported (SNPs located within ±500 kb of the 3,952 lead SNPs reported by Okbay et al.10).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Okbay et al. [10] Okbay, A. et al. Polygenic prediction of educational attainment within and between families from genome-wide association analyses in 3 million individuals. Nat. Genet. 54, 437–449 (2022). source

  • It may state two separate claims as one.

    Premisecitation

    EduYears is moderately heritable, with twin-study heritability of ~40% and SNP-based heritability of ~20% from GWAS.

    “EduYears has been shown to be moderately heritable with a heritability of ~40% from twin studies and a single-nucleotide polymorphism (SNP)-based heritability of ~20% from genome-wide association studies (GWASs)11.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [11] Rietveld, C. A. et al. GWAS of 126,559 individuals identifies genetic variants associated with educational attainment. Science 340, 1467–1471 (2013). source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    Okbay et al. found positive genetic correlations between EduYears and intracranial volume, bipolar disorder, schizophrenia, cognitive performance, and height and negative genetic correlations between EduYears and Alzheimer's disease, neuroticism and body mass index in the EUR population.

    “Okbay et al. investigated the genetic correlation of EduYears with 14 traits and found positive genetic correlations between EduYears and intracranial volume, bipolar disorder, schizophrenia, cognitive performance, and height and negative genetic correlations between EduYears and Alzheimer’s disease, neuroticism and body mass index in the EUR population9.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [9] Okbay, A. et al. Genome-wide association study identifies 74 loci associated with educational attainment. Nature 533, 539–542 (2016). source

      Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

      Experimental — accuracy not yet measured

  • It may state two separate claims as one.

    Premisecitation

    EduYears shows phenotypic correlations and shares genetic components with multiple traits and diseases relevant to medical research, including cognitive function, neurodegeneration and psychiatric disorders.

    “As previous studies and our results suggest, EduYears shows phenotypic correlations and shares genetic components with multiple traits and diseases relevant to medical research, including cognitive function, neurodegeneration and psychiatric disorders43–46”
    Quote in the paperSubstantial claim4 of 4 sources checked
    • cites [43] Trampush, J. W. et al. Independent evidence for an association between general cognitive ability and a genetic locus for educational attainment. Am. J. Med. Genet. Part B 168, 363–373 (2015). source

      The Introduction (page unavailable) reports that lower cognitive performance is associated with lower educational attainment, and the Results > COGENT Analysis (page unavailable) finds an education-associated variant also associated with general cognitive ability. The paper does not establish that EduYears shares genetic components with neurodegenerative or psychiatric disorders; its Introduction discusses cognitive deficits in those disorders, which is a narrower point. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
    • cites [44] Elliott, M. L. et al. A polygenic score for higher educational attainment is associated with larger brains. Cereb. Cortex 29, 3496–3504 (2019). source

      The Introduction and Discussion report associations between educational-attainment genetics and cognitive performance, supporting the cognitive-function part of the claim. The passages do not establish phenotypic correlations or shared genetic components with neurodegeneration or psychiatric disorders. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
    • cites [45] Raghavan, N. S., Vardarajan, B. & Mayeux, R. Genomic variation in educational attainment modifies Alzheimer disease risk. Neurol. Genet. 5, e310 (2019). source

      The Results and Discussion (page not provided) link educational attainment to Alzheimer disease risk, and the paper reports a genetic correlation between the two. The Discussion also notes that genes in implicated regions have been associated with bipolar disorder or major depression. It does not establish phenotypic correlations or shared genetic components between EduYears and cognitive function or psychiatric disorders broadly. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
    • cites [46] Mullins, N. et al. Genome-wide association study of more than 40,000 bipolar disorder cases provides new insights into the underlying biology. Nat. Genet. 53, 817–829 (2021). source

      The Discussion (page not provided) reports epidemiological associations between educational attainment and bipolar disorder and estimates that almost all variants influencing bipolar disorder also influence educational attainment. This supports the psychiatric-disorder example, but the paper does not establish the claimed relationships between EduYears and cognitive function or neurodegeneration. (LLM, conf=0.88)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationcitation

    Our results do not support any immediate medical or clinical applications, such as polygenic prediction in direct-to-consumer services.

    “our results do not support any immediate medical or clinical applications, such as polygenic prediction in direct-to-consumer services49,50.”
    Quote in the paperNot a substantial claim2 of 2 sources checked
    • cites [49] Turley, P. et al. Problems with using polygenic scores to select embryos. N. Engl. J. Med. 385, 78–86 (2021). source

      The Clinical Outcomes section (page unavailable) says the clinical utility of polygenic scores among adults still needs research and that predictive power is lower for embryo selection. The Discussion (page unavailable) cautions companies offering embryo-selection services about communicating uncertain benefits. These passages support caution about that application, but not the broader claim that the paper's results rule out any immediate medical or clinical use of polygenic prediction, including direct-to-consumer services generally. (LLM, conf=0.83)

      Experimental — accuracy not yet measured
    • cites [50] Karavani, E. et al. Screening human embryos for polygenic traits has limited utility. Cell 179, 1424–1435.e8 (2019). source

      The Discussion says limitations in polygenic-score accuracy and other constraints reduce the utility of embryo screening. It does not extend that conclusion to all immediate medical or clinical applications, or evaluate polygenic prediction in direct-to-consumer services generally; its mention of consumers concerns embryo selection specifically. The claim therefore applies the paper's narrower finding too broadly. (LLM, conf=0.83)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim.

    Study design and conductown data

    Population-specific LD scores based on the 1KG Project phase 3 data were used to estimate SNP-based heritability.

    “We performed LDSC v1.0.1 (ref. 18) to estimate the SNP-based heritability of EduYears in the TWB, KoGES, EAS and EUR populations14 using population-specific LD scores based on the 1KG Project phase 3 data.”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [18] Bulik-Sullivan, B. K. et al. LD score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat. Genet. 47, 291–295 (2015). source

      The Results (Overview of Methods) say LD Score regression can estimate SNP-based heritability under strong assumptions, and Online Methods (Estimation of LD Score) describe LD scores derived from the 1000 Genomes European reference panel. Online Methods (Intra-European LD Score Differences) also describe scores for European subpopulations. The paper does not specify phase 3 data or report using population-specific scores for TWB, KoGES, and EAS heritability estimates. (LLM, conf=0.87)

      Experimental — accuracy not yet measured
    • cites [14] Lee, J. J. et al. Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals. Nat. Genet. 50, 1112–1121 (2018). source

      The Online Methods (Cross-cohort genetic correlation; page unavailable) report using LDSC with a European reference population, and the Online Methods (Biological annotation; page unavailable) describe stratified LD Score regression to partition heritability. The paper does not report using population-specific LD scores from 1000 Genomes phase 3 to estimate SNP-based heritability across TWB, KoGES, EAS and EUR; its primary GWAS was restricted to European-descent individuals. (LLM, conf=0.85)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductcitation

    Covariates for genetic association analyses were selected based on previous GWAS for EduYears.

    “We adjusted for the birth year (BY), BY2, BY3, sex, BY by sex interaction, BY2 by sex interaction, BY3 by sex interaction, and the top ten PCs, based on previous GWAS for EduYears14.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [14] Lee, J. J. et al. Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals. Nat. Genet. 50, 1112–1121 (2018). source

      The Results section, “Polygenic Prediction” (page unavailable), lists sex, birth year, their interaction, and ten genetic principal components as baseline controls. That supports part of the cited covariate list, but it describes prediction analyses, not the GWAS association models, and our copy does not establish the claimed squared and cubed birth-year terms or their sex interactions. (LLM, conf=0.82)

      Experimental — accuracy not yet measured

  • It may state two separate claims as one.

    Findingown data

    Overall, the cross-population PGS from PRS-CSx showed better performance than the single-population PGS; the largest improvement was observed when applying PRS-CSx to the EAS testing cohorts.

    “Overall, the cross-population PGS from PRS-CSx showed better performance than the single-population PGS; the largest improvement was observed when applying PRS-CSx to the EAS testing cohorts.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Findingnone given

    Both genes have not been reported to be associated with other traits in GWAS.

    “Both genes have not been reported to be associated with other traits in GWAS.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    EduYears showed significant genetic correlations with 12 out of the 82 traits in the EAS population, all of which were also significant in the EUR population with concordant directions.

    “EduYears showed significant genetic correlations with 12 out of the 82 traits in the EAS population, all of which were also significant in the EUR population with concordant directions.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    The EAS GWAS meta-analysis results were similar to those of TWB and KoGES separately, except for one locus on chromosome 12 near ALDH2 that showed significant heterogeneity.

    “The results of the EAS GWAS meta-analysis were similar to those of the TWB and KoGES separately, except for one locus on chromosome 12 near ALDH2 that showed significant heterogeneity (Supplementary Fig. 3).”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    The direction of genetic effects was consistent between TWB and KoGES for most genome-wide significant SNPs, except for the genome-wide significant locus on chromosome 12.

    “we observed that the direction of genetic effects was consistent between TWB and KoGES for most genome-wide significant SNPs, except for the genome-wide significant locus on chromosome 12”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingcitation

    The EAS-EUR cross-population rg for EduYears is lower than that for schizophrenia (0.945), higher than major depressive disorder (0.342), and comparable to height (0.897) and haemoglobin A1c (0.875).

    “While the EAS–EUR cross-population rg for EduYears is lower than that for schizophrenia (EAS–EUR cross-population rg = 0.945), it is considerably higher than major depressive disorder (EAS–EUR cross-population rg = 0.342) and comparable to other physiological traits (EAS–EUR cross-population rg = 0.897 for height) and molecular phenotypes (EAS–EUR cross-population rg = 0.875 for haemoglobin A1c).”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Findingown data

    The nearest gene for rs2881903 was FAM81B, and the nearest gene for rs16930687 was a processed pseudogene (PR11-224P11.1).

    “The nearest gene for rs2881903 was FAM81B, and the nearest gene for rs16930687 was a processed pseudogene (PR11-224P11.1).”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingreasoning

    The cross-population PGS explained 6.1% of the phenotypic variance in EduYears in the EUR cohort, equivalent to ancestry-matched PGS in the EUR cohort.

    “However, the cross-population PGS explained 6.1% of the phenotypic variance in EduYears, equivalent to ancestry-matched PGS in the EUR cohort”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Among significantly enriched pathways, 14 were common across both populations, while 2 and 13 were exclusive to EAS and EUR populations, respectively.

    “Among these significantly enriched pathways, 14 pathways were common across both populations, while 2 and 13 pathways exhibited significant enrichment exclusively in EAS and EUR populations, respectively.”
    • Also stated in: § Discussion · plain
      “We showed that 14 pathways were significantly associated with EduYears in both populations.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    This study presents the largest-so-far EduYears GWAS in the EAS population and a cross-ancestry GWAS meta-analysis across EAS and EUR populations, including 176,400 EAS and 766,345 EUR samples.

    “We present the largest-so-far EduYears GWAS in the EAS population and cross-ancestry GWAS meta-analysis across EAS and EUR populations for EduYears, including 176,400 samples of EAS genetic ancestry from TWB and KoGES and 766,345 samples of EUR genetic ancestry from previous studies”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Cross-population GWAS improved fine-mapping resolution and PGS prediction performance in the context of educational attainment.

    “Furthermore, we demonstrated that cross-population GWAS improved fine-mapping resolution and PGS prediction performance in the context of educational attainment.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Compared with the largest GWAS for EduYears in EUR, considerably fewer genomic loci in EAS were identified (7 loci in EAS versus 3,952 loci in EUR), and all 7 loci reported in EAS were previously reported in the EUR GWAS.

    “However, compared with the largest GWAS for EduYears in EUR, considerably fewer genomic loci in EAS were identified (7 loci in EAS versus 3,952 loci in EUR) and all 7 loci reported in EAS were previously reported in the EUR GWAS.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductreasoning

    Participants might be classified into the same educational category regardless of whether they graduated from a specific educational level.

    “Regardless of whether the participants graduated from a specific educational level, they might be classified into the same category.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Premisereasoning

    Previous studies of the genetic architecture of EduYears focused solely on samples of European genetic ancestries, and evidence for generalizability to non-European populations is limited.

    “However, previous studies investigated the genetic architecture of EduYears solely focused on samples of EUR genetic ancestries, and the evidence for the generalizability of the findings to non-EUR populations is limited.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductnone given

    The difference in the years of compulsory education required in Taiwan and South Korea might limit the phenotypic variation in EduYears.

    “On the other hand, the difference in the years of compulsory education, described in detail in Supplementary Note, required in Taiwan and South Korea might limit the phenotypic variation in EduYears.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationreasoning

    The larger standard error of heritability estimates in the EAS population reflected the smaller EAS GWAS sample sizes relative to the EUR GWAS.

    “The larger s.e. of heritability estimates in the EAS reflected the smaller EAS GWAS sample sizes relative to the EUR GWAS.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Interpretationown data

    This result further supports a similar genetic architecture for EduYears between the two ancestries and highlights shared genetic components between EduYears and various socioeconomic and health-related outcomes.

    “This result further supports a similar genetic architecture for EduYears between the two ancestries and highlights the shared genetic components between EduYears and various socioeconomic and health-related outcomes.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. It may be too vague to check against evidence.

    Interpretationown data

    Studying diverse populations can bring insights in identifying gene-environment associations.

    “This finding suggests that studying diverse populations can bring insights in identifying gene–environment associations.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Interpretationreasoning

    Our EduYears GWAS in EAS and cross-population GWAS meta-analysis enhance comprehension of the genetic basis of EduYears and facilitate the transfer of genetic insights for EduYears across populations.

    “As the largest previous GWAS of EduYears was limited to the EUR population, our EduYears GWAS in EAS and cross-population GWAS meta-analysis enhance our comprehension of the genetic basis of EduYears and facilitate the transfer of genetic insights for EduYears across populations.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim.

    Study design and conductreasoning

    Bonferroni correction was applied to all tested genes and gene sets to account for multiple comparisons.

    “We then applied a Bonferroni correction to all tested genes and gene sets to account for multiple comparisons.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductown data

    FDR correction was applied to control for false positive discoveries in genetic correlation analysis with other traits.

    “We then applied FDR correction to control for false positive discoveries.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.

    Interpretationreasoning

    Findings on genetic overlaps between EduYears and health outcomes may shed light on the genetic basis of these relevant health outcomes.

    “and findings on genetic overlaps between EduYears and health outcomes may shed light on the genetic basis of these relevant health outcomes.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim.

    Study design and conductcitation

    Genetic association analyses used a two-step whole-genome regression model that accounts for sample relatedness and population structure.

    “We performed genetic association analyses on EduYears using post-QC imputed genotype data and a two-step whole-genome regression model implemented in Regenie v2.2.4 (ref. 57), which accounts for sample relatedness and population structure.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [57] Mbatchou, J. et al. Computationally efficient whole-genome regression for quantitative and binary traits. Nat. Genet. 53, 1097–1103 (2021). source

  • Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.

    Premisecitation

    The impact of EduYears on health outcomes is probably via complex mechanisms like mediation and interaction between genetic and environmental factors.

    “and the impact of EduYears on health outcomes is probably via complex mechanisms like mediation and interaction between genetic and environmental factors47.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [47] Schork, A. J., Peterson, R. E., Dahl, A., Cai, N. & Kendler, K. S. Indirect paths from genetics to education. Nat. Genet. 54, 372–373 (2022). source

  • Probably not a substantial claim. It may be too vague to check against evidence.

    Interpretationreasoning

    These results underscore the importance of combining diverse population cohorts in genetic studies.

    “These results underscore the importance of combining diverse population cohorts in genetic studies.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim.

    Study design and conductreasoning

    The top ten principal components were included as covariates to control for population stratification.

    “The top ten PCs were included as covariates to control for population stratification.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim.

    Study design and conductreasoning

    Each education category was mapped to the International Standard Classification of Education (ISCED) category to ensure comparability across cohorts.

    “To ensure comparability across cohorts, including TWB, KoGES and cohorts in GWAS meta-analysis in EUR population, we mapped each category in these questions to the International Standard Classification of Education (ISCED) category.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductcitation

    Confidence intervals were inferred using 1,000 bootstrap replicates over samples.

    “To infer confidence intervals, we used the boot package68 in R with 1,000 bootstrap replicates over samples.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [68] Davison, A. C. & Hinkley, D. V. Bootstrap Methods and Their Application (Cambridge Univ. Press, 1997). source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductown data

    Stringent pre-imputation quality control was conducted.

    “We conducted stringent pre-imputation QC, followed by the PBK genotype QC project pipeline (see URLs), for samples in TWB batch 1, TWB batch 2 and KoGES.”
    Quote in the paperNot a substantial claim
48 of 48

  • The authors state this as an assumption.

    EAS and EUR populations share underlying biological mechanisms for EduYears.

    “even though it is assumed that they share the underlying biological mechanisms37”
1 of 1

Citations

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Metadata

Authors
Chen, T.-T., Kim, J., Lam, M., Chuang, Y.-F., Chiu, Y.-L., Lin, S.-C., Jung, S.-H., Kim, B., Kim, S., Cho, C., Shim, I., Park, S., Ahn, Y., Okbay, A., Jang, H., Kim, H. J., Seo, S. W., Park, W.-Y., Ge, T., Huang, H., Feng, Y.-C. A., Lin, Y.-F., Myung, W., Chen, C.-Y., & Won, H.-H.
Affiliations
  • National Health Research Institutes (TW)
  • Seoul National University (KR)
  • Northwell Health (US)
  • National Yang Ming Chiao Tung University (TW)
  • Far Eastern Memorial Hospital (TW)
  • University of Pennsylvania (US)
  • Samsung Medical Center (KR)
  • Broad Institute (US)
  • Vrije Universiteit Amsterdam (NL)
  • National Taiwan University (TW)
  • Biogen (United States) (US)
Corresponding author
  • Feng, Yen-Chen Anne
  • Myung, Woojae
  • Chen, Chia-Yen
  • Won, Hong-Hee
ORCID
Venue
Nature Human Behaviour
Publisher
Springer Science and Business Media LLC
Published
2024-01-05
Volume
8
Issue
3
Pages
562-575
ISSN
2397-3374 (ISSN-L)
DOI
10.1038/s41562-023-01781-9
PMC
PMC10963262
OpenAlex
W4390614142
Cited by
18
References
68
Citations by year
2026: 6 · 2025: 8 · 2024: 4
FWCI
4.78
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Human Behaviour (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Pure Amsterdam UMC (publishedVersion, cc-by)
  • VU Research Portal (publishedVersion, other-oa)
  • RePEc: Research Papers in Economics (submittedVersion)
Funders
  • National Research Foundation of Korea (NRF-2022R1A2C2009998, NRF-2021R1A2C4001779)
  • Korea Health Industry Development Institute (HU22C0042, HU21C0111)
  • Ministry of Science and Technology, Taiwan (106-2628-B-010-001-MY4, 110-2314-B-A49A-524, 110-2321-B-418-001, 112-2314-B-002-200-MY3, 111-2314-B-002-299, 109-2314-B-400-017, 110-2314-B-400-028-MY3)
  • National Health Research Institutes (NP-110-PP-09, NP-111-PP-09, NP-112-PP-09)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genetic Associations and Epidemiology
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Cognitive Abilities and Testing
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genetic Mapping and Diversity in Plants and Animals
Keywords
educational attainment, genetic architecture, East Asian ancestry, European ancestry, genetic correlation, gene expression enrichment, fine mapping, polygenic prediction, cognitive ability
MeSH terms
Academic Success, East Asian People, Educational Status, Humans, Multifactorial Inheritance, White People, Genome-Wide Association Study

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 2026-10-29, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-29 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

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

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