Genetic determinants of plasma protein levels in the Estonian population

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The largest pQTL study to date was conducted with 35,571 samples and resulted in the detection of 10,248 independent autosomal pQTLs for 1463 proteins.

“We extended the analysis to include data from the largest pQTL study to date, conducted with 35,571 samples and resulting in the detection of 10,248 independent autosomal pQTLs for 1463 proteins9.”

• cites [9] Sun, B. B. et al. Genetic regulation of the human plasma proteome in 54,306 UK Biobank participants. 2022.06.17.496443 Preprint at https://doi.org/10.1101/2022.06.17.496443 (2022). sourceDisputed

  The Discovery of pQTLs section reports 34,557 discovery participants, 14,287 significant primary associations, and 2,414 proteins with at least one pQTL—not the figures in the claim. The Replication of pQTLs section also reports an analysis of 52,363 participants, so the cited study is not described by the claimed sample size. (LLM, conf=0.90)

  Checked by a reasoning model — may be wrong.

  Experimental — accuracy not yet measured

May combine two claims.

The 95% CSs contained an average of 15.7 variants in our study and 22.7 variants (9.6 cis and 29.4 trans) in Sun et al.

“The 95% CSs contained an average of 15.7 variants in our study and 22.7 variants (9.6 cis and 29.4 trans) in that of Sun et al.9.”

• cites [9] Sun, B. B. et al. Genetic regulation of the human plasma proteome in 54,306 UK Biobank participants. 2022.06.17.496443 Preprint at https://doi.org/10.1101/2022.06.17.496443 (2022). sourceDisputed

  The cited paper’s “Identification and fine-mapping of independent signals” section reports an average of 20.5 variants per 95% credible set, with means of 9.7 for cis and 26.7 for trans signals—not the attributed 22.7, 9.6, and 29.4. The 15.7 figure describes the citing authors’ own study and is not attributed to this source. (LLM, conf=0.90)

  Checked by a reasoning model — may be wrong.

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

To account for hidden batch effects on gene expression, the first two principal components of the gene expression data were included as covariates, as described in Lepik et al.

“To account for hidden batch effects on the gene expression, the first two principal components of the gene expression data were also included as covariates, as described in detail in Lepik et al.29.”

• cites Lepik et al. [29] Lepik, K. et al. C-reactive protein upregulates the whole blood expression of CD59—An integrative analysis. PLOS Comput. Biol. 13, e1005766 (2017). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Sun et al. reported that approximately 5.6% (570/10,248) and 1.5% (155/10,248) of the variants with primary associations had MAFs < 0.01 and < 0.005, respectively.

“Sun et al.9 reported that approximately 5.6% (570/10,248) and 1.5% (155/10,248) of the variants with primary associations had MAFs < 0.01 and < 0.005, respectively.”

• cites Sun et al. [9] Sun, B. B. et al. Genetic regulation of the human plasma proteome in 54,306 UK Biobank participants. 2022.06.17.496443 Preprint at https://doi.org/10.1101/2022.06.17.496443 (2022). sourcePartly supported

  The Discovery of pQTLs section (page unavailable) reports nearly the same proportions at the stated MAF thresholds, but different counts: 783/14,287 (5.5%) below 1% and 235/14,287 (1.6%) below 0.5%. It does not support the cited numerators or denominator. (LLM, conf=0.88)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim.

To provide more conservative results accounting for the number of tested proteins, the study used a strategy described by Gao et al. and Kettunen et al. that accommodates the correlation between protein levels.

“To also provide the more conservative results accounting for the number of tested proteins, we used a strategy described by Gao et al.35 and Kettunen et al.36, which accommodates the correlation between protein levels.”

• cites Gao et al. [35] Gao, X., Starmer, J. & Martin, E. R. A multiple testing correction method for genetic association studies using correlated single nucleotide polymorphisms. Genet. Epidemiol. 32, 361–369 (2008). sourceSupported
• cites Kettunen et al. [36] Kettunen, J. et al. Genome-wide study for circulating metabolites identifies 62 loci and reveals novel systemic effects of LPA. Nat. Commun. 7, 11122 (2016). sourceSupported

May combine two claims.

Of the pQTLs from Pietzner et al. that were nominally significant and accessible for comparison, 147 (52.88%) were nominally significant, all 147 remained significant after Benjamini-Hochberg FDR < 0.05, and 91.84% (135/147) were directionally concordant with the current study.

“From the Pietzner et al. study7, 147 pQTLs (52.88%) were nominally significant (P < 0.05) and accessible for comparisons. After correcting for multiple testing, 147 pQTLs remained significant (Benjamini–Hochberg FDR < 0.05) and 91.84% (135/147) of pQTLs were directionally concordant with the current study (Supplementary Table S2).”

• cites [7] Pietzner, M. et al. Mapping the proteo-genomic convergence of human diseases. Science 374, eabj1541 (2021). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Zheng et al. used MR and colocalisation analysis to examine associations of 1002 plasma proteins with 153 diseases and 72 disease-related risk factors, and identified 413 protein-trait associations supported by MR, 130 (31.5%) of which were not supported by the colocalisation analysis.

“Zheng et al.16 used MR and colocalisation analysis to examine associations of 1002 plasma proteins with 153 diseases and 72 disease-related risk factors, and identified 413 protein–trait associations supported by MR, 130 (31.5%) of which were not supported by the colocalisation analysis.”

• cites Zheng et al. [16] Zheng, J. et al. Phenome-wide Mendelian randomization mapping the influence of the plasma proteome on complex diseases. Nat. Genet. 52, 1122–1131 (2020). sourceSupported

Classified as probably not a substantial claim.

To retain independent signals, associated variants were clumped in PLINK using a 1 Mb window with LD thresholds of R2 = 0.1 and P < 5 × 10-8.

“To retain independent signals, associated variants were clumped in PLINK (version 1.9)31, using a 1 Mb window with the LD thresholds of R2 = 0.1 and P < 5 × 10–8.”

• cites [31] Chang, C. C. et al. Second-generation PLINK: Rising to the challenge of larger and richer datasets. GigaScience 4, 7 (2015). sourceSupported

May combine two claims.

Of the 278 pQTLs, 184 (66.19%) were tested in the Sun et al. study; 103 (55.98%) were significant at Benjamini-Hochberg FDR < 0.05 and 89.32% (92/103) were directionally concordant.

“66.19% (184/278) of pQTLs were tested in the Sun et al. study4. Of them, 55.98% (103/184) were significant (Benjamini–Hochberg FDR < 0.05) and 89.32% (92/103) were directionally concordant (Supplementary Table S2).”

• cites [4] Sun, B. B. et al. Genomic atlas of the human plasma proteome. Nature 558, 73–79 (2018). sourceSupported

Classified as probably not a substantial claim.

To flag potential pseudo-pQTL signals caused by the epitope effect, the study followed the strategy described by Folkersen et al.

“To flag potential ‘pseudo-pQTL’ signals caused by the epitope effect, i.e. altered assay binding affinity due to a change in protein structure instead of an actual change in protein expression level, we followed the strategy described by Folkersen et al.6.”

• cites Folkersen et al. [6] Folkersen, L. et al. Genomic and drug target evaluation of 90 cardiovascular proteins in 30,931 individuals. Nat. Metab. 2, 1135–1148 (2020). sourceSupported

Classified as probably not a substantial claim.

In the pQTL-eQTL colocalisation analysis, significant pQTL loci were compared to eQTL Catalogue datasets, excluding those of Kasela et al. and Lepik et al. due to sample overlap.

“In an pQTL–eQTL colocalisation analysis, we compared our significant pQTL loci to all eQTL Catalogue datasets18, excluding those of Kasela et al.57 and Lepik et al.29 due to sample overlap, containing gene expression, exon expression, transcript usage and txrevise event usage data, and GTEx (version 8)58 datasets containing gene expression data”

• cites [18] Kerimov, N. et al. A compendium of uniformly processed human gene expression and splicing quantitative trait loci. Nat. Genet. 53, 1290–1299 (2021). sourceSupported
• cites Kasela et al. [57] Kasela, S. et al. Pathogenic implications for autoimmune mechanisms derived by comparative eQTL analysis of CD4+ versus CD8+ T cells. PLoS Genet. 13, e1006643 (2017). sourceSupported
• cites Lepik et al. [29] Lepik, K. et al. C-reactive protein upregulates the whole blood expression of CD59—An integrative analysis. PLOS Comput. Biol. 13, e1005766 (2017). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

IL6R was selected because it has been widely reported by previous pQTL studies as an example of the successful linking of molecular traits and diseases to discover drug targets.

“IL6R was selected because it has been widely reported by previous pQTL studies as an example of the successful linking of molecular traits and diseases to discover drug targets50,59.”

• cites [50] Folkersen, L. et al. Mapping of 79 loci for 83 plasma protein biomarkers in cardiovascular disease. PLoS Genet. 13, e1006706 (2017). sourcePartly supported

  The Discussion (‘Insights into potential causal involvement of the plasma proteins in CVD’; page unavailable) identifies IL6R as a protein with evidence of potentially causal involvement in coronary artery disease and notes trials of an IL6R inhibitor. The Discussion also says the IL6R pQTL association was previously known and replicated. It does not establish that previous pQTL studies widely reported IL6R as a successful example of drug-target discovery; the inhibitor trials were still underway. (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites [59] Bretherick, A. D. et al. Linking protein to phenotype with Mendelian Randomization detects 38 proteins with causal roles in human diseases and traits. PLoS Genet. 16, e1008785 (2020). sourcePartly supported

  The Results section, “One-to-Many: Linking IL6R levels to atopy, rheumatoid arthritis, and coronary artery disease” (page not provided), links plasma IL6R abundance to several diseases and notes that an IL6R-targeting drug is in clinical use. It cites prior MR and genetic evidence, but does not establish that IL6R was widely reported by previous pQTL studies as an example of drug-target discovery. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

For the common variant pQTL analysis power calculations with n = 500, the study assumed linear regression with an additive genetic model with no gene-gene interactions using genpwr, following the strategy previously used by Yao et al.

“For the common variant pQTL analysis power calculations with n = 500, we assumed linear regression with an additive genetic model with no gene–gene interactions using genpwr (version 1.0.4) R package33, following the strategy previously used by Yao et al.34.”

• cites [33] Moore, C. M., Jacobson, S. A. & Fingerlin, T. E. Power and sample size calculations for genetic association studies in the presence of genetic model misspecification. Hum. Hered. 84, 256–271 (2020). sourceSupported
• cites Yao et al. [34] Yao, C. et al. Genome-wide mapping of plasma protein QTLs identifies putatively causal genes and pathways for cardiovascular disease. Nat. Commun. 9, 3268 (2018). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Investigation of the effects of other structural variants, such as copy number variants (CNVs), on protein levels has thus far been limited.

“Investigation of the effects of other structural variants, such as copy number variants (CNVs), on protein levels has thus far been limited15.”

• cites [15] Png, G. et al. Population-wide copy number variation calling using variant call format files from 6,898 individuals. Genet. Epidemiol. 44, 79–89 (2020). sourcePartly supported

  The Introduction (page unavailable) says population-wide CNV detection remains challenging, and the Results’ Association analysis examines CNV deletions against measured protein levels. These passages support the context for limited investigation, but they do not specifically establish that research on CNVs’ effects on protein levels has been limited. (LLM, conf=0.78)

  Experimental — accuracy not yet measured

May combine two claims.

Of the 278 pQTLs, 21 (7.55%) were tested in the Suhre et al. study; 12 (57.14%) were significant at Benjamini-Hochberg FDR < 0.05 and all significant pQTLs were directionally concordant.

“7.55% (21/278) pQTLs were also tested in the Suhre et al. study51 and 57.14% (12/21) were significant (Benjamini–Hochberg FDR < 0.05), and all the significant pQTLs were directionally concordant with the current study (Supplementary Table S2).”

• cites [51] Suhre, K. et al. Connecting genetic risk to disease end points through the human blood plasma proteome. Nat. Commun. 8, 14357 (2017). sourceSupported

May combine two claims.

Five of 145 significant independent signals for CAD from mixed-ancestry samples and 2 of 7 significant loci for rheumatoid arthritis from the study conducted by Stahl et al. were pQTLs in our dataset.

“For example, 5 of 145 significant independent signals for CAD from mixed-ancestry samples67 and 2 of 7 significant loci for rheumatoid arthritis from the study conducted by Stahl et al.68 were pQTLs in our dataset.”

• cites [67] van der Harst, P. & Verweij, N. Identification of 64 novel genetic loci provides an expanded view on the genetic architecture of coronary artery disease. Circ. Res. 122, 433–443 (2018). sourcePartly supported

  The Results section, “Genome-Wide Analyses of 34 541 Cases and 261 984 Controls” (page unavailable), confirms that this is a CAD association study reporting significant independent SNPs. It does not substantiate the cited figure of 145 signals or describe them as coming from mixed-ancestry samples; it reports 161 genome-wide significant CAD loci instead. The pQTL overlap is presented as the citing authors’ own finding, and the rheumatoid-arthritis figure is attached to a different citation. (LLM, conf=0.80)

  Experimental — accuracy not yet measured
• cites [68] Stahl, E. A. et al. Genome-wide association study meta-analysis identifies seven new rheumatoid arthritis risk loci. Nat. Genet. 42, 508–514 (2010). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

A previous study has shown a link between IL6R and CAD.

“A previous study has shown a link between IL6R and CAD69.”

• cites [69] Interleukin-6 Receptor Mendelian Randomisation Analysis (IL6R MR) Consortium et al. The interleukin-6 receptor as a target for prevention of coronary heart disease: A mendelian randomisation analysis. Lancet Lond. Engl. 379, 1214–1224 (2012). sourceSupported

May combine two claims.

At the low confidence interaction score of 0.15, TMEM119 and SELPLG have a shared connection due to co-expression and co-mentioning in PubMed abstracts, and the functional link was also supported by GeneMANIA.

“However, at the low confidence interaction score of 0.15, TMEM119 and SELPLG have a shared connection due to co-expression and co-mentioning in PubMed abstracts, functional link was also supported by GeneMANIA.”

• cites [41] Szklarczyk, D. et al. The STRING database in 2023: Protein–protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucleic Acids Res. 51, D638–D646 (2023). sourcePartly supported

  The DATABASE CONTENT section (page unavailable) says STRING uses co-mentions in PubMed abstracts, and IMPROVED CO-EXPRESSION ANALYSIS (page unavailable) describes co-expression as evidence for functional associations. But the paper does not report a TMEM119–SELPLG connection at score 0.15 or say GeneMANIA supported that link; the claim’s specific result is unsupported. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites [39] Franz, M. et al. GeneMANIA update 2018. Nucleic Acids Res. 46, W60–W64 (2018). sourcePartly supported

  The Introduction (Use cases; page unavailable) says GeneMANIA returns connections among query genes and can help generate hypotheses about gene function. It does not report a TMEM119–SELPLG connection, an interaction score of 0.15, or co-expression and co-mentioning in PubMed abstracts; searches for the gene names and score found no such passage. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites [40] Warde-Farley, D. et al. The GeneMANIA prediction server: Biological network integration for gene prioritization and predicting gene function. Nucleic Acids Res. 38, W214-220 (2010). sourcePartly supported

  The Introduction (page not provided) describes GeneMANIA as displaying functional association networks, and Data sources (page not provided) confirms that it uses co-expression data. The paper does not report a TMEM119–SELPLG link, a 0.15 interaction score, PubMed co-mentioning, or a GeneMANIA result supporting that specific link. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

May combine two claims.

Trans associations between rare variants and SELPLG and MUC-16 levels were supported by the GeneMANIA-based identification of two shared networks: between TMEM119 and SELPLG, as well as GAL3ST2 and MUC-16, respectively.

“Trans associations between rare variants and SELPLG and MUC-16 levels were supported by the GeneMANIA-based identification of two shared networks: between TMEM119 and SELPLG, as well as GAL3ST2 and MUC-16, respectively.”

• cites [39] Franz, M. et al. GeneMANIA update 2018. Nucleic Acids Res. 46, W60–W64 (2018). sourceSupported
• cites [40] Warde-Farley, D. et al. The GeneMANIA prediction server: Biological network integration for gene prioritization and predicting gene function. Nucleic Acids Res. 38, W214-220 (2010). sourceSupported

May combine two claims.

At the low confidence interaction score of 0.15 in STRING, connections were established between JAKMIP1, RNF112, CKAP5 and TUBGCP4 based on co-expression and co-mentioning in PubMed abstracts but no links were detected with GDF-15.

“On the other hand, at the low confidence interaction score of 0.15, connections were established between JAKMIP1, RNF112, CKAP5 and TUBGCP4 based on co-expression and co-mentioning in PubMed abstracts but no links were detected with GDF-15.”

• cites [41] Szklarczyk, D. et al. The STRING database in 2023: Protein–protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucleic Acids Res. 51, D638–D646 (2023). sourceCould not verify

  The Improved Co-expression Analysis and Database Content sections describe STRING’s co-expression and text-mining methods, but do not report this particular network. Searches for JAKMIP1, GDF-15, the other named proteins, and the 0.15 threshold found no passage confirming the claimed connections or absence of links to GDF-15; page numbers were not available in our copy. (LLM, conf=0.76)

  Experimental — accuracy not yet measured

May combine two claims.

Based on BioGRID, trans associations between TMEM119 and SELPLG, as well as GAL3ST2 and MUC-16 had no supporting interactions.

“Based on BioGRID42,43, trans associations between TMEM119 and SELPLG, as well as GAL3ST2 and MUC-16 had no supporting interactions.”

• cites [42] Stark, C. et al. BioGRID: A general repository for interaction datasets. Nucleic Acids Res. 34, D535–D539 (2006). sourcePartly supported

  The Introduction identifies BioGRID as a repository of physical and genetic interactions, supporting its use as the cited resource. Neither that section nor the retrieved passages report whether TMEM119–SELPLG or GAL3ST2–MUC-16 have supporting interactions, so the specific negative finding is not verified by this paper. (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites [43] Oughtred, R. et al. The BioGRID database: A comprehensive biomedical resource of curated protein, genetic, and chemical interactions. Protein Sci. 30, 187–200 (2021). sourcePartly supported

  The Protein and Genetic Interactions section describes BioGRID as a database of experimentally supported interactions that users can search by gene or protein. The paper does not report whether TMEM119–SELPLG or GAL3ST2–MUC-16 have supporting interactions, so our copy cannot verify the cited sentence’s specific finding. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

May combine two claims.

Of the 278 pQTLs, 34 (12.23%) were tested in the Folkersen et al. study; 29 (85.29%) were significant at Benjamini-Hochberg FDR < 0.05 and all significant pQTLs were directionally concordant.

“12.23% (34/278) pQTLs were tested in the Folkersen et al. study50 and 85.29% (29/34) of the pQTLs were significant (Benjamini–Hochberg FDR < 0.05) and all the significant pQTLs were also directionally concordant with the current study (Supplementary Table S2).”

• cites [50] Folkersen, L. et al. Mapping of 79 loci for 83 plasma protein biomarkers in cardiovascular disease. PLoS Genet. 13, e1006706 (2017). sourceSupported

May combine two claims.

Associations were identified between IL6R pQTLs and CAD, rheumatoid arthritis and 7 other disease traits, thereby supporting the findings of the study.

“We also identified associations between IL6R pQTLs and CAD, rheumatoid arthritis and 7 other disease traits (Supplementary Table S12), thereby supporting the findings of the study69.”

• cites [69] Interleukin-6 Receptor Mendelian Randomisation Analysis (IL6R MR) Consortium et al. The interleukin-6 receptor as a target for prevention of coronary heart disease: A mendelian randomisation analysis. Lancet Lond. Engl. 379, 1214–1224 (2012). sourcePartly supported

  The Results report an association between an IL6R variant and reduced coronary heart disease risk. The paper discusses rheumatoid arthritis in connection with trials of the IL6R-blocking drug tocilizumab, but does not report an association between IL6R variants and rheumatoid arthritis. The seven other traits are presented as the citing authors’ own findings, not as findings attributed to this source. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

May combine two claims.

Only one of those associations [proxy SNP rs10935473 with the CNV on chromosome 3 (98,410,653-98,414,807; deletion frequency = 0.651)] affecting FLT4/VEGF-sR3 levels, was identified in our study because the other proteins were not measured in our cohort.

“Only one of those associations [proxy SNP rs10935473 with the CNV on chromosome 3 (98,410,653–98,414,807; deletion frequency = 0.651)] affecting FLT4/VEGF-sR3 levels, was identified in our study because the other proteins were not measured in our cohort.”

• cites [4] Sun, B. B. et al. Genomic atlas of the human plasma proteome. Nature 558, 73–79 (2018). sourceCould not verify

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

  Experimental — accuracy not yet measured

May combine two claims.

Sun et al. fine mapped CSs in 127 (67.6%) genetic regions for 117 proteins, matching our findings.

“and Sun et al.9 fine mapped CSs in 127 (67.6%) genetic regions for 117 proteins, matching our findings.”

• cites [9] Sun, B. B. et al. Genetic regulation of the human plasma proteome in 54,306 UK Biobank participants. 2022.06.17.496443 Preprint at https://doi.org/10.1101/2022.06.17.496443 (2022). sourcePartly supported

  The ‘Identification and fine-mapping of independent signals’ section (page unavailable) says Sun et al. fine-mapped 95% credible sets for pQTL signals. It does not report the attributed figures of 127 regions, 67.6%, or 117 proteins; those specific figures are unsupported by our copy. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Previous studies indicate that few proteins are driven by rare variants.

“Previous studies indicate that few proteins are driven by rare variants11–13.”

• cites [11] Gilly, A. et al. Whole-genome sequencing analysis of the cardiometabolic proteome. Nat. Commun. 11, 6336 (2020). sourceSupported
• cites [12] Gilly, A. et al. Gene-based whole genome sequencing meta-analysis of 250 circulating proteins in three isolated European populations. Mol. Metab. 61, 101509 (2022). sourceSupported
• cites [13] Kierczak, M. et al. Contribution of rare whole-genome sequencing variants to plasma protein levels and the missing heritability. Nat. Commun. 13, 2532 (2022). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Recent findings support GDF-15's role as a biomarker of metabolic stress.

“and recent findings support its role as a biomarker of metabolic stress86.”

• cites [86] Lemmelä, S. et al. Integrated analyses of growth differentiation factor-15 concentration and cardiometabolic diseases in humans. eLife 11, e76272 (2022). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

JAKMIP1 is associated with type 2 diabetes and medications used to treat it.

“associated with type 2 diabetes and medications used to treat it82–84.”

• cites [82] Mahajan, A. et al. Refining the accuracy of validated target identification through coding variant fine-mapping in type 2 diabetes. Nat. Genet. 50, 559–571 (2018). sourceCould not verify

  The Results sections on coding-variant fine-mapping and T2D loci discuss associations for other genes, but the passages found do not mention JAKMIP1. Searches for JAKMIP1 and diabetes medications or treatment found no passage linking that gene to either T2D or its medications; the claim cannot be verified from our copy. (LLM, conf=0.72)

  Experimental — accuracy not yet measured
• cites [83] Wu, Y. et al. Genome-wide association study of medication-use and associated disease in the UK Biobank. Nat. Commun. 10, 1891 (2019). sourceCould not verify

  The Results section reports associations involving diabetes medications (A10), and the drug-target analysis names IDE as a gene meriting further study for type 2 diabetes. A search for JAKMIP1 in our copy found no passage linking that gene to type 2 diabetes or its medications, so the specific claim cannot be verified. (LLM, conf=0.75)

  Experimental — accuracy not yet measured
• cites [84] Vujkovic, M. et al. Discovery of 318 new risk loci for type 2 diabetes and related vascular outcomes among 1.4 million participants in a multi-ancestry meta-analysis. Nat. Genet. 52, 680–691 (2020). sourceCould not verify

  The Results section on gene–drug relationships discusses type 2 diabetes loci and drug interactions, but does not identify JAKMIP1. Searches for JAKMIP1 and its proposed association with diabetes medications found no passage in our copy addressing that gene. (LLM, conf=0.72)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Kierczak et al. used a maximum MAF threshold 0.0239, whereas we used a standard conservative threshold of 0.01.

“they used a maximum MAF threshold 0.0239, whereas we used a standard conservative threshold of 0.01.”

• cites [13] Kierczak, M. et al. Contribution of rare whole-genome sequencing variants to plasma protein levels and the missing heritability. Nat. Commun. 13, 2532 (2022). sourcePartly supported

  The Results (SKAT analyses) and Methods (Statistical analyses) state that 0.0239 was the cutoff between rare and common variants in the CommonRare analyses. It was not a maximum MAF for the study overall: the Methods say the primary SKAT analyses included all variants passing quality control, regardless of MAF. The citing authors’ own 0.01 threshold is not a claim about Kierczak et al. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

May combine two claims.

A positive colocalisation finding typically implies a non-zero MR estimate, but the reverse is not generally true.

“As they have highlighted, a positive colocalisation finding typically implies a non-zero MR estimate, the reverse is not generally true17.”

• cites [17] Zuber, V. et al. Combining evidence from Mendelian randomization and colocalization: Review and comparison of approaches. Am. J. Hum. Genet. 109, 767–782 (2022). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

The FGF5 locus has been linked to cardiovascular conditions in previous GWASs.

“The FGF5 locus has been linked to cardiovascular conditions in previous GWASs67,75.”

• cites [67] van der Harst, P. & Verweij, N. Identification of 64 novel genetic loci provides an expanded view on the genetic architecture of coronary artery disease. Circ. Res. 122, 433–443 (2018). sourceCould not verify

  The Results section, ‘Insights in Loci by Associations With Other Phenotypes’ (page unavailable), reports prior GWAS associations for several loci, but does not identify FGF5 among them. A search of our copy for ‘FGF5’ found no passage linking that locus to cardiovascular conditions. (LLM, conf=0.72)

  Experimental — accuracy not yet measured
• cites [75] Nikpay, M. et al. A comprehensive 1,000 Genomes-based genome-wide association meta-analysis of coronary artery disease. Nat. Genet. 47, 1121–1130 (2015). sourceCould not verify

  The Scanning for additive associations section discusses previously reported coronary artery disease loci, but the passages found do not identify FGF5 among them. Searches for “FGF5” and “fibroblast growth factor 5” found no passage linking that locus to a cardiovascular condition in a previous GWAS; the claim therefore cannot be verified from our copy. (LLM, conf=0.73)

  Experimental — accuracy not yet measured

May combine two claims.

Rare genetic variants tend to have greater population specificity, making replication of findings from rare variant analyses more difficult.

“Rare genetic variants tend to have greater population specificity87, making replication of findings from rare variant analyses more difficult.”

• cites [87] Momozawa, Y. & Mizukami, K. Unique roles of rare variants in the genetics of complex diseases in humans. J. Hum. Genet. 66, 11–23 (2021). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Only a limited number of pQTL studies have conducted fine mapping as a post-GWAS analysis.

“Only a limited number of pQTL studies have conducted fine mapping9,71 as one of the post-GWAS analyses.”

• cites [9] Sun, B. B. et al. Genetic regulation of the human plasma proteome in 54,306 UK Biobank participants. 2022.06.17.496443 Preprint at https://doi.org/10.1101/2022.06.17.496443 (2022). sourceSupported
• cites [71] Zhang, J. et al. Plasma proteome analyses in individuals of European and African ancestry identify cis-pQTLs and models for proteome-wide association studies. Nat. Genet. 54, 593–602 (2022). sourceSupported

Classified as probably not a substantial claim.

The study applied MR to significant colocalisation events to assess causality, a strategy recommended by Zuber et al.

“We then applied MR to significant colocalisation events to assess causality, a strategy recommended by Zuber et al.17.”

• cites [17] Zuber, V. et al. Combining evidence from Mendelian randomization and colocalization: Review and comparison of approaches. Am. J. Hum. Genet. 109, 767–782 (2022). sourcePartly supported

  The Discussion recommends that positive cis-MR analyses be accompanied by colocalization, supporting the use of the two methods together. It does not recommend the specific strategy attributed here—selecting significant colocalization events and then applying MR to them. Page numbers are unavailable in our copy. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Pietzner et al. recently detected a significant colocalisation of 50.1% of the cis-pQTLs with corresponding gene eQTLs using GTEx.

“Similar to our findings, Pietzner et al.7 recently detected a significant colocalisation of 50.1% of the cis-pQTLs with corresponding gene eQTLs using GTEx.”

• cites [7] Pietzner, M. et al. Mapping the proteo-genomic convergence of human diseases. Science 374, eabj1541 (2021). sourcePartly supported

  The Results section (‘Shared genetic architecture with gene expression and splicing’) reports GTEx v8 colocalization and a shared gene-expression signal for 50.1% of 1,584 protein targets with at least one cis-pQTL. The cited percentage is therefore for protein targets, not explicitly for individual cis-pQTLs as the claim states. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

May combine two claims.

Of the cis associations, 25.5% (2021/7936) colocalised with eQTLs for the corresponding protein-encoding gene from the full eQTL Catalogue, while for the GTEx dataset alone it was 54.3% (273/503).

“Of the cis associations, 25.5% (2021/7936) colocalised with the eQTLs for the corresponding protein-encoding gene from the full eQTL Catalogue, while for the GTEx dataset alone it was 54.3% (273/503).”

• cites [18] Kerimov, N. et al. A compendium of uniformly processed human gene expression and splicing quantitative trait loci. Nat. Genet. 53, 1290–1299 (2021). sourceCould not verify

  The Methods (Co-localization; page unavailable) describe analyses using the eQTL Catalogue and GTEx, but not co-localization of cis associations with eQTLs for corresponding protein-encoding genes. The Results (Detection of additional co-localizations missed in GTEx; page unavailable) report different counts for GWAS loci. Searches for 2021/7936, 273/503, and protein-encoding-gene cis associations found no passage supporting the quoted percentages. (LLM, conf=0.76)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Gene-based pQTL analyses of rare variants constitute an emerging approach.

“Gene-based pQTL analyses of rare variants constitute an emerging approach10–13”

• cites [10] Solomon, T. et al. Identification of common and rare genetic variation associated with plasma protein levels using whole-exome sequencing and mass spectrometry. Circ. Genom. Precis. Med. 11, e002170 (2018). sourceSupported
• cites [11] Gilly, A. et al. Whole-genome sequencing analysis of the cardiometabolic proteome. Nat. Commun. 11, 6336 (2020). sourceSupported
• cites [12] Gilly, A. et al. Gene-based whole genome sequencing meta-analysis of 250 circulating proteins in three isolated European populations. Mol. Metab. 61, 101509 (2022). sourceSupported
• cites [13] Kierczak, M. et al. Contribution of rare whole-genome sequencing variants to plasma protein levels and the missing heritability. Nat. Commun. 13, 2532 (2022). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The lack of replication of Png et al. associations may be due to partial overlap between assayed protein sets, cohort differences (European ancestry vs a Greek population isolate with population-specific CNVs), and differences in CNV detection approach.

“possibly because there was only a partial overlap between the assayed protein sets, differences between cohorts (European ancestry vs a Greek population isolate with population-specific CNVs)76, and differences in the approach used for CNV detection.”

• cites [15] Png, G. et al. Population-wide copy number variation calling using variant call format files from 6,898 individuals. Genet. Epidemiol. 44, 79–89 (2020). sourceSupported
• cites [76] Panoutsopoulou, K. et al. Genetic characterization of Greek population isolates reveals strong genetic drift at missense and trait-associated variants. Nat. Commun. 5, 5345 (2014). sourcePartly supported

  The Abstract and Discussion identify the studied Greek cohorts as genetic isolates, supporting the cohort distinction for which this citation is placed. Our copy describes drifted SNPs and missense variants, but does not establish that these cohorts have population-specific CNVs. The citation is attached to the cohort comparison, not the separate explanations about protein-set overlap or CNV detection. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Most pQTL studies have been conducted using serum or plasma measurements from blood samples, and only a limited number of studies has involved examination of liver and brain tissue-specific pQTLs.

“Second, most pQTL studies have been conducted using serum or plasma measurements from blood samples4,6,8,51 and only a limited number of studies has involved the examination of liver and brain tissue–specific pQTLs88,89.”

• cites [4] Sun, B. B. et al. Genomic atlas of the human plasma proteome. Nature 558, 73–79 (2018). sourceSupported
• cites [6] Folkersen, L. et al. Genomic and drug target evaluation of 90 cardiovascular proteins in 30,931 individuals. Nat. Metab. 2, 1135–1148 (2020). sourceSupported
• cites [8] Ferkingstad, E. et al. Large-scale integration of the plasma proteome with genetics and disease. Nat. Genet. 53, 1712–1721 (2021). sourceNot checked — no full text of the source
• cites [51] Suhre, K. et al. Connecting genetic risk to disease end points through the human blood plasma proteome. Nat. Commun. 8, 14357 (2017). sourceSupported
• cites [88] He, B., Shi, J., Wang, X., Jiang, H. & Zhu, H.-J. Genome-wide pQTL analysis of protein expression regulatory networks in the human liver. BMC Biol. 18, 97 (2020). sourceSupported
• cites [89] Robins, C. et al. Genetic control of the human brain proteome. Am. J. Hum. Genet. 108, 400–410 (2021). sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Kierczak et al. detected cis-region rare variant associations for four proteins (CTSZ, CYR61, GDF-15 and PON3) and trans associations of rare GAL3ST2 variants affecting the MUC16 level.

“Kierczak et al.13 detected cis-region rare variant associations for four proteins (CTSZ, CYR61, GDF-15 and PON3) and trans associations of rare GAL3ST2 variants affecting the MUC16 level”

• cites [13] Kierczak, M. et al. Contribution of rare whole-genome sequencing variants to plasma protein levels and the missing heritability. Nat. Commun. 13, 2532 (2022). sourcePartly supported

  The Discussion (page unavailable) identifies rare variants as likely contributors to a CTSZ association, supporting that part of the claim. The retrieved passages do not establish the claimed cis associations for CYR61, GDF-15, or PON3. The Results section on rare-variant trans associations (page unavailable) lists six proteins with significant associations, none of them MUC16, and does not identify GAL3ST2. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

May combine two claims.

Our CSs for cis associations contained an average of 15.76 variants, whereas Zhang et al. used imputed genotyping data and reported an average of 21.29 variants.

“Our CSs for cis associations contained an average of 15.76 variants, whereas Zhang et al. used imputed genotyping data and reported an average of 21.29 variants71.”

• cites [71] Zhang, J. et al. Plasma proteome analyses in individuals of European and African ancestry identify cis-pQTLs and models for proteome-wide association studies. Nat. Genet. 54, 593–602 (2022). sourcePartly supported

  The Methods section (‘Plasma protein data and genetic data’; page unavailable) says the ARIC genotypes were imputed to the TOPMed reference panel. The Results (‘Fine Mapping’; page unavailable) report an average of 21.29 variants per 95% credible set for the European-ancestry group, but 12.11 for the African-ancestry group. The cited sentence omits that 21.29 applies specifically to the European-ancestry group, not to the study overall. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

FGF5 plays essential roles in the regulation of cell proliferation, including in cardiac myocytes, and cell differentiation.

“For example, FGF5 plays essential roles in the regulation of cell proliferation, including in cardiac myocytes, and cell differentiation73”

• cites [73] Ornitz, D. M. et al. Receptor specificity of the fibroblast growth factor family. J. Biol. Chem. 271, 15292–15297 (1996). sourcePartly supported

  The paper reports testing FGF5’s mitogenic activity in FGFR-expressing BaF3 cells and states in the Discussion that FGF5 activates certain receptor splice forms. That supports a limited connection to cell proliferation, but our copy does not establish an essential role for FGF5 in cardiac-myocyte proliferation or cell differentiation. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

May combine two claims.

The study detected a cis signal for the FGF5 level and associated variants in the region, which overlapped with previous GWAS findings for cardiovascular diseases and medications used to treat them.

“We detected a cis signal for the FGF5 level and associated variants in the region, which overlapped with previous GWAS findings for cardiovascular diseases and medications used to treat them.”

• cites [67] van der Harst, P. & Verweij, N. Identification of 64 novel genetic loci provides an expanded view on the genetic architecture of coronary artery disease. Circ. Res. 122, 433–443 (2018). sourcePartly supported

  The Results section, “Insights in Loci by Associations With Other Phenotypes” (page unavailable), reports CAD-associated loci and their overlap with previously reported GWAS variants, including variants associated with blood-pressure traits. It does not identify an FGF5-region overlap or associations with medications used to treat cardiovascular disease. The citing authors’ statement that they detected an FGF5 cis signal is their own finding, not a finding attributed to this source. (LLM, conf=0.77)

  Experimental — accuracy not yet measured
• cites [75] Nikpay, M. et al. A comprehensive 1,000 Genomes-based genome-wide association meta-analysis of coronary artery disease. Nat. Genet. 47, 1121–1130 (2015). sourceCould not verify

  The Scanning for additive associations section reports CAD-associated variants, and the Discussion mentions drugs used to lower CAD risk. Neither identifies an FGF5-region signal or shows that its variants overlap GWAS findings for cardiovascular disease or its medications. Searches for FGF5 and the claimed medication overlap found no passage addressing that specific point; page numbers were unavailable in our copy. (LLM, conf=0.74)

  Experimental — accuracy not yet measured

May combine two claims.

The study identified 184 cis and 94 trans signals for 157 protein traits, which were further fine-mapped to credible sets for 101 cis and 87 trans signals for 151 proteins.

“We identified 184 cis and 94 trans signals for 157 protein traits, which were further fine-mapped to credible sets for 101 cis and 87 trans signals for 151 proteins.”

May combine two claims.

Rare genetic variation contributed to the levels of 7 proteins, with 5 cis and 14 trans associations.

“Rare genetic variation contributed to the levels of 7 proteins, with 5 cis and 14 trans associations.”

May combine two claims.

The ABO locus was a trans-signal hotspot, with intronic variants additionally associated with ICAM2, galectin-4 (LGALS4), PODXL and LIFR protein levels.

“This locus was a trans-signal hotspot, with intronic variants additionally associated with the ICAM2, galectin-4 (LGALS4), PODXL and LIFR protein levels.”

May combine two claims.

Those connections were based mainly on physical interactions (77.64%) and co-expression (8.01%).

“Those connections were based mainly on physical interactions (77.64%) and co-expression (8.01%).”

May combine two claims.

IL6R had a total of 11 significant colocalisations, which included colocalisations with CAD as well as immunological conditions such as asthma.

“IL6R had a total of 11 significant colocalisations, which included colocalisations with CAD as well as immunological conditions such as asthma.”

May combine two claims.

The deletion in the 3q12.1 intergenic region (chromosome 3: 98,410,653-98,414,807 bp; frequency = 0.651) acted as a hub, having multiple trans associations with protein levels: ICAM2 (P = 1.31 × 10-29), FLT4 (P = 2.34 × 10-24), PDCD1LG2 (P = 2.88 × 10-15) and IL1R1 (P = 8.19 × 10-8).

“For example, the deletion in the 3q12.1 intergenic region (chromosome 3: 98,410,653–98,414,807 bp; frequency = 0.651) acted as a hub, having multiple trans associations with protein levels: ICAM2 (P = 1.31 × 10–29), FLT4 (P = 2.34 × 10–24), PDCD1LG2 (P = 2.88 × 10–15) and IL1R1 (P = 8.19 × 10–8).”

May combine two claims.

Six proteins (CD6, PRSS27, CEACAM5, CD40, TNFRSF6B and IL1RL1) had significant colocalisations with eQTLs from brain tissue, but no evidence of shared conditions with direct effects on the brain tissue in the PheWAS.

“Six of these proteins (CD6, PRSS27, CEACAM5, CD40, TNFRSF6B and IL1RL1) had significant colocalisations with eQTLs from brain tissue, but no evidence of shared conditions with direct effects on the brain tissue in the PheWAS.”

May combine two claims.

Eleven variants (all cis) were previously established eQTLs and had evidence for transcription factor binding- and/or DNase peak-related functions.

“Eleven variants (all cis) were previously established eQTLs and had evidence for transcription factor binding– and/or DNase peak–related functions.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The 95% credible sets contained an average of 15.7 variants (cis sets, 15.76; trans sets, 15.6).

“The 95% CSs contained an average of 15.7 variants (cis sets, 15.76; trans sets, 15.6).”

May combine two claims.

Cis colocalisations consisted of 2021 (25.5%) cases in which colocalising eQTLs and pQTLs affected the same gene product and 5915 (74.5%) cases in which the colocalising loci affected different gene products in the cis regions.

“Colocalisations classified as cis consisted of 2021 (25.5%) cases in which colocalising eQTLs and pQTLs affected the same gene product and 5915 (74.5%) cases in which the colocalising loci affected different gene products in the cis regions.”

May combine two claims.

For the MICA-MICB dimer pQTL, associations were detected between CNV on chromosome 6 (31,292,078-31,293,977 bp; deletion frequency = 0.876) and medium HDL triglycerides (P = 8.82 × 10-5), and between a CNV on chromosome 6 (31,337,848-31,341,642 bp; deletion frequency = 0.074) and lower-limb oedema (ICD-10 code R60; P = 9.06 × 10-5).

“For the MICA-MICB dimer pQTL, associations were detected between CNV on chromosome 6 (31,292,078–31,293,977 bp; deletion frequency = 0.876) and medium HDL triglycerides (P = 8.82 × 10–5), and between a CNV on chromosome 6 (31,337,848–31,341,642 bp; deletion frequency = 0.074) and lower-limb oedema (ICD-10 code R60; P = 9.06 × 10–5).”

May combine two claims.

TNFRSF6B and SULT1A1 colocalised with inflammatory bowel disease, and TNFRSF6B also separately colocalised with its two main forms: Crohn's disease and ulcerative colitis.

“TNFRSF6B and SULT1A1 colocalised with inflammatory bowel disease, and TNFRSF6B also separately colocalised with its two main forms: Crohn’s disease and ulcerative colitis.”

May combine two claims.

For MICA/B, ten independent signals were identified in the cis region of one subunit on chromosome 6 (strongest signal for intronic variant of MICA: rs3132467; MAF = 0.30, P = 3.04 × 10-68) and two trans associations.

“For MICA/B, we identified ten independent signals in the cis region of one subunit on chromosome 6 (the strongest signal was for an intronic variant of MICA: rs3132467; MAF = 0.30, P = 3.04 × 10–68) and two trans associations.”

May combine two claims.

Associations for CXCL1, CCL3 and VEGFA proteins were validated by multiple independent arrays, with the same genetic regions reaching genome-wide significance and showing concordant effect directions.

“Associations for the CXCL1, CCL3 and VEGFA proteins were validated by multiple independent arrays, in which the same genetic regions reached genome-wide significance and showed concordant effect directions.”

May combine two claims.

The variant with the most (n = 10) histone modifications was rs10415777, a cis-pQTL for OSCAR.

“The variant with the most (n = 10) histone modifications was rs10415777, a cis-pQTL for OSCAR.”

May combine two claims.

eQTL analysis indicated that the deletion was also associated with SIRPB1, but not SIRPA, expression (P = 3.5 × 10-87).

“eQTL analysis indicated that the deletion was also associated with SIRPB1, but not SIRPA, expression (P = 3.5 × 10–87).”

May combine two claims.

This locus was also an eQTL for NAIP (P = 6.4 × 10-48), but not for IL18 expression (P > 0.001).

“This locus was also an eQTL for NAIP (P = 6.4 × 10–48), but not for IL18 expression (P > 0.001).”

Probably not a substantial claim: may not pass S3 (load-bearing).

Due to tissue availability, 78.7% (4906/6236) of the identified meQTLs were from whole blood studies.

“Methylation data originates from five tissues: cord blood, monocytes, neutrophils, T cells and whole blood; due to tissue availability, 78.7% (4906/6236) of the identified meQTLs were from whole blood studies.”

May combine two claims.

The largest number of colocalisations was found for pQTLs of the MICA-MICB heterodimer (n = 6583), followed by OSCAR (n = 1207) and ACP5 (n = 1105) pQTLs.

“The largest number of colocalisations was found for pQTLs of the MICA-MICB heterodimer (n = 6583), followed by OSCAR (n = 1207) and ACP5 (n = 1105) pQTLs.”

May combine two claims.

We also detected moderate correlation between IL-18 protein expression and NAIP gene expression (Spearman's R = 0.17); Spearman correlation coefficient between IL-18 protein and gene expressions was 0.05.

“We also detected moderate correlation between IL-18 protein expression and NAIP gene expression (Spearman’s R = 0.17); Spearman correlation coefficient between IL-18 protein and gene expressions was 0.05.”

May combine two claims.

The majority [n = 158 (52.3%)] of the mQTLs discovered were for the APOE missense variant rs7412, which had a trans association with the level of LDLR.

“The majority [n = 158 (52.3%)] of the mQTLs discovered were for the APOE missense variant rs7412, which had a trans association with the level of LDLR.”

May combine two claims.

Strong trans associations were between rs8176671 and CDH5 level (16q21; MAF = 0.19, P = 8.83 × 10-40) and between deletion rs8176643 and SELE level (1q24.2; MAF = 0.18, P = 7.98 × 10-36); both variants are intronic for the 9q34.2 locus of the ABO gene.

“Strong trans associations were between the rs8176671 and the CDH5 level (16q21; MAF = 0.19, P = 8.83 × 10–40) as well as between the deletion rs8176643 and the SELE level (1q24.2; MAF = 0.18, P = 7.98 × 10–36); both of these variants are intronic variants for the 9q34.2 locus of the ABO gene.”

May combine two claims.

IL1RL1 pQTLs colocalised with IL1RL1, IL18R1 and IL18RAP eQTLs detected in multiple cell types with direct effects on the immune system (e.g. T-cells); these variants were associated with asthma and allergic reactions in the PheWAS.

“As another example, IL1RL1 pQTLs colocalised with IL1RL1, IL18R1 and IL18RAP eQTLs detected in multiple cell types with direct effects on the immune system (e.g. T-cells; Supplementary Table S9); these variants were associated with asthma and allergic reactions in the PheWAS.”

May combine two claims.

ICAM2 and FLT4 had trans associations with rs12493830 on chromosome 3 and a CNV (chromosome 3: 98,410,653-98,414,807 bp) in the same intergenic region, separated by 3859 bp.

“For example, ICAM2 and FLT4 had trans associations with rs12493830 on chromosome 3 and a CNV (chromosome 3: 98,410,653–98,414,807 bp) in the same intergenic region, separated by 3859 bp.”

May combine two claims.

The detected associations included 278 independent pQTL variants, 184 (66.2%) cis and 94 (33.8%) trans, and 9.71% of them were indels.

“The detected associations included 278 independent pQTL variants [184 (66.2%) cis and 94 (33.8%) trans], 9.71% of which were indels.”

May combine two claims.

A frameshift-causing insertion rs139130389 was associated with FOLR3 level (MAF = 0.12, P = 3.91 × 10-91), with four independent signals in the FOLR3 cis region.

“and a frameshift-causing insertion rs139130389 and the FOLR3 level (MAF = 0.12, P = 3.91 × 10–91) association, with four independent signals in the FOLR3 cis region.”

May combine two claims.

Thirty-three (32.7%) out of 101 cis regions were fine-mapped to more than one signal (mean, 1.4 signals/region), with the CCL24 cis region having the largest number of independent credible sets (n = 5).

“Thirty-three (32.7%) out of 101 cis regions were fine-mapped to more than one signal (mean, 1.4 signals/region), with the CCL24 cis region having the largest number of independent CSs (n = 5).”

May combine two claims.

For 55 proteins with cis-pQTLs, 503 (68.1%) colocalising eQTLs were identified; for 22 proteins with trans-pQTLs, 236 (31.9%) colocalising eQTLs were identified.

“For 55 proteins with cis-pQTLs, 503 (68.1%) colocalising eQTLs were identified; for 22 proteins with trans-pQTLs, 236 (31.9%) colocalising eQTLs were identified.”

May combine two claims.

Trans-pQTLs for the CTRC and TEK proteins were in the ABO locus and colocalised with ABO eQTLs; in the PheWAS, they were linked to multiple self-reported diagnoses (e.g. 'blood clot in the leg') from the UK Biobank sample, and to haematological traits.

“Trans-pQTLs for the CTRC and TEK proteins were in the ABO locus and colocalised with ABO eQTLs; in the PheWAS, they were linked to multiple self-reported diagnoses (e.g. ‘blood clot in the leg') from the UK Biobank sample, and to haematological traits.”

May combine two claims.

The CNV was also associated with the expression of VSTM1 itself (P = 1.81 × 10−39) and both gene-protein expression pairs showed moderate correlation (OSCAR-OSCAR, Spearman's R = 0.32; VSTM1-OSCAR, Spearman's R = 0.34).

“However, the CNV was also associated with the expression of VSTM1 itself (P = 1.81 × 10−39) and both gene–protein expression pairs showed moderate correlation (OSCAR–OSCAR, Spearman’s R = 0.32; VSTM1–OSCAR, Spearman’s R = 0.34).”

May combine two claims.

LDLR, TNFRSF11B, TNFRSF6B, WISP1, CXCL1 and PLAU proteins showed significant pQTL effects but yielded no credible set.

“LDLR, TNFRSF11B, TNFRSF6B, WISP1, CXCL1 and PLAU proteins showed significant pQTL effects but yielded no CS.”

May combine two claims.

A 19q13.42 deletion (frequency = 0.291) overlapping the VSTM1 gene was an eQTL and a pQTL for nearby gene OSCAR (P = 1.77 × 10-14 and P = 5.64 × 10-9, respectively).

“Similarly, we determined that a 19q13.42 deletion (frequency = 0.291) overlapping the VSTM1 gene was an eQTL and a pQTL for nearby gene OSCAR (P = 1.77 × 10–14 and P = 5.64 × 10–9, respectively).”

May combine two claims.

Cis and trans pairs were specific to 73 and 26 proteins, respectively, and 6 proteins (IL1R2, TEK, MIA, FCRLB, PDCD1LG2 and MICA-MICB) had colocalisations for both cis and trans associations.

“Cis and trans pairs were specific to 73 and 26 proteins, respectively, and 6 proteins (IL1R2, TEK, MIA, FCRLB, PDCD1LG2 and MICA-MICB) had colocalisations for both cis and trans associations.”

May combine two claims.

We detected 12 significant (Bonferroni-corrected P-value threshold 1.12 × 10-7) pQTL associations between CNVs and plasma protein levels (7 cis and 5 trans, 11 proteins), with two cis associations detected for the MICA-MICB heterodimer.

“We detected 12 significant (Bonferroni-corrected P-value threshold 1.12 × 10–7) pQTL associations between CNVs and plasma protein levels (7 cis and 5 trans, 11 proteins; Supplementary Table S16), with two cis associations detected for the MICA-MICB heterodimer.”

May combine two claims.

Gene expression data was available for 109 proteins with 201 pQTLs, including two heterodimers with two subunits encoding the protein.

“Gene expression data was available for 109 proteins with 201 pQTLs, including two heterodimers with two subunits encoding the protein.”

May combine two claims.

Nominally significant associations were detected for a CNV on chromosome 19 (41,381,588-41,387,347 bp, deletion frequency = 0.054 and duplication frequency = 0.022) with the pQTL of the MIA protein level (P = 2.38 × 10-6) and migraine (ICD-10 code G43; P = 3.14 × 10-5).

“Additionally, we detected nominally significant associations for a CNV on chromosome 19 (41,381,588–41,387,347 bp, deletion frequency = 0.054 and duplication frequency = 0.022) with the pQTL of the MIA protein level (P = 2.38 × 10–6) and migraine (ICD-10 code G43; P = 3.14 × 10–5).”

May combine two claims.

Additional strong cis associations included rs1569960 with SIRPA level (MAF = 0.34, P = 2.67 × 10-106), with four independent signals in the SIRPA cis region.

“Additional strong cis associations included the rs1569960 and the SIRPA level (MAF = 0.34, P = 2.67 × 10–106) association, with four independent signals in the SIRPA cis region;”

May combine two claims.

In SNP pQTL fine mapping, we detected two independent CSs, overlapping SIRPB1 [variant with the largest posterior inclusion probability (PIP) = 0.295] and at SIRPA (variant with the largest PIP = 0.242).

“In SNP pQTL fine mapping, we detected two independent CSs, overlapping SIRPB1 [variant with the largest posterior inclusion probability (PIP) = 0.295] and at SIRPA (variant with the largest PIP = 0.242; Fig. 5a).”

Probably not a substantial claim: may not pass S3 (load-bearing).

Fourteen (4.3%) of the proteins were measured in multiple arrays.

“Fourteen (4.3%) of the proteins were measured in multiple arrays.”

May combine two claims.

Cis-pQTLs colocalising with eQTLs were detected in 49 tissues, and trans-pQTLs colocalising with eQTLs were identified in 46 tissues (not in Epstein-Barr virus-transformed lymphocytes or uterine or vaginal tissue).

“Cis-pQTLs colocalising with eQTLs were detected in 49 tissues, and trans-pQTLs colocalising with eQTLs were identified in 46 tissues (not in Epstein-Barr virus–transformed lymphocytes or uterine or vaginal tissue).”

May combine two claims.

Of the 184 cis associations detected for 104 proteins, 31 (16.85%) were with protein-altering primary lead cis-pQTL variants and an additional 5 were with cis-pQTL variants in high LD with PAVs.

“Of the 184 cis associations detected for 104 proteins, 31 (16.85%) were with protein-altering primary lead cis-pQTL variants and an additional 5 were with cis-pQTL variants in high LD with PAVs (Supplementary Table S2).”

May combine two claims.

Detected colocalisations were 56.4% for cis- and 43.6% for trans-pQTLs.

“Detected colocalisations were 56.4% for cis- and 43.6% for trans-pQTLs.”

May combine two claims.

The study generated a comprehensive pQTL resource and interpreted it using eQTL and publicly available GWAS data.

“In conclusion, we have generated a comprehensive pQTL resource and interpreted it by using eQTL, as well as publicly available GWAS data.”

May combine two claims.

A CNV on chromosome 5 overlapping with NAIP has trans effects on the IL18 protein level and a cis effect on the NAIP gene expression level, but no significant effect on IL18 gene expression.

“We found that a CNV on chromosome 5 overlapping with NAIP has trans effects on the IL18 protein level and a cis effect on the NAIP gene expression level, but there is no significant effect on the IL18 gene expression.”

May combine two claims.

A deletion in the same locus overlapping SIRPB1 affects the SIRPA level and (more strongly) SIRPB1 gene expression.

“and a deletion in the same locus overlapping SIRPB1 and affecting the SIRPA level and (more strongly) SIRPB1 gene expression”

Classified as probably not a substantial claim. May combine two claims.

All data included in the PheWAS were obtained using the lm function with custom R scripts from 2115 unrelated Estonian Biobank samples for which WGS data were available, and were corrected for age, sex and six genotype principal components calculated from common SNPs.

“All data included in the PheWAS were obtained using the lm function with custom R scripts from 2115 unrelated Estonian Biobank samples for which WGS data were available, and were corrected for age, sex and six genotype principal components (PCs; calculated from common SNPs).”

Classified as probably not a substantial claim.

The genome-wide significance threshold for CNV pQTL analysis was set to 1.12 × 10-7 (0.05/2465/181).

“The genome-wide significance threshold was set to 1.12 × 10–7 (0.05/2465/181).”

May combine two claims.

The dataset had 80% power to detect pQTL effects explaining at least 7.6% of variance in protein expression at P < 5 × 10-8 and at least 9.7% at P < 2.76 × 10-10.

“The power calculations suggested that our dataset had 80% power to detect the pQTL effects that explained at least 7.6% and 9.7% of the variance in protein expression (when using P < 5 × 10–8 and P < 2.76 × 10–10 respectively, Supplementary Fig. S1, Supplementary Table S5).”

Probably not a substantial claim: may not pass S3 (load-bearing).

The LTA and MICA-MICB proteins associated with variants in the major histocompatibility complex region on chromosome 6 were excluded due to the complexity of the associated HLA region.

“We excluded the LTA and MICA-MICB proteins associated with variants in the major histocompatibility complex region on chromosome 6, due to the complexity of the associated HLA region.”

Classified as probably not a substantial claim. May combine two claims.

The PheWAS significance threshold was set to P < 0.05/420, as 420 PCs calculated on all included phenotypes explained 95% of the variability.

“The PheWAS significance threshold was set to P < 0.05/420, as 420 PCs calculated on all included phenotypes explained 95% of the variability.”

Classified as probably not a substantial claim. May combine two claims.

The study excluded multiallelic sites and genetic variants based on quality/depth < 2, Hardy-Weinberg equilibrium test failure (P > 1 × 10-9), and call rate < 90%.

“We excluded multiallelic sites and genetic variants, based on quality/depth < 2, Hardy–Weinberg equilibrium test failure (P > 1 × 10–9), and call rate < 90%.”

Classified as probably not a substantial claim.

To account for population structure, a kinship matrix was generated in EPACTS using genetic variants with MAF > 0.01 and call rate > 95%.

“To account for population structure, a kinship matrix was generated in EPACTS using genetic variants with MAF > 0.01 and call rate > 95%.”

Classified as probably not a substantial claim.

To discard rare CNV events, all CNV sites with in-sample frequencies of the most frequent copy number > 0.95 were excluded.

“To discard rare CNV events, all CNV sites with in-sample frequencies of the most frequent copy number > 0.95 were excluded.”

Classified as probably not a substantial claim. May combine two claims.

To account for multiple testing, P-values were adjusted using false discovery rate correction; pQTLs were considered replicated at Benjamini-Hochberg FDR ≤ 0.05 and concordant allelic direction with the discovery pQTLs.

“To account for multiple testing, we adjusted P-values using false discovery rate (FDR) correction; pQTLs were considered as replicated at Benjamini–Hochberg FDR ≤ 0.05 and concordant allelic direction with the discovery pQTLs.”

Classified as probably not a substantial claim.

The effects of age, sex, season of sample collection, smoking status, blood sample processing time, plasma sample storage time and protein analysis batch were regressed out using a custom R script.

“We regressed out the effects of age, sex, the season of sample collection, smoking status, blood sample processing time (days), plasma sample storage time (in days) and protein analysis batch using a custom R script.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

A recent landmark study conducted on the Icelandic population revealed 18,084 genetic associations with protein levels, 19% of which were with rare variants.

“a recent landmark study conducted on the Icelandic population revealed 18,084 genetic associations with protein levels, 19% of which were with rare variants8”

• cites [8] Ferkingstad, E. et al. Large-scale integration of the plasma proteome with genetics and disease. Nat. Genet. 53, 1712–1721 (2021). sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

To ensure that each protein was represented by a single association, comparisons were restricted to instances where either one subunit or the entire heterodimer complex was available.

“In order to ensure that each protein was represented by a single association, we restricted our comparisons to instances where either one subunit or the entire heterodimer complex was available.”

Classified as probably not a substantial claim.

Only samples that passed quality control on every panel (n = 478) were included.

“Only samples that passed quality control on every panel (n = 478) were included.”

Classified as probably not a substantial claim. May combine two claims.

To correct for multiple testing, P-values were adjusted using false discovery rate correction; results were considered significant at Benjamini-Hochberg FDR ≤ 0.05.

“To correct for multiple testing, we adjusted P-values using false discovery rate (FDR) correction; results were considered significant at Benjamini–Hochberg FDR ≤ 0.05.”

Classified as probably not a substantial claim. May combine two claims.

For rare variant eQTL analysis, P-values were adjusted using false discovery rate correction; rare variant eQTLs were considered replicated at Benjamini-Hochberg FDR ≤ 0.05 and directionally concordant with the rare variant pQTLs.

“Similar to single variant eQTL analysis, to account for multiple testing, we adjusted P-values using false discovery rate (FDR) correction; rare variant eQTLs were considered as replicated at Benjamini–Hochberg FDR ≤ 0.05 and directionally concordant with the rare variant pQTLs.”

Classified as probably not a substantial claim. May combine two claims.

Low-quality CNV calls and sites with call rates < 90% were excluded.

“Low-quality calls and sites with call rates < 90% were excluded.”

Classified as probably not a substantial claim.

The eQTL results were corrected for multiple testing and a Bonferroni-corrected threshold of P < 1.61 × 10-9 [0.05/(2465 CNVs × 12,619 genes)] was applied.

“The eQTL results were corrected for multiple testing and a Bonferroni-corrected threshold of P < 1.61 × 10–9 [0.05/(2465 CNVs × 12,619 genes)] was applied.”

Classified as probably not a substantial claim.

Age, sex, body mass index, blood components and RNA sequencing batch were included as covariates in the eQTL analysis.

“We included age, sex, body mass index, blood components (neutrophils, eosinophils, basophils, lymphocytes, monocytes, erythrocytes and thrombocytes) and RNA sequencing batch as covariates.”

Classified as probably not a substantial claim.

For each significantly associated CNV, all SNP markers within a 500-kbp proximity were tested for potential tagging effects.

“For each significantly associated CNV, all SNP markers within a 500-kbp proximity were tested for potential tagging effects.”

Classified as probably not a substantial claim. May combine two claims.

To correct for multiple testing, P-values were adjusted using false discovery rate correction; eQTLs were considered replicated at Benjamini-Hochberg FDR ≤ 0.05 and with concordant allelic direction with the pQTLs.

“To correct for multiple testing, we adjusted P-values using false discovery rate (FDR) correction; eQTLs were considered as replicated at Benjamini–Hochberg FDR ≤ 0.05 and with concordant allelic direction with the pQTLs.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Ferkingstad et al. found that 12% of 45,334 lead associations in the GWAS Catalog were with variants in high linkage disequilibrium (LD) with pQTLs.

“Ferkingstad et al.8 found that 12% of 45,334 lead associations in the GWAS Catalog were with variants in high linkage disequilibrium (LD) with pQTLs.”

• cites Ferkingstad et al. [8] Ferkingstad, E. et al. Large-scale integration of the plasma proteome with genetics and disease. Nat. Genet. 53, 1712–1721 (2021). sourceNot checked — no full text of the source

May combine two claims.

As 181 principal components cumulatively explained > 95% of the total variance in the proteomics data, the stricter significance threshold was set to 2.76 × 10-10 (5 × 10-8/181).

“As 181 principal components cumulatively explained > 95% of the total variance in the proteomics data, the stricter significance threshold was set to 2.76 × 10–10 (5 × 10–8/181).”

Probably not a substantial claim: may not pass S3 (load-bearing).

Duplicate associations happening due to data resource overlap were removed.

“Duplicate associations happening due to data resource overlap were removed.”

Probably not a substantial claim: may not pass S3 (load-bearing).

As part of quality control, individual samples that did not pass the Olink internal quality control system were excluded.

“As part of the quality control, we excluded individual samples that did not pass the Olink internal quality control system.”

Classified as probably not a substantial claim.

The rare variant results were corrected for multiple testing based on a Bonferroni-corrected threshold of P < 1.48 × 10-8 [0.05 / (18,717 genes × 181 protein traits)].

“The results were corrected for multiple testing based on Bonferroni-corrected threshold of P < 1.48 × 10–8 [0.05 / (18,717 genes × 181 protein traits)].”

May combine two claims.

Those results suggest that CNV works through gene expression, although it remains unclear whether the effect on the OSCAR level is through OSCAR or VSTM1 gene expression.

“Those results suggest that CNV works through gene expression, although it remains unclear whether the effect on the OSCAR level is through OSCAR or VSTM1 gene expression.”

May combine two claims.

The TNFRSF6B level was causally linked to a reduced risk of inflammatory bowel disease and its subtypes (inflammatory bowel disease (A294), P = 4.00 × 10-20, Benjamini-Hochberg FDR = 9.19 × 10-19; Crohn's disease (A12), P = 1.18 × 10-16, Benjamini-Hochberg FDR = 1.82 × 10-15; ulcerative colitis (A970), P = 2.14 × 10-8, Benjamini-Hochberg FDR = 7.56 × 10-8).

“The TNFRSF6B level was causally linked to a reduced risk of inflammatory bowel disease and its subtypes (inflammatory bowel disease (A294), P = 4.00 × 10–20, Benjamini–Hochberg FDR = 9.19 × 10–19; Crohn’s disease (A12), P = 1.18 × 10–16, Benjamini–Hochberg FDR = 1.82 × 10–15; ulcerative colitis (A970), P = 2.14 × 10–8, Benjamini–Hochberg FDR = 7.56 × 10–8).”

• AssumptionRests on an assumption: The genetic variants used as instruments in Mendelian randomisation analyses satisfy the instrumental variable assumptions (relevance, independence, exclusion restriction).

May combine two claims.