CoChat

A spectrum of nonsense-mediated mRNA decay efficiency along the degree of mutational constraint

Open access

The DOI matches its Crossref record; 45 of the 45 references it cites were found in a registry or an index; none of the 16 sources its claims cite is checked for support yet.

Abstract

Despite its importance for regulating gene expression, nonsense-mediated mRNA decay (NMD) remains poorly understood. Here, we extend the findings of a previous landmark study that proposed several factors associated with NMD efficiency using matched genome and transcriptome data from The Cancer Genome Atlas Program (TCGA) by incorporating additional data including Genotype-Tissue Expression (GTEx), gnomAD, and metrics for mutational constraints. Factors affecting NMD efficiency are analyzed using an allele-specific expression (ASE)-based measure to reduce noise caused by random variations. Additionally, by combining our data with the updated allele frequency database of gnomAD, we demonstrate the spectrum of NMD efficiency according to the degree of gene-level mutational constraints (degree of a gene-tolerating loss-of-function variants). The NMD prediction model, trained on TCGA data, shows that gene-level mutational constraint is an important predictor of NMD efficiency. Findings of this study suggest the potential role of NMD on shaping the landscape of mutational constraints.

Overview

CoChat Summary

TL;DR

This study uses matched genome/transcriptome data from TCGA and GTEx plus gnomAD mutational-constraint metrics to model nonsense-mediated mRNA decay (NMD) efficiency with an allele-specific expression (ASE)-based measure. It reports a spectrum of NMD efficiency along gene-level mutational constraint (LOEUF), with gene-level constraint emerging as a predictor in a TCGA-trained random forest model.

Key findings

  • In TCGA, 4,257 nonsense variants from 9,235 samples across 32 tumor types showed wide NMD-efficiency variability among predicted NMD-sensitive variants and concentration near zero among predicted NMD-escaping variants; adding start-proximal and long-exon rules gave only marginal discriminative benefit over the two canonical rules.
  • NMD efficiency increased with greater mutational constraint (lower LOEUF) in TCGA variants, most evident across LOEUF deciles; this trend was not evident in GTEx, which the authors attribute to depletion of high-NMD-efficiency, low-LOEUF germline variants.
  • The final random forest model used 14 features and explained 43.7% of total variance; against a non-random variance of 46%, this corresponded to 95% (43.7/46.0) of explainable variance. LOEUF ranked eighth in feature importance, and adding cancer type did not improve performance (R² 43.7% to 43.1%).
  • ASE-based models had significantly higher explainable variance than TPM-based models across all parameter combinations; predicted NMD efficiency was higher than observed in all AF subsets of GTEx, suggesting depletion of high-NMD-efficiency variants in the general population.

Methods

  • TCGA: downloaded WES, copy-number, and expression data from GDC for 9,235 tumor samples/32 tumor types; extracted nonsense variants in copy-neutral genes (CN=2), excluded variants with <5 reads and matched-normal VAF >0 and <0.3, and excluded genes with >2 NMD-triggering variants; initial 50,591 nonsense variants; final 4,257 after tumor VAF ≥0.3, tumor DOC ≥30, RNA DOC ≥30, and collapsing. RNA-seq variant calling used Mutect2/hg38; NMD efficiency = −log2(VAFRNA/VAFDNA).
  • GTEx: from ASE data of 844 individuals (V8), extracted 131,584 nonsense variants, used WASP-filtered ASE tables, excluded RNA DOC <30, collapsed repeated tissue samples, and calculated NMD efficiency as −log2(VAFRNA/0.5); final dataset 18,055 variants from 54 tissue types. De novo DB: 415,516 variants, 1,817 nonsense, final 864. MMRF/TARGET validation followed the same procedures but lacked copy-number information.
  • Modeling/statistics: random forest regression with 10,000 trees, max three features, 5-fold cross-validation, 14 features, target NMD efficiency; TCGA split into training n=3,301 and test n=826. Wilcoxon signed-rank, Pearson/Spearman, and Shapiro-Wilk tests were performed in R 4.2.2.

Limitations

  • The paper states: the ASE-based VAF measure is an indirect estimate still subject to random variance; future work should use higher-DOC data or direct measures such as single-molecule imaging. It also states frameshift variants were excluded due to mapping bias; WASP filtering was used for GTEx but not TCGA/MMRF-TARGET, though the authors assume mapping-bias inaccuracy was minor.
  • The paper states: in MMRF-TARGET validation, predicted NMD efficiencies were considerably higher than observed despite correlation; the authors assume this was due to high expression of truncated transcripts in amplified genetic regions because copy-number information was unavailable. It also states somatic variants may not be free from selective pressure because driver mutations can be under selection, though they assume the effect was minor.
  • Evident from the text: GTEx analyses are limited by depletion of low-LOEUF variants, and the AF[0] GTEx subset is younger with lesser negative-selection effect; de novo variants lacked RNA-seq, so observed NMD efficiency could not be derived. Some start-proximal trend analyses used limited samples (n=419 and 517 for Supplementary Fig. 9B/C).

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

Integrity signals

Registry

Crossref · 10.1038/s42003-024-07136-y

Cites

45 of 45 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1 Distribution plots of NMD efficiency observed from TCGA variants. Mean NMD efficiency and number of variants in each distribution are noted on the plot. A Last-exon nonsense variants versus other nonsense variants. B Nonsense variants in the last 50 nt of the penultimate exon versus other nonsense variants. C Nonsense variants located in the last exon or last 50nt of the penultimate exon versus other nonsense variants. D Nonsense variants closer than 150nt from the start codon versus other nonsense variants. E Nonsense variants located in the exon longer than 407 nt versus other nonsense variants. F Nonsense variants located in the last exon, the last 50 nt of the penultimate exon, closer than 150 nt from the start codon, or the exon longer than 407 nt versus other nonsense variants.
Fig. 1 Distribution plots of NMD efficiency observed from TCGA variants. Mean NMD efficiency and number of variants in each distribution are noted on the plot. A Last-exon nonsense variants versus other nonsense variants. B Nonsense variants in the last 50 nt of the penultimate exon versus other nonsense variants. C Nonsense variants located in the last exon or last 50nt of the penultimate exon versus other nonsense variants. D Nonsense variants closer than 150nt from the start codon versus other nonsense variants. E Nonsense variants located in the exon longer than 407 nt versus other nonsense variants. F Nonsense variants located in the last exon, the last 50 nt of the penultimate exon, closer than 150 nt from the start codon, or the exon longer than 407 nt versus other nonsense variants.CC BY-NC-ND · PubMed Central
Fig. 2 Distribution plots of NMD efficiency observed from GTEx variants. Mean NMD efficiency and number of variants in each distribution are noted on the plot. A Last-exon nonsense variants versus other nonsense variants. B Nonsense variants in the last 50nt of the penultimate exon versus other nonsense variants. C Nonsense variants located in the last exon or last 50nt of the penultimate exon versus other nonsense variants. D Nonsense variants closer than 150nt from the start codon versus other nonsense variants. E Nonsense variants located in the exon longer than 407nt versus other nonsense variants. F Nonsense variants located in the last exon, the last 50nt of the penultimate exon, closer than 150nt from the start codon, or the exon longer than 407nt versus other nonsense variants.
Fig. 2 Distribution plots of NMD efficiency observed from GTEx variants. Mean NMD efficiency and number of variants in each distribution are noted on the plot. A Last-exon nonsense variants versus other nonsense variants. B Nonsense variants in the last 50nt of the penultimate exon versus other nonsense variants. C Nonsense variants located in the last exon or last 50nt of the penultimate exon versus other nonsense variants. D Nonsense variants closer than 150nt from the start codon versus other nonsense variants. E Nonsense variants located in the exon longer than 407nt versus other nonsense variants. F Nonsense variants located in the last exon, the last 50nt of the penultimate exon, closer than 150nt from the start codon, or the exon longer than 407nt versus other nonsense variants.CC BY-NC-ND · PubMed Central
Fig. 3 The effect of AF and gene-level mutational constraint (LOEUF) on NMD efficiency. A NMD efficiency of AF-stratified subsets of TCGA variants and B GTEx variants. C NMD efficiency of TCGA variants binned by LOEUF decile. D NMD efficiency of GTEx variants binned by LOEUF decile. E NMD efficiency of TCGA variants (AF = 0) binned by LOEUF decile. F NMD efficiency of GTEx variants (AF = 0) binned by LOEUF decile. AF, allele frequency; LOEUF, loss-of-function observed/expected upper-bound fraction.
Fig. 3 The effect of AF and gene-level mutational constraint (LOEUF) on NMD efficiency. A NMD efficiency of AF-stratified subsets of TCGA variants and B GTEx variants. C NMD efficiency of TCGA variants binned by LOEUF decile. D NMD efficiency of GTEx variants binned by LOEUF decile. E NMD efficiency of TCGA variants (AF = 0) binned by LOEUF decile. F NMD efficiency of GTEx variants (AF = 0) binned by LOEUF decile. AF, allele frequency; LOEUF, loss-of-function observed/expected upper-bound fraction.CC BY-NC-ND · PubMed Central
Fig. 4 Observed (from RNAseq) and predicted (by the RFR model trained in this study) NMD efficiency. A Observed and predicted NMD efficiencies of AF-stratified subsets of GTEx variants, TCGA variants, MMRF-TARGET variants, and de novo variants (observed NMD efficiency from de novo DB was unavailable due to a lack of RNAseq data) are presented. For this experiment, TCGA data were randomly separated into a training set (n = 3301) and a test set (n = 826). B Correlation between observed and predicted NMD efficiency in the AF [0.01, ∞) subset of GTEx. C Correlation between observed and predicted NMD efficiency in the AF [0] subset of GTEx. D Correlation between observed and predicted NMD efficiency in MMRF-TARGET data. Pearson’s test was performed.
Fig. 4 Observed (from RNAseq) and predicted (by the RFR model trained in this study) NMD efficiency. A Observed and predicted NMD efficiencies of AF-stratified subsets of GTEx variants, TCGA variants, MMRF-TARGET variants, and de novo variants (observed NMD efficiency from de novo DB was unavailable due to a lack of RNAseq data) are presented. For this experiment, TCGA data were randomly separated into a training set (n = 3301) and a test set (n = 826). B Correlation between observed and predicted NMD efficiency in the AF [0.01, ∞) subset of GTEx. C Correlation between observed and predicted NMD efficiency in the AF [0] subset of GTEx. D Correlation between observed and predicted NMD efficiency in MMRF-TARGET data. Pearson’s test was performed.CC BY-NC-ND · PubMed Central
Fig. 5 NMD efficiencies of each cancer type in TCGA and tissue type in GTEx that consist of more than 100 variants. The NMD efficiency did not exhibit conspicuous differences across cancer or tissue types A Cancer types in TCGA data. BLCA bladder urothelial carcinoma. BLCA bladder urothelial carcinoma, BRCA breast invasive carcinoma, CESC cervical squamous cell carcinoma and endocervical adenocarcinoma, COAD colon adenocarcinoma, LGG brain lower grade glioma, READ rectum adenocarcinoma, SKCM skin cutaneous melanoma, STAD stomach adenocarcinoma, UCEC uterine corpus endometrial carcinoma). B Tissue types in GTEx data. ADPSBQ adipose-subcutaneous, aDPVSC adipose-visceral (Omentum), ADRNLG adrenal gland, ARTAORT artery-aorta, ARTCRN artery-coronary, ARTTBL artery-tibial, BREAST breast, BRNACC brain-anterior cingulate cortex (BA24), BRNAMY brain-amygdala, BRNCDT brain-caudate (basal ganglia), BRNCHA brain-cerebellum, BRNCHB brain-cerebellar Hemisphere, BRNCTXA brain-Cortex, BRNCTXB brain-frontal cortex (BA9), BRNHPP brain-hippocampus, BRNHPT brain-hypothalamus, BRNNCC brain-nucleus accumbens (basal ganglia), BRNPTM brain-putamen (basal ganglia), BRNSNG brain-substantia nigra, BRNSPC brain-spinal cord (cervical c-1), CLNSGM colon-sigmoid, CLNTRN colon-transverse, ESPGEJ esophagus-gastroesophageal junction, ESPMCS esophagus-mucosa, ESPMSL esophagus-muscularis, FIBRBLS cell-cultured fibroblasts, HRTAA heart-atrial appendage, HRTLV heart-left ventricle, KDNCTX kidney-cortex, LCL cell-EBV-transformed lymphocytes, LIVER liver, LUNG lung, MSCLSK muscle-skeletal, NERVET nerve-tibial, OVARY ovary, PNCREAS pancreas, PRSTTE prostate, PTTARY pituitary, SKINNS skin-not Sun Exposed (suprapubic), SKINS skin-sun exposed (lower leg), SLVRYG minor salivary gland, SNTTRM small intestine-terminal ileum, SPLEEN spleen, STMACH stomach, TESTIS testis, THYROID thyroid, UTERUS uterus, VAGINA vagina, WHLBLD whole blood.
Fig. 5 NMD efficiencies of each cancer type in TCGA and tissue type in GTEx that consist of more than 100 variants. The NMD efficiency did not exhibit conspicuous differences across cancer or tissue types A Cancer types in TCGA data. BLCA bladder urothelial carcinoma. BLCA bladder urothelial carcinoma, BRCA breast invasive carcinoma, CESC cervical squamous cell carcinoma and endocervical adenocarcinoma, COAD colon adenocarcinoma, LGG brain lower grade glioma, READ rectum adenocarcinoma, SKCM skin cutaneous melanoma, STAD stomach adenocarcinoma, UCEC uterine corpus endometrial carcinoma). B Tissue types in GTEx data. ADPSBQ adipose-subcutaneous, aDPVSC adipose-visceral (Omentum), ADRNLG adrenal gland, ARTAORT artery-aorta, ARTCRN artery-coronary, ARTTBL artery-tibial, BREAST breast, BRNACC brain-anterior cingulate cortex (BA24), BRNAMY brain-amygdala, BRNCDT brain-caudate (basal ganglia), BRNCHA brain-cerebellum, BRNCHB brain-cerebellar Hemisphere, BRNCTXA brain-Cortex, BRNCTXB brain-frontal cortex (BA9), BRNHPP brain-hippocampus, BRNHPT brain-hypothalamus, BRNNCC brain-nucleus accumbens (basal ganglia), BRNPTM brain-putamen (basal ganglia), BRNSNG brain-substantia nigra, BRNSPC brain-spinal cord (cervical c-1), CLNSGM colon-sigmoid, CLNTRN colon-transverse, ESPGEJ esophagus-gastroesophageal junction, ESPMCS esophagus-mucosa, ESPMSL esophagus-muscularis, FIBRBLS cell-cultured fibroblasts, HRTAA heart-atrial appendage, HRTLV heart-left ventricle, KDNCTX kidney-cortex, LCL cell-EBV-transformed lymphocytes, LIVER liver, LUNG lung, MSCLSK muscle-skeletal, NERVET nerve-tibial, OVARY ovary, PNCREAS pancreas, PRSTTE prostate, PTTARY pituitary, SKINNS skin-not Sun Exposed (suprapubic), SKINS skin-sun exposed (lower leg), SLVRYG minor salivary gland, SNTTRM small intestine-terminal ileum, SPLEEN spleen, STMACH stomach, TESTIS testis, THYROID thyroid, UTERUS uterus, VAGINA vagina, WHLBLD whole blood.CC BY-NC-ND · PubMed Central

Verification

Claims the paper makes

0 Valid · 2 Not valid · 10 Not checked · 76 Own results and interpretations

48 more claims

Create a free account to see all 88 claims and to verify the 16 references still open.

See all claims — free account
  • It may state two separate claims as one. 14
  • The quote only partly states this claim. 1
  • Probably not a substantial claim. 24

Create a free account to see these claims.

Self-check

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

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Show all 45 references

Research works printed but not linked: 1

  1. Kang, H. & Kim, Y.-G. NMDeff v1.0 10.1016/j.tig.2020.11.002 (2024). · DOI · Paper page

Cited by

See all 24

Metadata

Authors
Kim, Y.-G., Kang, H., Lee, B., Jang, H.-J., Park, J.-H., Ha, C., Park, H., & Kim, J.-W.
Affiliations
  • Samsung Medical Center (KR)
  • Sungkyunkwan University (KR)
  • Korea Advanced Institute of Science and Technology (KR)
  • Samsung (South Korea) (KR)
Corresponding author
  • Park, Jong-ho
  • Park, Hogun
  • Kim, Jong-Won
Venue
Communications Biology
Published
2024-11-07
Volume
7
Issue
1
Pages
1461-1461
ISSN
2399-3642 (ISSN-L)
OpenAlex
W4404130307
Cited by
24
References
45
Citations by year
2026: 17 · 2025: 7
FWCI
2.28
Type
journal article
Language
en
Open access
Licence
Locations
  • Communications Biology (publishedVersion, cc-by-nc-nd)
  • Research Square (submittedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • National Research Foundation
  • Sungkyunkwan University (2019-0-00421)
  • National Research Foundation of Korea (No.1711180477, 2022R1A2C2091571)
  • Ministry of Science and ICT, South Korea (2019-0-00421)
  • Institute for Information and Communications Technology Promotion (2019‐0‐00421)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › RNA Research and Splicing
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genomics and Rare Diseases
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › RNA and protein synthesis mechanisms
Keywords
nonsense-mediated mRNA decay, mutational constraint, loss-of-function mutations, allele-specific expression, gene expression regulation, TCGA, GTEx
MeSH terms
Alleles, Gene Frequency, Humans, Mutation, Neoplasms, RNA, Messenger, Codon, Nonsense, Nonsense Mediated mRNA Decay, Transcriptome

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

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

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

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: A spectrum of nonsense-mediated mRNA decay efficiency along the degree of mutational constraint. https://app.cochat.ai/papers/10.1038/s42003-024-07136-y
The paper
Kim, Y.-G., Kang, H., Lee, B., Jang, H.-J., Park, J.-H., Ha, C., Park, H., & Kim, J.-W. (2024). A spectrum of nonsense-mediated mRNA decay efficiency along the degree of mutational constraint. Communications Biology. https://doi.org/10.1038/s42003-024-07136-y

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.