The DOI matches its Crossref record; 55 of the 55 references it cites were found in a registry or an index; 39 of the 51 sources its claims cite were checked for support.
Human 5′ UTR design and variant effect prediction from a massively parallel translation assay
Open accessPublished version of a preprint
Abstract
The ability to predict the impact of cis-regulatory sequences on gene expression would facilitate discovery in fundamental and applied biology. Here we combine polysome profiling of a library of 280,000 randomized 5' untranslated regions (UTRs) with deep learning to build a predictive model that relates human 5' UTR sequence to translation. Together with a genetic algorithm, we use the model to engineer new 5' UTRs that accurately direct specified levels of ribosome loading, providing the ability to tune sequences for optimal protein expression. We show that the same approach can be extended to chemically modified RNA, an important feature for applications in mRNA therapeutics and synthetic biology. We test 35,212 truncated human 5' UTRs and 3,577 naturally occurring variants and show that the model predicts ribosome loading of these sequences. Finally, we provide evidence of 45 single-nucleotide variants (SNVs) associated with human diseases that substantially change ribosome loading and thus may represent a molecular basis for disease.
Overview
AI snapshot
Written from the paper’s full text.
TL;DR
The paper combines polysome profiling of a 280,000-member randomized human 5′ UTR library with a convolutional neural network, Optimus 5-Prime, to predict ribosome loading from 5′ UTR sequence. The model explains up to 93% of mean ribosome load variation, predicts effects of human variants and modified RNA, and enables genetic-algorithm design of 5′ UTRs for targeted translation.
Key findings
- Optimus 5-Prime explained 93% of MRL variation in a held-out test set of 20,000 sequences, versus at best 66% for position-specific k-mer linear models; it also predicted full polysome distributions with an average 83% of variation across fractions.
- The model generalized to an mCherry reporter library (77% and 78% of MRL variation in two experiments), to 35,212 common human 5′ UTRs plus 3,577 ClinVar variants (81% of MRL variation), and to Ψ- and m1Ψ-modified RNA (68% and 76% of variability, respectively).
- For human variant effects, the model explained 55% of observed MRL change and predicted direction for 64% of variants; 45 SNVs had log2 differences >0.5 or <−0.5, including CPOX rs867711777 (−0.89), RPL5 rs376208311 (−0.87), and TMEM127 rs121908813 (−1.5).
- A genetic algorithm designed 12,000 UTRs for target MRLs; after retraining on 6,082 target-MRL UTRs and 2,695 unseen random UTRs, held-out target-MRL performance improved from r² = 0.386 to 0.772.
Methods
- A 280,000-member eGFP library with 50-nt random 5′ UTRs was in vitro transcribed, transfected into HEK293T cells, polysome-profiled, sequenced, and converted to mean ribosome load (MRL); validation included mCherry, modified UTP, human/ClinVar, and 25–100 nt length libraries.
- A CNN was trained on 260,000 sequences and tested on 20,000, using Keras/TensorFlow, grid-search hyperparameters, three convolution layers, and a dense layer; separate models predicted polysome fractions and were used with a genetic algorithm for UTR design.
- Human analyses used the first 50 nt upstream of annotated translation initiation sites for 35,212 Ensembl transcripts and 3,577 ClinVar variants; a generalized model used 76,319 random sequences of 25–100 nt and test sets of 7,600 random and 7,600 human UTRs.
Limitations
- The paper states that the model was trained on synthetic random sequences, that the 50-nt model covers only 13% of human 5′ UTRs, that the 25–100 nt extension covers 29%, and that longer UTRs had lower read coverage and lower test-set quality.
- The paper notes lower accuracy for variant effect prediction because it compares two noisy measurements and most variants have little effect; it also reports divergence at very high MRLs due to unusual poly-U composition, corrected by retraining, and lower mCherry accuracy partly due to protocol differences.
- Evident from the text: experiments use HEK293T/HEK293 cells, transfected IVT mRNA, and eGFP/mCherry reporters rather than endogenous chromatin/transcription contexts; polysome profiling measures ribosome loading, with protein output validated only for selected UTRs, and the model does not include 5′ cap structure or 3′ UTR sequence.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
41 Own result · 18 Own interpretation · 11 Partly supported · 9 Own method · 6 No basis given · 5 Supported · 3 Sources not found · 2 Supported so far · 2 No full text of the sources · 1 Mixed support · accuracy not yet measured
Some cited sources support this claim and some don’t — read each source’s verdict below.
In the paper
“While there has been some success in this effort in prokaryotes, yeast and even mammalian cells27,34–36, a fully rational approach to designing functional 5′ UTRs has not yet been implemented.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > 5′ UTR design for targeted ribosome loading
What the sources say
- Partly supportedcites [27] Ferreira, J. P., Overton, K. W. & Wang, C. L. Tuning gene expression with synthetic upstream open reading frames. Proc. Natl Acad. Sci. USA 110, 11284–11289 (2013). sourceExperimental — accuracy not yet measured
The opening discussion says there were few systematic approaches for precisely controlling recombinant-protein translation in mammalian cells, which supports the claim’s general concern about design methods. It does not say that a fully rational approach to designing functional 5′ UTRs had never been implemented; the Discussion instead describes the authors’ own systematic approach using engineered RNA leaders. The claim is broader than the source supports. (LLM, conf=0.80)
- Supportedcites [34] Seo, S. W. et al. Predictive design of mRNA translation initiation region to control prokaryotic translation efficiency. Metab. Eng. 15, 67–74 (2013). source
The citation is attached to the clause about success in prokaryotes, not to the subsequent assertion that a fully rational approach has not been implemented. The Software implementation section describes UTR Designer generating 5′-UTR sequences to meet a target expression level, supporting the prokaryotic-success clause. This verdict does not validate the uncited, broader assertion. (LLM, conf=0.90)
- Could not verifycites [35] Jensen, M. K. & Keasling, J. D. Recent applications of synthetic biology tools for yeast metabolic engineering. FEMS Yeast Res. 15, 1–10 (2015). sourceExperimental — accuracy not yet measured
The editorial’s section “Yeast synthetic biology: new tools to unlock cellular function” discusses advances in synthetic-part design, but does not address 5′ UTRs or whether a fully rational approach to designing them has been implemented. Searches for 5′ UTR design and the claimed limitation found no passage that settles it; the copy has no page numbers. (LLM, conf=0.75)
- Disputedcites [36] Salis, H. M., Mirsky, E. A. & Voigt, C. A. Automated design of synthetic ribosome binding sites to control protein expression. Nat. Biotechnol. 27, 946–950 (2009). sourceExperimental — accuracy not yet measured
The Materials and Methods section (Software Implementation; page unavailable) says the authors implemented an RBS Calculator that optimizes ribosome-binding-site sequences to rationally control protein expression. This contradicts the unqualified assertion that a rational design approach had not been implemented, although the paper addresses bacterial ribosome-binding sites rather than every feature of a 5′ UTR. (LLM, conf=0.77)
Checked by a reasoning model — may be wrong.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Our data clearly show the expected decrease in ribosome loading for sequences with either out-of-frame upstream start codons (uAUGs) (Fig. 1c) or upstream open reading frames (uORFs) (Supplementary Fig. 2b)18,19.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > MPRA design and validation
What the sources say
- Partly supportedcites [18] Morris, D. R. & Geballe, A. P. Upstream open reading frames as regulators of mRNA translation. Mol. Cell. Biol. 20, 8635–8642 (2000). sourceExperimental — accuracy not yet measured
In “uORF RECOGNITION BY SCANNING RIBOSOMES” (page unavailable), the review reports that a uORF reduces downstream translation by about tenfold. “NASCENT PEPTIDES MEDIATE REGULATION BY SOME uORFs” explains that ribosomes can still load onto the mRNA but be blocked from reaching the downstream cistron. This supports reduced downstream translation, not a general decrease in ribosome loading for sequences with uORFs or out-of-frame uAUGs. (LLM, conf=0.84)
- Supportedcites [19] Johnstone, T. G., Bazzini, A. A. & Giraldez, A. J. Upstream ORFs are prevalent translational repressors in vertebrates. EMBO J. 35, 706–723 (2016). source
The section headed “uORFs are associated with widespread translational repression” (page unavailable) reports that transcripts containing uORFs or overlapping upstream ORFs had significantly lower downstream translation efficiency, measured using ribosome-protected fragments normalized to mRNA levels. The section on uORF sequence features also reports stronger repression for CDS-overlapping upstream ORFs. This supports the cited expectation of reduced ribosome loading, though the paper describes its measure as translation efficiency. (LLM, conf=0.87)
boosteddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Because only 13% of human 5′ UTRs are shorter than 50 nt and thus covered by Optimus 5-Prime, we next asked whether the approach introduced here could be extended to longer 5′ UTRs.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Modeling human 5′ UTRs of varying length
What the sources say
- Partly supportedcites [44] Mignone, F. et al. Untranslated regions of mRNAs. Genome Biol. 3, reviews0004.1 (2002). sourceExperimental — accuracy not yet measured
The Structural features of untranslated regions section (page not available) says 5′ UTRs average 100–200 nucleotides and vary widely in length. It does not give the proportion of human 5′ UTRs shorter than 50 nucleotides or discuss coverage by Optimus 5-Prime. (LLM, conf=0.78)
- Partly supportedcites [45] Leppek, K., Das, R. & Barna, M. Functional 5′ UTR mRNA structures in eukaryotic translation regulation and how to find them. Nat. Rev. Mol. Cell Biol. 19, 158–174 (2018). sourceExperimental — accuracy not yet measured
The section ‘5′ UTRs as platforms for RNA structure’ (page unavailable) reports a median human 5′ UTR length of 218 nucleotides, supporting the general point that many are longer than 50 nt. It does not give the 13% figure or discuss which UTRs Optimus 5-Prime covers. (LLM, conf=0.84)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Our model’s MRL predictions correlated well with the independently reported fluorescence levels (r2: 0.73–0.85, Supplementary Fig. 9).”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Modeling 5′ UTRs and ribosome loading
What the sources say
- Supportedcites [27] Ferreira, J. P., Overton, K. W. & Wang, C. L. Tuning gene expression with synthetic upstream open reading frames. Proc. Natl Acad. Sci. USA 110, 11284–11289 (2013). source
The Results describe Ferreira et al.’s engineered upstream sequences and explain that GFP fluorescence normalized to RFP was used to measure translation. The Fig. 2 and Fig. 3 captions also report fluorescence measurements for those constructs. This supports citing the paper as the source of the independently reported fluorescence levels; the Optimus model correlation is the citing authors’ own result, not a result attributed to Ferreira et al. (LLM, conf=0.88)
boostedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“Earlier MPRAs designed to learn aspects of 5′ UTR cis-regulation relied on FACS10,11 or growth selection12 to stratify libraries by activity.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Abstract
What the sources say
- Supportedcites [10] Noderer, W. L. et al. Quantitative analysis of mammalian translation initiation sites by FACS-seq. Mol. Syst. Biol. 10, 748 (2014). source
The Results section describes a library varying bases around the translation start codon and sorting cells into 20 FACS gates by GFP/RFP activity. This supports the FACS-based stratification attributed to this source; the separate growth-selection citation is not part of this verdict. (LLM, conf=0.93)
- Not checked — no full text of the sourcecites [11] Kosuri, S. et al. Composability of regulatory sequences controlling transcription and translation in Escherichia coli. Proc. Natl Acad. Sci. USA 110, 14024–14029 (2013). source
- Supportedcites [12] Cuperus, J. T. et al. Deep learning of the regulatory grammar of yeast 5′ untranslated regions from 500,000 random sequences. Genome Res. 27, 2015–2024 (2017). source
The Results section (“5′ UTR library and assay”) describes a competitive growth assay for a library of 5′ UTR variants. It measured each variant’s change in abundance after selection as an indicator of activity. This supports the growth-selection part attributed to this source; the FACS citations are separate. (LLM, conf=0.92)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Translation initiation is largely dependent on start codons and their context and position relative to a CDS12,17.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > MPRA design and validation
What the sources say
- Supportedcites [12] Cuperus, J. T. et al. Deep learning of the regulatory grammar of yeast 5′ untranslated regions from 500,000 random sequences. Genome Res. 27, 2015–2024 (2017). source
In Results > Effects of 5′ UTR features (page not provided), the paper reports that nucleotides immediately before the start codon affect protein expression. The same section finds that upstream start codons affect expression differently depending on their position and reading frame relative to the primary ORF, supporting the citing sentence’s broad paraphrase about translation initiation. (LLM, conf=0.84)
- Supportedcites [17] Hinnebusch, A. G., Ivanov, I. P. & Sonenberg, N. Translational control by 5′-untranslated regions of eukaryotic mRNAs. Science 352, 1413–1416 (2016). source
The section “The scanning mechanism of translation initiation” explains that scanning ribosomes select start codons according to codon identity, surrounding sequence context, and order along the mRNA. It also describes how an upstream start codon can affect translation of the downstream main ORF. Page numbers were not provided in the supplied copy. (LLM, conf=0.90)
boosteddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“The sequence of the 5′ untranslated region (5′ UTR) is a primary determinant of translation efficiency1,2.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Abstract
What the sources say
- Supportedcites [1] Araujo, P. R. et al. Before it gets started: regulating translation at the 5′ UTR. Comp. Funct. Genom. 2012, 475731 (2012). source
The paper’s Translation Regulation section identifies 5′ UTR features as main regulatory elements, including secondary structures, protein-binding sites, uAUGs and uORFs. Its 5′ UTR section also states that GC content can affect translation efficiency. Together, these passages support the broad claim that 5′ UTR sequence is a major determinant of translation efficiency; the copy does not quantify its importance relative to all other determinants. (LLM, conf=0.86)
- Supportedcites [2] Jackson, R. J., Hellen, C. U. T. & Pestova, T. V. The mechanism of eukaryotic translation initiation and principles of its regulation. Nat. Rev. Mol. Cell Biol. 11, 113–127 (2010). source
The Ribosome scanning section explains that initiation complexes traverse the 5′ UTR and that its secondary structure affects the factors required for scanning. The Regulation by specific 5′-UTR/protein interactions section calls 5′-UTR sequences crucial for regulating translation of a large group of mRNAs and describes a 5′-UTR motif necessary for that regulation. Together, these passages support the broad claim that 5′-UTR sequence is a major determinant of translation efficiency. (LLM, conf=0.86)
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“This is likely due to the model learning the impact of Ψ and m1Ψ on the formation of secondary structure33.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Evaluation of mRNA containing Ψ and m1Ψ
What the sources say
- Partly supportedcites [33] Kierzek, E. et al. The contribution of pseudouridine to stabilities and structure of RNAs. Nucleic Acids Res. 42, 3492–3501 (2014). sourceExperimental — accuracy not yet measured
The Results section reports that replacing U with Ψ usually increases RNA-duplex stability, supporting the premise that Ψ can affect RNA structure. The paper does not establish an effect of m1Ψ or examine whether a model learned either modification’s effects, so it cannot substantiate the proposed explanation for improved prediction accuracy. (LLM, conf=0.88)
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“These techniques require the expression of a single library variant per cell that must be transcribed within the cell from a DNA template, making it difficult to distinguish between the effects of transcriptional and translational control.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Abstract
What the sources say
- Partly supportedcites [10] Noderer, W. L. et al. Quantitative analysis of mammalian translation initiation sites by FACS-seq. Mol. Syst. Biol. 10, 748 (2014). sourceExperimental — accuracy not yet measured
Split into 2 atoms — 0 confirmed, 2 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.
- Not checked — no full text of the sourcecites [11] Kosuri, S. et al. Composability of regulatory sequences controlling transcription and translation in Escherichia coli. Proc. Natl Acad. Sci. USA 110, 14024–14029 (2013). source
- Partly supportedcites [12] Cuperus, J. T. et al. Deep learning of the regulatory grammar of yeast 5′ untranslated regions from 500,000 random sequences. Genome Res. 27, 2015–2024 (2017). sourceExperimental — accuracy not yet measured
Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“By calculating UTR minimum free energies (MFE)26 and comparing them to UTR MRLs, we captured and quantitated this repressive effect of secondary structure on ribosome load (Supplementary Fig. 2c)17,25.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > MPRA design and validation
What the sources say
- Not checked — no full text of the sourcecites [26] Zadeh, J. N. et al. NUPACK: analysis and design of nucleic acid systems. J. Comput. Chem. 32, 170–173 (2011). source
- Supportedcites [17] Hinnebusch, A. G., Ivanov, I. P. & Sonenberg, N. Translational control by 5′-untranslated regions of eukaryotic mRNAs. Science 352, 1413–1416 (2016). source
The cited paper’s “Translational control by 5′UTR structure” section states that sufficiently stable mRNA structures inhibit downstream scanning-dependent initiation, supporting the repressive effect attributed to it. The citing sentence says “we captured and quantitated” that effect using MFE and MRL; it presents the quantification as the citing authors’ work, not the cited paper’s. (LLM, conf=0.88)
- Partly supportedcites [25] Kozak, M. Influences of mRNA secondary structure on initiation by eukaryotic ribosomes. Proc. Natl Acad. Sci. USA 83, 2850–2854 (1986). sourceExperimental — accuracy not yet measured
The paper reports that highly stable 5′-UTR hairpins inhibit translation by 85–95%, supporting a repressive effect of secondary structure under those conditions. It does not report quantifying ribosome load; the Discussion instead distinguishes inhibition of translation from initial 40S-subunit binding. The citing authors’ calculation of MFE and comparison with ribosome load are their own analysis, not a measurement in this source. (LLM, conf=0.84)
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Probably not a substantial claim. 23
- It may state two separate claims as one. 21
- The quote only partly states this claim. 1
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Genetic variants play a major role in phenotypic differences between individuals38 and how these sequences affect translation is only beginning to be understood39,40.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
What the sources say
- Partly supportedcites [39] Battle, A. et al. Impact of regulatory variation from RNA to protein. Science 347, 664–667 (2015). sourceExperimental — accuracy not yet measured
In the section labeled Abstract (page unavailable), the paper says that how genetic differences affect gene expression through translation has not been fully characterized. That supports the general point that understanding is incomplete, but the cited passage does not specifically address how 5′ UTR sequences affect translation. (LLM, conf=0.83)
- Partly supportedcites [40] Cenik, C. et al. Integrative analysis of RNA, translation, and protein levels reveals distinct regulatory variation across humans. Genome Res. 25, 1610–1621 (2015). sourceExperimental — accuracy not yet measured
The Discussion (page unavailable) says the molecular mechanisms linking sequence to translation remain an interesting question and that less is known about positive regulation by upstream open reading frames. The Results show that 5′ UTR variants affect translation, but the paper does not characterize understanding of all 5′ UTR sequence effects as only beginning. (LLM, conf=0.78)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Another SNV, rs121908813, is implicated in familial pheochromocytoma, a condition characterized by tumors found in the neuroendocrine system that secrete high levels of catecholamines43.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
What the sources say
- Partly supportedcites [43] Qin, Y. et al. Germline mutations in TMEM127 confer susceptibility to pheochromocytoma. Nat. Genet. 42, 229–233 (2010). sourceExperimental — accuracy not yet measured
The Abstract describes pheochromocytomas as catecholamine-secreting tumors of neural crest origin and reports TMEM127 mutations in affected families. Our copy does not identify rs121908813 or establish that it is one of those mutations, so the claim about that specific SNV is not supported. (LLM, conf=0.84)
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
- Partly supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Polysome profiling has been used extensively to measure translation of native RNA isoforms14,15 but isolating the role of 5′ UTR regulation has been difficult due to differences in the size and sequence of the concomitant coding sequences and 3′ UTRs.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Abstract
What the sources say
- Supportedcites [14] Floor, S. N. & Doudna, J. A. Tunable protein synthesis by transcript isoforms in human cells. elife 5, e10921 (2016). source
Split into 2 atoms — 2 confirmed, 0 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.
- Supportedcites [15] Wang, X., Hou, J., Quedenau, C. & Chen, W. Pervasive isoform‐specific translational regulation via alternative transcription start sites in mammals. Mol. Syst. Biol. 12, 875 (2016). source
Split into 2 atoms — 2 confirmed, 0 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“The two uridine analogs pseudouridine (Ψ) or 1-methyl-pseudouridine (m1Ψ) are widely used for mRNA therapeutics because they can increase mRNA stability and help modulate the host immune response31,32.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Evaluation of mRNA containing Ψ and m1Ψ
What the sources say
- Partly supportedcites [31] Karikó, K. et al. Incorporation of pseudouridine into mRNA yields superior nonimmunogenic vector with increased translational capacity and biological stability. Mol. Ther. 16, 1833–1840 (2008). sourceExperimental — accuracy not yet measured
The Introduction reports that pseudouridine (Ψ) suppresses RNA-mediated immune activation, and the Discussion describes increased biological stability of Ψ-modified mRNA as a likely contributor to enhanced translation. The paper does not address 1-methyl-pseudouridine (m1Ψ) or establish that both analogs are widely used in mRNA therapeutics. Page numbers are unavailable in our copy. (LLM, conf=0.88)
- Partly supportedcites [32] Anderson, B. R. et al. Incorporation of pseudouridine into mRNA enhances translation by diminishing PKR activation. Nucleic Acids Res. 38, 5884–5892 (2010). sourceExperimental — accuracy not yet measured
The Introduction and Discussion (page numbers unavailable) support pseudouridine’s role in reducing innate immune activation and its potential for therapeutic mRNA delivery. They do not establish that either analog increases mRNA stability, and the paper does not address 1-methyl-pseudouridine or establish that both analogs are widely used in therapeutics. (LLM, conf=0.88)
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Massively parallel reporter assays (MPRAs) – methods that assess thousands to millions of sequence variants in a single experiment – coupled with machine learning have proven useful in addressing similar voids by producing quantitative biological insight that would be difficult to achieve through traditional approaches3–9.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Abstract
What the sources say
- Partly supportedcites [3] Angermueller, C., Pärnamaa, T., Parts, L. & Stegle, O. Deep learning for computational biology. Mol. Syst. Biol. 12, 878 (2016). sourceExperimental — accuracy not yet measured
The Introduction and ‘Deep learning for regulatory genomics’ describe machine learning as a way to derive biological predictions and report advantages over conventional approaches. But the passages from our copy do not establish the specific claim that coupling massively parallel reporter assays with machine learning has produced those insights; a search for ‘massively parallel reporter assays’ found no matching passage. (LLM, conf=0.82)
- Not checked — no full text of the sourcecites [4] Alipanahi, B., Delong, A., Weirauch, M. T. & Frey, B. J. Predicting the sequence specificities of DNA- and RNA-binding proteins by deep learning. Nat. Biotechnol. 33, 831–838 (2015). source
- Partly supportedcites [5] Zhou, J. & Troyanskaya, O. G. Predicting effects of noncoding variants with deep learning-based sequence model. Nat. Methods 12, 931–934 (2015). sourceExperimental — accuracy not yet measured
The Abstract describes DeepSEA, a deep-learning model that learns from large-scale chromatin-profiling data to predict effects of noncoding variants. This supports the broader point that machine learning can provide quantitative biological insight, but the paper does not describe coupling machine learning with massively parallel reporter assays. A search for “reporter” found no relevant passage. (LLM, conf=0.86)
- Could not verifycites [6] Kleftogiannis, D., Kalnis, P. & Bajic, V. B. DEEP: a general computational framework for predicting enhancers. Nucleic Acids Res. 43, e6 (2015). sourceExperimental — accuracy not yet measured
The Introduction discusses machine learning applied to histone-mark and other high-throughput data, while the Materials and Methods describe DEEP’s use of ChIP-Seq data or DNA sequences. Searches for massively parallel reporter assays and reporter assays coupled with machine learning found no passage addressing that combination or the claimed quantitative insight. (LLM, conf=0.76)
- Partly supportedcites [7] Liu, F., Li, H., Ren, C., Bo, X. & Shu, W. PEDLA: predicting enhancers with a deep learning-based algorithmic framework. Sci. Rep. 6, 28517 (2016). sourceExperimental — accuracy not yet measured
The Abstract reports that PEDLA uses deep learning and heterogeneous genomic data to improve enhancer prediction, supporting the general point that machine learning can yield useful biological insight. But the paper does not report coupling machine learning with massively parallel reporter assays; its Materials and Methods describe other data types and validation using enhancer markers. The MPRA-specific part of the attributed claim is therefore unsupported. (LLM, conf=0.85)
- Partly supportedcites [8] Kelley, D. R., Snoek, J. & Rinn, J. L. Basset: learning the regulatory code of the accessible genome with deep convolutional neural networks. Genome Res. 26, 990–999 (2016). sourceExperimental — accuracy not yet measured
The Results section (‘In silico saturation mutagenesis’; page unavailable) describes parallel reporter assays as a way to measure the activity of many mutations, and presents Basset’s machine-learning predictions as an in silico approach to the same problem. It does not report coupling an MPRA with machine learning; Basset was trained on DNase-seq accessibility data, so the cited paper supports the broader point about quantitative insight but not the claimed combination. (LLM, conf=0.86)
- Partly supportedcites [9] Zhao, W. et al. Massively parallel functional annotation of 3′ untranslated regions. Nat. Biotechnol. 32, 387–391 (2014). sourceExperimental — accuracy not yet measured
The main text (page unavailable) describes fast-UTR, a massively parallel reporter approach that quantified effects of 3′ UTR sequences and genetic variants. It does not report coupling the assay with machine learning; it says the resulting data could be useful for refining computational prediction methods. Thus the paper supports the assay’s quantitative value, but not the machine-learning part of the claim. (LLM, conf=0.86)
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Human 5′ UTR sequences vary in length from tens to thousands of nucleotides with a median length of 218 nt44,45.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Modeling human 5′ UTRs of varying length
What the sources say
- Partly supportedcites [44] Mignone, F. et al. Untranslated regions of mRNAs. Genome Biol. 3, reviews0004.1 (2002). sourceExperimental — accuracy not yet measured
The Structural features of untranslated regions section (page not available in our copy) says that 5′ UTR lengths vary within a species from about a dozen to a few thousand nucleotides, supporting the claimed range. It does not give a median of 218 nt for human 5′ UTRs; it reports an average of 100–200 nt across diverse taxonomic classes instead. (LLM, conf=0.87)
- Supportedcites [45] Leppek, K., Das, R. & Barna, M. Functional 5′ UTR mRNA structures in eukaryotic translation regulation and how to find them. Nat. Rev. Mol. Cell Biol. 19, 158–174 (2018). source
The ‘5′ UTRs as platforms for RNA structure’ section states that humans have a median 5′ UTR length of 218 nucleotides. It also reports that individual 5′ UTR lengths in higher eukaryotes range from a few to thousands of base pairs, consistent with the claimed broad range. No page number is supplied in our copy. (LLM, conf=0.91)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Partly supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“One cause of the disease is a result of either the disruption or downregulation of RPL542.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
What the sources say
- Supportedcites [42] Boria, I. et al. The ribosomal basis of Diamond–Blackfan anemia: mutation and database update. Hum. Mutat. 31, 1269–1279 (2010). source
The Variants > RPL5 section reports RPL5 mutations in patients with Diamond-Blackfan anemia, including new mutations identified by this paper. The Biological Relevance > RPL5 section describes the pathogenic effects of RPL5 haploinsufficiency and knockdown, supporting disruption or downregulation as a cause. Page numbers were not supplied in our copy. (LLM, conf=0.92)
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Translation initiation and elongation is also affected by RNA secondary structure that forms within 5′ UTRs and coding sequences, with strong structures showing the most negative effect on translation17,25.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > MPRA design and validation
What the sources say
- Partly supportedcites [17] Hinnebusch, A. G., Ivanov, I. P. & Sonenberg, N. Translational control by 5′-untranslated regions of eukaryotic mRNAs. Science 352, 1413–1416 (2016). sourceExperimental — accuracy not yet measured
The section “Translational control by 5′UTR structure” says sufficiently stable structures inhibit scanning-dependent initiation, supporting an inhibitory effect of strong 5′UTR structures. It does not establish an effect on elongation from structures in coding sequences, or that stronger structures always have the most negative effect; it also describes a strong stem-loop that improves start-codon recognition. (LLM, conf=0.88)
- Partly supportedcites [25] Kozak, M. Influences of mRNA secondary structure on initiation by eukaryotic ribosomes. Proc. Natl Acad. Sci. USA 83, 2850–2854 (1986). sourceExperimental — accuracy not yet measured
The Discussion reports that a highly stable hairpin near or upstream of the initiator codon markedly reduced translation, while a moderately stable one did not. The study tested structures in the 5′ noncoding region or involving the initiator codon; it does not establish an effect of coding-sequence structure on elongation. (LLM, conf=0.88)
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Partly supported
None of the sources cited for this claim could be matched to a paper in Crossref, so there is nothing to check it against.
In the paper
“Of note, several filters did not fall into either of these categories and also did not match previously described PWMs for RNA binding proteins (Tomtom29 and the Homo sapiens RBP database30), suggesting the possibility for previously undescribed regulatory interactions.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Modeling 5′ UTRs and ribosome loading
What the sources say
- Not matchedcites Tomtom29
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Not matchedcites Homo sapiens RBP database30
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
- Not matched
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“A model trained on data from another polysome profiling experiment performed similarly (Supplementary Fig. 5a).”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Modeling 5′ UTRs and ribosome loading
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We also identified 2,308 additional SNVs that resulted from oligo synthesis errors, and found that 103 of them showed MRL changes greater than 0.5 or less than −0.5 in log2 scale (Supplementary Fig. 14b).”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The eGFP library contained approximately 750,000 unique sequences and the mCherry library contained approximately 500,000 sequences.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > Sequence Processing.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“As an example, one of the ClinVar variants with sizeable differences in MRL, rs867711777, is found in the 5′ UTR of the CPOX gene and shows a log2 difference of −0.89.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Together with a genetic algorithm, we use the model to engineer new 5′ UTRs that accurately direct specified levels of ribosome loading, providing the ability to tune sequences for optimal protein expression.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Rests on an assumption: The field lacks a means to accurately predict protein expression from 5′ UTR sequence alone.
- Rests on an assumption: The first 50 bases upstream of the CDS contain the regulatory signals that mediate translation initiation beyond ribosomal recruitment to the 5′ cap.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Optimus 5-Prime, the CNN trained on the data has notable performance, explaining up to 93% of mean ribosome load variation in the test set and up to 81% of variation for 38,789 truncated human UTRs.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Discussion
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The eGFP library contained approximately 750,000 unique sequences and the mCherry library contained approximately 500,000 sequences.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > Sequence Processing.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Using this expanded dataset, we retrained the Optimus 5-Prime model in Figure 2, which showed increased accuracy with all sub-libraries and unchanged performance with random library sequences (Supplementary Fig. 13b).”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > 5′ UTR design for targeted ribosome loading
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The variant rs376208311 lies in the 5′ UTR of the ribosomal subunit gene RPL5 and shows a −0.87 log2 difference in MRL.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“sometime after training this first model we determined that adding a third convolution layer and additional filters to each layer showed improved performance.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > Genetic algorithm for designing new 5′ UTR sequences.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, the model performs nearly as well after randomly splitting the training (260,000) and test (20,000) sets (Supplementary Fig. 15d).”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > Convolution Neural Network.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The quote only partly states this claim.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The majority of SNVs had little to no effect, but 45 had log2 differences greater than 0.5 or less than −0.5 (Supplementary Table 4, Clinvar SUB4797518).”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“No sequences between the eGFP and mCherry library matched.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > Sequence Processing.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The generalized Optimus 5-Prime model performed well on both the human (r2: 0.78) and random (r2: 0.84) sequences (Supplementary Fig. 15c,d) and for 5′ UTRs of any length (Fig. 4e, r2: 0.75–0.84).”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Modeling human 5′ UTRs of varying length
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Overall, Optimus 5-Prime could explain 55% of the observed MRL change (Fig. 4b) and accurately predicted the direction of change for 64% of the variants.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
hedgedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Visualization of the filters in the first and second convolution layer revealed recognizable motifs including strong TIS sequences (e.g. ACCAUG), stop codons (TAA, TGA, TAG), uORFs, non-canonical start codons (CUG, GUG), and sequences composed of multiples of CG or AU likely involved in secondary structure formation (Fig. 2f and Supplementary Fig. 10).”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Modeling 5′ UTRs and ribosome loading
hedgedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“We performed 10-fold cross-validation grid search to exhaustively test hyperparameter combinations of convolution layers (2, 3), convolution filter lengths (8, 10, 12), number of convolution filters (40, 80, 120), number of nodes in the dense layer (40, 80, 120), dropout probability between all layers (0, 0.2, 0.4).”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > Convolution Neural Network.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“Genetic variants play a major role in phenotypic differences between individuals38 and how these sequences affect translation is only beginning to be understood39,40.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
What the sources say
- Not checked — no full text of the sourcecites [38] Hernandez, R. D. et al. Singleton variants dominate the genetic architecture of human gene expression. Preprint https://doi.org/10.2139/ssrn.3151998 (2018). source
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“Both RT primers have unique molecular indices (UMIs).”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > Polysome Fraction Processing and Next-Generation Sequencing.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Of note, predictions are not limited to common variants or even those that have been previously described; instead the model can be used to screen every possible SNV, insertion or deletion in the 100 bases upstream of the start codon – there are millions in the human genome - and select those for further study that have the strongest impact on ribosome loading and thus the highest likelihood of be”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Discussion
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Rests on an assumption: Changes in mean ribosome load (MRL) measured in the assay reflect changes in protein expression that can affect disease pathogenicity.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“But existing approaches to this problem, such as quantitative trait locus (QTL) analysis and genome wide association studies (GWAS) are limited to common variants and cannot scale to the enormous number of rare 5′ UTR variants occurring in the human population.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds. It may state two separate claims as one.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“Training involved regularization to limit overfitting and 5-fold cross validation.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > k-mer Linear Model.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our model also proved capable of predicting the effect of disease-relevant 5′ UTR variants on translation, even suggesting mechanisms of action.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Discussion
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“UTRs were removed if the CDS sequence did not match the intended sequence.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > Sequence Processing.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“To normalize differences in total read counts between fractions, relative reads were calculated within each fraction.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > Sequence Processing.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
In the paper
“The depletion of CPOX reduces heme biosynthesis and is the cause of hereditary coproporphyria (HCP)41.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
What the sources say
- Not matchedcites [41] Wang, B. & Bissell, D. M. Hereditary Coproporphyria (University of Washington, 2012). .
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
- Not matched
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“The model was trained over three epochs before overfitting occurred.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > Convolution Neural Network.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“A small portion of the electroporation was plated and resulted in ~750,000 cfu and the rest was grown in liquid culture overnight (both under kanamycin selection).”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Online Methods > Randomized 50-mer 5′ UTR Library.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“We also created a second test set consisting of 7,600 human 5′ UTRs – 100 UTRs for every length from 25 to 100 nt.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Modeling human 5′ UTRs of varying length
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Moreover, a majority of variants do not impact translation, resulting in a large cluster of variants for which the MRL change is close to zero where measurements are dominated by noise.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Predicting the effect of human 5′ UTR variants on ribosome loading
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“To ensure that 5′ UTRs of every length would be represented equally, we took the 100 5′ UTRs with the deepest read coverage at every length (~10% of the library) as the test set, rather than using the top 10% of the entire population.”
§ Human 5′ UTR design and variant effect prediction from a massively parallel translation assay > Results > Modeling human 5′ UTRs of varying length
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
The paper relies on this without support.
The field lacks a means to accurately predict protein expression from 5′ UTR sequence alone.
“While many cis-regulatory elements within human 5′ UTRs have been characterized individually, the field still lacks a means to accurately predict protein expression from 5′ UTR sequence alone, limiting the ability to estimate the effects of genome-encoded variants and the ability to engineer 5′ UTRs for precise translation control.”
The paper relies on this without support.
The region surrounding the start codon (the translation initiation site or Kozak sequence) is a primary determinant of whether a ribosome will begin translation.
“The region surrounding the start codon, known as the translation initiation site (TIS) or the Kozak sequence, is a primary determinant of whether a ribosome will begin translation.”
The paper relies on this without support.
The first 50 bases upstream of the CDS contain the regulatory signals that mediate translation initiation beyond ribosomal recruitment to the 5′ cap.
“We initially focused on the first 50 bases upstream of the CDS to specifically investigate the regulatory signals that mediate the initiation of translation beyond ribosomal recruitment to the 5′ cap.”
The paper relies on this without support.
TMEM127 acts as a tumor suppressor.
“TMEM127 acts as a tumor suppressor and decreased expression of it could explain the observed pathogenicity of this variant.”
The paper relies on this without support.
The variant rs376208311 is associated with Diamond-Blackfan anemia (BDA).
“This variant is associated with Diamond-Blackfan anemia (BDA).”
The paper relies on this without support.
Changes in mean ribosome load (MRL) measured in the assay reflect changes in protein expression that can affect disease pathogenicity.
“select those for further study that have the strongest impact on ribosome loading and thus the highest likelihood of being pathogenic.”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source doesn’t support the claim as written.
In the citing paper
“It should be pointed out that the authors of ref. [46] also demonstrated that MRL inversely correlates with minimal free energy values.”
From Effects of Combinations of Untranslated-Region Sequences on Translation of mRNA (2023) · cites this paper as [46]
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“On the basis of 280,000 different 5′UTR sequences, the Sample et al. [37] study demonstrated that MRL values range from two to 11; however, among the highest MRLs, only in the range six–eight was there a high correlation (r2 = 0.77) between the predicted and observed values.”
From Untranslated Region Sequences and the Efficacy of mRNA Vaccines against Tuberculosis (2024) · cites this paper as [37]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“In addition to greater length, the TPL sequence has a higher GC content and highly negative folding minimal free energy (ΔG), which correlates with inhibition of translation [4,7,45,46] (Table 1).”
From Effects of Combinations of Untranslated-Region Sequences on Translation of mRNA (2023) · cites this paper as [46]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“In TPL, the presence of unique cis- and/or trans-regulatory elements that ensure efficient translation is indirectly confirmed by evidence obtained via assessment of predicted mean ribosomal loading (MRL) with the Optimus 5-Prime algorithm [46].”
From Effects of Combinations of Untranslated-Region Sequences on Translation of mRNA (2023) · cites this paper as [46]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“On the basis of 280,000 different 5′UTR sequences, the authors of that paper demonstrated that MRL values range from 2 to 11.”
From Effects of Combinations of Untranslated-Region Sequences on Translation of mRNA (2023) · cites this paper as [46]
Experimental — accuracy not yet measured
The cited source says what the claim attributes to it.
In the citing paper
“Here, we improved the mRNA translation by using a new set of 5′ UTR synthetic sequences based on the selection of synthetic 5′ UTRs with high ribosome loading.22”
From Combining an optimized mRNA template with a double purification process allows strong expression of in vitro transcribed mRNA (2021) · cites this paper as [22]
The cited source says what the claim attributes to it.
In the citing paper
“It is worth noting that the predictive model in question was tested only for 5′UTR sequences up to 100 bp in length.”
From Untranslated Region Sequences and the Efficacy of mRNA Vaccines against Tuberculosis (2024) · cites this paper as [37]
The cited source says what the claim attributes to it.
In the citing paper
“The latter is a method for predicting MRL of a transcript through the use of convolutional neural networks.”
From Effects of Combinations of Untranslated-Region Sequences on Translation of mRNA (2023) · cites this paper as [46]
The check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“the Pfizer-BioNTech mRNA vaccine contains the 5′ UTR of human α-globin, which was identified as one of the top 20% 5′ UTRs in massive screening studies,10,11,13 and the 3′ UTR AES-mtRNR1 combination of two genes, AES (Amino Enhancer of Split) and mtRNR1 (12S mitochondrial RNA).12”
From Evaluation of synthetic mRNA with selected UTR sequences and alternative poly(A) tail, in vitro and in vivo (2025) · cites this paper as [11]
Experimental — accuracy not yet measuredThe check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“TPL as the 5′UTR had the highest MRL (7.7), whereas MRL of the other tested UTRs was within 5.7–6.7.”
From Effects of Combinations of Untranslated-Region Sequences on Translation of mRNA (2023) · cites this paper as [46]
Experimental — accuracy not yet measuredIn the citing paper
“However, because UTR performance can vary by cell type, alternative UTR sequences may be used that have been optimized for the desired application and intended cell target16–18.”
From mRNA vaccines for infectious diseases: principles, delivery and clinical translation (2021) · cites this paper as [16]
Citations
Cites
Cutadapt removes adapter sequences from high-throughput sequencing reads
Martin, M. (2011) · EMBnet.journal · doi:10.14806/ej.17.1.200 · Cited by 33,889
TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., Corrado, G. S., Davis, A., Dean, J., Devin, M., et al. (2016) · arXiv · doi:10.48550/arxiv.1603.04467 · Cited by 9,408
Shrestha, S., & Offer, S. M. (2016) · Medical Epigenetics · doi:10.1159/000444713 · Cited by 8,951
Kozak, M. (1986) · Cell · doi:10.1016/0092-8674(86)90762-2 · Cited by 4,741
Predicting the sequence specificities of DNA- and RNA-binding proteins by deep learning
Alipanahi, B., Delong, A., Weirauch, M. T., & Frey, B. J. (2015) · Nature Biotechnology · doi:10.1038/nbt.3300 · Cited by 2,873
The mechanism of eukaryotic translation initiation and principles of its regulation
Jackson, R. J., Hellen, C. U. T., & Pestova, T. V. (2010) · Nature Reviews Molecular Cell Biology · doi:10.1038/nrm2838 · Cited by 2,759
ClinVar: public archive of interpretations of clinically relevant variants
Landrum, M. J., Lee, J. M., Benson, M., Brown, G., Chao, C., Chitipiralla, S., Gu, B., Hart, J., Hoffman, D., Hoover, J., et al. (2016) · Nucleic Acids Research · doi:10.1093/nar/gkv1222 · Cited by 2,758
Predicting effects of noncoding variants with deep learning–based sequence model
Zhou, J., & Troyanskaya, O. G. (2015) · Nature Methods · doi:10.1038/nmeth.3547 · Cited by 2,338
Quantifying similarity between motifs
Gupta, S., Stamatoyannopoulos, J. A., Bailey, T. L., & Noble, W. S. (2007) · Genome Biology · doi:10.1186/gb-2007-8-2-r24 · Cited by 2,058
Karikó, K., Muramatsu, H., Welsh, F. A., Ludwig, J., Kato, H., Akira, S., & Weissman, D. (2008) · Molecular Therapy · doi:10.1038/mt.2008.200 · Cited by 1,930
Cited by
Sorted by citations
mRNA vaccines for infectious diseases: principles, delivery and clinical translation
Chaudhary, N., Weissman, D., & Whitehead, K. A. (2021) · Nature Reviews Drug Discovery · doi:10.1038/s41573-021-00283-5 · Cited by 1,555
mRNA-based therapeutics: powerful and versatile tools to combat diseases
Qin, S., Tang, X., Chen, Y., Chen, K., Fan, N., Xiao, W., Zheng, Q., Li, G., Teng, Y., Wu, M., et al. (2022) · Signal Transduction and Targeted Therapy · doi:10.1038/s41392-022-01007-w · Cited by 852
Unlocking the promise of mRNA therapeutics
Rohner, E., Yang, R., Foo, K. S., Goedel, A., & Chien, K. R. (2022) · Nature Biotechnology · doi:10.1038/s41587-022-01491-z · Cited by 612
Kim, J., Eygeris, Y., Gupta, M., & Sahay, G. (2021) · Advanced Drug Delivery Reviews · doi:10.1016/j.addr.2020.12.014 · Cited by 514
Recent advances in mRNA vaccine technology
Pardi, N., Hogan, M. J., & Weissman, D. (2020) · Current Opinion in Immunology · doi:10.1016/j.coi.2020.01.008 · Cited by 512
The landscape of mRNA nanomedicine
Huang, X., Kong, N., Zhang, X., Cao, Y., Langer, R., & Tao, W. (2022) · Nature Medicine · doi:10.1038/s41591-022-02061-1 · Cited by 511
Modifications in an Emergency: The Role of N1-Methylpseudouridine in COVID-19 Vaccines
Nance, K. D., & Meier, J. L. (2021) · ACS Central Science · doi:10.1021/acscentsci.1c00197 · Cited by 466
Combinatorial optimization of mRNA structure, stability, and translation for RNA-based therapeutics
Leppek, K., Byeon, G. W., Kladwang, W., Wayment-Steele, H. K., Kerr, C. H., Xu, A. F., Kim, D. S., Topkar, V. V., Choe, C., Rothschild, D., et al. (2022) · Nature Communications · doi:10.1038/s41467-022-28776-w · Cited by 399
Combinatorial optimization of mRNA structure, stability, and translation for RNA-based therapeutics
Leppek, K., Byeon, G. W., Kladwang, W., Wayment-Steele, H. K., Kerr, C. H., Xu, A. F., Kim, D. S., Topkar, V. V., Choe, C., Rothschild, D., et al. (2021) · doi:10.1101/2021.03.29.437587 · Cited by 399
mRNA therapeutics in cancer immunotherapy
Beck, J. D., Reidenbach, D., Salomon, N., Sahin, U., Türeci, Ö., Vormehr, M., & Kranz, L. M. (2021) · Molecular Cancer · doi:10.1186/s12943-021-01348-0 · Cited by 385
Tailoring mRNA Vaccine to Balance Innate/Adaptive Immune Response
Linares-Fernández, S., Lacroix, C., Exposito, J.-Y., & Verrier, B. (2020) · Trends in Molecular Medicine · doi:10.1016/j.molmed.2019.10.002 · Cited by 376
Nucleic acid delivery for therapeutic applications
Gupta, A., Andresen, J. L., Manan, R. S., & Langer, R. (2021) · Advanced Drug Delivery Reviews · doi:10.1016/j.addr.2021.113834 · Cited by 351
mRNA vaccines for COVID-19: what, why and how
Park, J. W., Lagniton, P. N. P., Liu, Y., & Xu, R.-H. (2021) · International Journal of Biological Sciences · doi:10.7150/ijbs.59233 · Cited by 321
Leveraging artificial intelligence in the fight against infectious diseases
Wong, F. J., de la Fuente‐Núñez, C., & Collins, J. J. (2023) · Science · doi:10.1126/science.adh1114 · Cited by 331
Recommendations for clinical interpretation of variants found in non-coding regions of the genome
Ellingford, J. M., Ahn, J. W., Bagnall, R. D., Baralle, D., Barton, S., Campbell, C., Downes, K., Ellard, S., Duff‐Farrier, C. R. A., FitzPatrick, D. R., et al. (2022) · Genome Medicine · doi:10.1186/s13073-022-01073-3 · Cited by 274
Esposito, D., Weile, J., Shendure, J., Starita, L. M., Papenfuss, A. T., Roth, F. P., Fowler, D. M., & Rubin, A. F. (2019) · Genome biology · doi:10.1186/s13059-019-1845-6 · Cited by 260
Decoding mRNA translatability and stability from the 5′ UTR
Jia, L., Mao, Y., Ji, Q., Dersh, D., Yewdell, J. W., & Qian, S.-B. (2020) · Nature Structural & Molecular Biology · doi:10.1038/s41594-020-0465-x · Cited by 248
Chen, J., Hu, Z., Sun, S., Tan, Q., Wang, Y., Yu, Q., Zong, L., Hong, L., Xiao, J., Shen, T., et al. (2022) · doi:10.1101/2022.08.06.503062 · Cited by 247
Melamed, J. R., Yerneni, S. S., Arral, M. L., LoPresti, S. T., Chaudhary, N., Sehrawat, A., Muramatsu, H., Alameh, M., Pardi, N., Weissman, D., et al. (2023) · Science Advances · doi:10.1126/sciadv.ade1444 · Cited by 218
A Deep Neural Network for Predicting and Engineering Alternative Polyadenylation
Bogard, N., Linder, J., Rosenberg, A. B., & Seelig, G. (2019) · Cell · doi:10.1016/j.cell.2019.04.046 · Cited by 215
Self-citationChemically Modified Platforms for Better RNA Therapeutics
Shi, Y., Zhen, X., Zhang, Y., Li, Y., Koo, S., Saiding, Q., Kong, N., Liu, G., Chen, W., & Tao, W. (2024) · Chemical Reviews · doi:10.1021/acs.chemrev.3c00611 · Cited by 189
Characterising the loss-of-function impact of 5’ untranslated region variants in 15,708 individuals
Whiffin, N., Karczewski, K. J., Zhang, X., Chothani, S. P., Smith, M. J., Evans, D. G., Roberts, A. M., Quaife, N. M., Schäfer, S., Rackham, O. J. L., et al. (2020) · Nature Communications · doi:10.1038/s41467-019-10717-9 · Cited by 182
Griesemer, D. S., Xue, J. R., Reilly, S. K., Ulirsch, J. C., Kukreja, K., Davis, J. R., Kanai, M., Yang, D. K., Butts, J. C., Guney, M. H., et al. (2021) · Cell · doi:10.1016/j.cell.2021.08.025 · Cited by 176
Goshisht, M. K. (2024) · ACS Omega · doi:10.1021/acsomega.3c05913 · Cited by 173
High-throughput 5’ UTR engineering for enhanced protein production in non-viral gene therapies
Cao, J., Novoa, E. M., Zhang, Z., Chen, W. C. W., Liu, D., Choi, G. C. G., Wong, A. S. L., Wehrspaun, C., Kellis, M., & Lu, T. K. (2021) · Nature Communications · doi:10.1038/s41467-021-24436-7 · Cited by 159
Human-specific genetics: new tools to explore the molecular and cellular basis of human evolution
Pollen, A. A., Kilik, U., Lowe, C. B., & Camp, J. G. (2023) · Nature Reviews Genetics · doi:10.1038/s41576-022-00568-4 · Cited by 165
The genetic and biochemical determinants of mRNA degradation rates in mammals
Agarwal, V., & Kelley, D. R. (2022) · Genome Biology · doi:10.1186/s13059-022-02811-x · Cited by 155
Tailor made: the art of therapeutic mRNA design
Metkar, M., Pepin, C. S., & Moore, M. J. (2023) · Nature Reviews Drug Discovery · doi:10.1038/s41573-023-00827-x · Cited by 158
Leveraging mRNA Sequences and Nanoparticles to Deliver SARS‐CoV‐2 Antigens In Vivo
Zeng, C., Hou, X., Yan, J., Zhang, C., Li, W., Zhao, W., Du, S., & Dong, Y. (2020) · Advanced Materials · doi:10.1002/adma.202004452 · Cited by 152
RiNALMo: general-purpose RNA language models can generalize well on structure prediction tasks
Penić, R. J., Vlašić, T., Huber, R. G., Wan, Y., & Šikić, M. (2025) · Nature Communications · doi:10.1038/s41467-025-60872-5 · Cited by 169
Model-driven generation of artificial yeast promoters
Kotopka, B. J., & Smolke, C. D. (2020) · Nature Communications · doi:10.1038/s41467-020-15977-4 · Cited by 146
A 5′ UTR language model for decoding untranslated regions of mRNA and function predictions
Chu, Y., Yu, D., Li, Y., Huang, K., Shen, Y., Cong, L., Zhang, J. Y., & Wang, M. (2024) · Nature Machine Intelligence · doi:10.1038/s42256-024-00823-9 · Cited by 148
Machine-guided design of cell-type-targeting cis-regulatory elements
Gosai, S. J., Castro, R. I., Fuentes, N., Butts, J. C., Mouri, K., Alasoadura, M., Kales, S., Nguyen, T. T. L., Noche, R. R., Rao, A. S., et al. (2024) · Nature · doi:10.1038/s41586-024-08070-z · Cited by 142
The Untranslated Regions of mRNAs in Cancer
Schuster, S. L., & Hsieh, A. C. (2019) · Trends in cancer · doi:10.1016/j.trecan.2019.02.011 · Cited by 145
Antigen Presentation of mRNA-Based and Virus-Vectored SARS-CoV-2 Vaccines
Rijkers, G. T., Weterings, N., Obregón‐Henao, A., Dutt, T., Dutt, T. S., van Overveld, F. J., & Henao-Tamayo, M. I. (2021) · Vaccines · doi:10.3390/vaccines9080848 · Cited by 139
Direct in vivo CAR T cell engineering
Short, L., Holt, R. A., Cullis, P. R., & Evgin, L. (2024) · Trends in Pharmacological Sciences · doi:10.1016/j.tips.2024.03.004 · Cited by 141
Singh, S., Sharma, P. C., Pal, N., Sarma, D. K., Tiwari, R. R., & Kumar, M. (2024) · ACS Infectious Diseases · doi:10.1021/acsinfecdis.3c00625 · Cited by 122
Machine Learning for Designing Next-Generation mRNA Therapeutics
Castillo-Hair, S. M., & Seelig, G. (2021) · Accounts of Chemical Research · doi:10.1021/acs.accounts.1c00621 · Cited by 122
Self-citationNon-immunotherapy application of LNP-mRNA: Maximizing efficacy and safety
Vlatkovic, I. (2021) · Biomedicines · doi:10.3390/biomedicines9050530 · Cited by 112
Sapkota, D., Lake, A. M., Yang, W., Yang, C., Wesseling, H., Guise, A. J., Uncu, C., Dalal, J. S., Kraft, A. W., Lee, J., et al. (2019) · Cell Reports · doi:10.1016/j.celrep.2018.12.077 · Cited by 113
Recent advances in nanoparticulate RNA delivery systems
Witten, J. J., Hu, Y., Langer, R., & Anderson, D. G. (2024) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2307798120 · Cited by 113
Hold out the genome: a roadmap to solving the cis-regulatory code
de Boer, C. G., & Taipale, J. (2024) · Nature · doi:10.1038/s41586-023-06661-w · Cited by 107
Linder, J., Bogard, N., Rosenberg, A. B., & Seelig, G. (2020) · Cell Systems · doi:10.1016/j.cels.2020.05.007 · Cited by 107
Self-citationHeterogeneity in mRNA Translation
Sonneveld, S., Verhagen, B. M. P., & Tanenbaum, M. E. (2020) · Trends in Cell Biology · doi:10.1016/j.tcb.2020.04.008 · Cited by 106
Controlling gene expression with deep generative design of regulatory DNA
Zrimec, J., Fu, X., Muhammad, A. S., Skrekas, C., Jauniškis, V., Speicher, N. K., Börlin, C. S., Verendel, V., Chehreghani, M. H., Dubhashi, D., et al. (2022) · Nature Communications · doi:10.1038/s41467-022-32818-8 · Cited by 113
Sato, K., & Hamada, M. (2023) · Briefings in Bioinformatics · doi:10.1093/bib/bbad186 · Cited by 104
Determinants of genome-wide distribution and evolution of uORFs in eukaryotes
Zhang, H., Wang, Y., Wu, X., Tang, X., Wu, C., & Lü, J. (2021) · Nature Communications · doi:10.1038/s41467-021-21394-y · Cited by 105
Advancements of in vitro transcribed mRNA (IVT mRNA) to enable translation into the clinics
Kang, D. D., Li, H., & Dong, Y. (2023) · Advanced Drug Delivery Reviews · doi:10.1016/j.addr.2023.114961 · Cited by 100
Optimizing 5’UTRs for mRNA-delivered gene editing using deep learning
Castillo-Hair, S., Fedak, S., Wang, B., Linder, J., Havens, K., Certo, M., & Seelig, G. (2024) · Nature Communications · doi:10.1038/s41467-024-49508-2 · Cited by 98
Self-citationRNA chemistry and therapeutics
Wang, S., Weissman, D., & Dong, Y. (2025) · Nature Reviews Drug Discovery · doi:10.1038/s41573-025-01237-x · Cited by 109
Emerging Delivery Systems for Enabling Precision Nucleic Acid Therapeutics
Bian, X., Zhou, L., Luo, Z., Liu, G., Hang, Z., Li, H., Li, F., & Wen, Y. (2025) · ACS Nano · doi:10.1021/acsnano.4c11858 · Cited by 101
Characteristics of seismic activity of the Upper Silesian Coal Basin in Poland
Stec, K. (2006) · Geophysical Journal International · doi:10.1111/j.1365-246x.2006.03227.x · Cited by 97
Recent progress in mRNA cancer vaccines
Yao, R., Xie, C., & Xia, X. (2024) · Human Vaccines & Immunotherapeutics · doi:10.1080/21645515.2024.2307187 · Cited by 91
Eliciting antitumor immunity via therapeutic cancer vaccines
Peng, K., Zhao, X., Fu, Y., & Liang, Y. (2025) · Cellular and Molecular Immunology · doi:10.1038/s41423-025-01316-4 · Cited by 104
Movva, R., Greenside, P., Marinov, G. K., Nair, S., Shrikumar, A., & Kundaje, A. B. (2019) · PLoS ONE · doi:10.1371/journal.pone.0218073 · Cited by 90
Application of deep learning in genomics
Liu, J., Li, J., Wang, H., & Yan, J. (2020) · Science China Life Sciences · doi:10.1007/s11427-020-1804-5 · Cited by 89
Synthetic Messenger RNA-Based Vaccines: From Scorn to Hype
Pascolo, S. (2021) · Viruses · doi:10.3390/v13020270 · Cited by 87
Ding, K., Chin, M., Zhao, Y., Huang, W., Mai, B. K., Wang, H., Liu, P., Yang, Y., & Luo, Y. (2024) · Nature Communications · doi:10.1038/s41467-024-50698-y · Cited by 86
mRNA ageing shapes the Cap2 methylome in mammalian mRNA
Despic, V., & Jaffrey, S. R. (2023) · Nature · doi:10.1038/s41586-022-05668-z · Cited by 86
mRNA vaccine sequence and structure design and optimization: Advances and challenges
Jin, L., Zhou, Y., Zhang, S., & Chen, S. (2024) · Journal of Biological Chemistry · doi:10.1016/j.jbc.2024.108015 · Cited by 89
Deep generative models design mRNA sequences with enhanced translational capacity and stability
Zhang, H., Liu, H., Xu, Y., Huang, H., Liu, Y., Wang, J., Qin, Y., Wang, H., Ma, L., Xun, Z., et al. (2025) · Science · doi:10.1126/science.adr8470 · Cited by 98
Scalable Functional Assays for the Interpretation of Human Genetic Variation
Tabet, D. R., Parikh, V. N., Mali, P., Roth, F. P., & Claussnitzer, M. (2022) · Annual Review of Genetics · doi:10.1146/annurev-genet-072920-032107 · Cited by 84
mRNA vaccines: the most recent clinical applications of synthetic mRNA
Kwon, S., Kwon, M., Im, S., Lee, K., & Lee, H. (2022) · Archives of Pharmacal Research · doi:10.1007/s12272-022-01381-7 · Cited by 85
Lim, S., Khoo, R., Juang, Y.-C., Gopal, P., Zhang, H., Yeo, C., Peh, K. M., Teo, J., Ng, S., Henry, B., et al. (2021) · ACS Central Science · doi:10.1021/acscentsci.0c01337 · Cited by 80
iRQC, a surveillance pathway for 40S ribosomal quality control during mRNA translation initiation
Garshott, D. M., An, H., Sundaramoorthy, E., Leonard, M., Vicary, A. C., Harper, J. W., & Bennett, E. J. (2021) · Cell Reports · doi:10.1016/j.celrep.2021.109642 · Cited by 81
Computational biology and artificial intelligence in mRNA vaccine design for cancer immunotherapy
İmani, S., Li, X., Chen, K., Maghsoudloo, M., Kaboli, P. J., Hashemi, M., Khoushab, S., & Li, X. (2025) · Frontiers in Cellular and Infection Microbiology · doi:10.3389/fcimb.2024.1501010 · Cited by 87
Computational Methods for the Pharmacogenetic Interpretation of Next Generation Sequencing Data
Zhou, Y., Fujikura, K., Mkrtchian, S., & Lauschke, V. M. (2018) · Frontiers in Pharmacology · doi:10.3389/fphar.2018.01437 · Cited by 77
mRNA therapeutics: New vaccination and beyond
Wei, H. Y., Zheng, L., & Wang, Z. (2023) · Fundamental Research · doi:10.1016/j.fmre.2023.02.022 · Cited by 77
Fast activation maximization for molecular sequence design
Linder, J., & Seelig, G. (2021) · BMC Bioinformatics · doi:10.1186/s12859-021-04437-5 · Cited by 79
Self-citationLim, Y., Arora, S., Schuster, S. L., Corey, L., Fitzgibbon, M., Wladyka, C. L., Wu, X., Coleman, I. M., Delrow, J. J., Corey, E., et al. (2021) · Nature Communications · doi:10.1038/s41467-021-24445-6 · Cited by 70
Accuracy and data efficiency in deep learning models of protein expression
Nikolados, E.-M., Wongprommoon, A., Aodha, O. M., Cambray, G., & Oyarzún, D. A. (2022) · Nature Communications · doi:10.1038/s41467-022-34902-5 · Cited by 78
Wright, C. F., Quaife, N. M., Ramos-Hernández, L., Danecek, P., Ferla, M. P., Samocha, K. E., Kaplanis, J., Gardner, E. J., Eberhardt, R. Y., Chao, K. R., et al. (2021) · The American Journal of Human Genetics · doi:10.1016/j.ajhg.2021.04.025 · Cited by 71
Bicknell, A. A., Reid, D. W., Licata, M. C., Jones, A. K., Cheng, Y. M., Li, M., Hsiao, C. J., Pepin, C. S., Metkar, M., Levdansky, Y., et al. (2024) · Cell Reports · doi:10.1016/j.celrep.2024.114098 · Cited by 68
Self-citationSiegel, D. A., Tonquèze, O. L., Biton, A., Zaitlen, N., & Erle, D. J. (2021) · G3 Genes Genomes Genetics · doi:10.1093/g3journal/jkab404 · Cited by 66
Höllerer, S., Meng-Papaxanthos, L., Gumpinger, A., Fischer, K., Beisel, C., Borgwardt, K. M., Benenson, Y., & Jeschek, M. (2020) · Nature Communications · doi:10.1038/s41467-020-17222-4 · Cited by 66
mRNA-Based Therapeutics in Cancer Treatment
Sun, H., Zhang, Y., Wang, G., Yang, W., & Xu, Y. (2023) · Pharmaceutics · doi:10.3390/pharmaceutics15020622 · Cited by 65
Big data and deep learning for RNA biology
Hwang, H., Jeon, H., Yeo, N., & Baek, D. (2024) · Experimental & Molecular Medicine · doi:10.1038/s12276-024-01243-w · Cited by 64
Linares-Fernández, S., Moreno, J., Lambert, E., Mercier-Gouy, P., Vachez, L., Verrier, B., & Exposito, J.-Y. (2021) · Molecular Therapy - Nucleic Acids · doi:10.1016/j.omtn.2021.10.007 · Cited by 62
Peña-Martínez, E. G., & Rodríguez‐Martínez, J. A. (2024) · Frontiers in Bioscience-Scholar · doi:10.31083/j.fbs1601004 · Cited by 63
Direct analysis of ribosome targeting illuminates thousand-fold regulation of translation initiation
Niederer, R. O., Rojas-Durán, M. F., Zinshteyn, B., & Gilbert, W. V. (2022) · Cell Systems · doi:10.1016/j.cels.2021.12.002 · Cited by 60
UNI-RNA: UNIVERSAL PRE-TRAINED MODELS REVOLUTIONIZE RNA RESEARCHPreprint
Wang, X., Gu, R., Chen, Z., Li, Y., Ji, X.-H., Ke, G., & Wen, H. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.07.11.548588 · Cited by 65
Predicting mean ribosome load for 5’UTR of any length using deep learning
Karollus, A., Avsec, Ž., & Gagneur, J. (2021) · PLOS Computational Biology · doi:10.1371/journal.pcbi.1008982 · Cited by 57
Messenger RNA vaccines for cancer immunotherapy: progress promotes promise
Huff, A. L., Jaffee, E. M., & Zaidi, N. (2022) · Journal of Clinical Investigation · doi:10.1172/jci156211 · Cited by 60
ERNIE-RNA: an RNA language model with structure-enhanced representations
Yin, W., Zhang, Z., Zhang, S., He, L., Zhang, R., Jiang, R., Liu, G., Wang, J., Zhang, X., Qin, T., et al. (2025) · Nature Communications · doi:10.1038/s41467-025-64972-0 · Cited by 55
Findlay, G. M. (2021) · Human Molecular Genetics · doi:10.1093/hmg/ddab219 · Cited by 58
Integrated mRNA sequence optimization using deep learning
Gong, H., Wen, J., Luo, R., Feng, Y., Guo, J., Fu, H., & Zhou, X. (2023) · Briefings in Bioinformatics · doi:10.1093/bib/bbad001 · Cited by 57
Predicting the translation efficiency of messenger RNA in mammalian cells
Zheng, D., Persyn, L., Wang, J., Liu, Y., Ulloa-Montoya, F., Cenik, C., & Agarwal, V. (2026) · Nature Biotechnology · doi:10.1038/s41587-025-02712-x · Cited by 52
pegIT - a web-based design tool for prime editing
Anderson, M. V., Haldrup, J., Thomsen, E. A., Wolff, J. H., & Mikkelsen, J. G. (2021) · Nucleic Acids Research · doi:10.1093/nar/gkab427 · Cited by 52
Therapeutic mRNA Engineering from Head to Tail
Jia, L., & Qian, S. (2021) · Accounts of Chemical Research · doi:10.1021/acs.accounts.1c00541 · Cited by 51
Massively parallel identification of mRNA localization elements in primary cortical neurons
Mendonsa, S., von Kügelgen, N., Dantsuji, S., Ron, M., Breimann, L., Baranovskii, A., Loedige, I., Kirchner, M., Fischer, M., Zerna, N., et al. (2023) · Nature Neuroscience · doi:10.1038/s41593-022-01243-x · Cited by 51
Model-based reinforcement learning for biological sequence design
Angermueller, C., Dohan, D., Belanger, D., Deshpande, R., Murphy, K. J., & Colwell, L. J. (2020) · International Conference on Learning Representations · Cited by 51
Advances in the design and delivery of RNA vaccines for infectious diseases
Lokras, A. G., Bobak, T. R., Baghel, S. S., Sebastiani, F., & Foged, C. (2024) · Advanced Drug Delivery Reviews · doi:10.1016/j.addr.2024.115419 · Cited by 52
Exogenous RNA surveillance by proton-sensing TRIM25
Kim, M., Pyo, Y., Hyun, S., Jeong, M., Choi, Y., & Kim, V. N. (2025) · Science · doi:10.1126/science.ads4539 · Cited by 49
An Update on mRNA-Based Viral Vaccines
Jeeva, S., Kim, K., Shin, C. H., Wang, B., & Kang, S. (2021) · Vaccines · doi:10.3390/vaccines9090965 · Cited by 49
Self-citationDeciphering the impact of genetic variation on human polyadenylation using APARENT2
Linder, J., Koplik, S. E., Kundaje, A., & Seelig, G. (2022) · Genome Biology · doi:10.1186/s13059-022-02799-4 · Cited by 49
Self-citationComputational design of mRNA vaccines
Kim, Y.-A., Mousavi, K., Yazdi, A. K., Zwierzyna, M., Cardinali, M., Fox, D. M., Peel, T., Coller, J., Aggarwal, K., & Maruggi, G. (2023) · Vaccine · doi:10.1016/j.vaccine.2023.07.024 · Cited by 47
Tang, X., Huo, M., Chen, Y., Huang, H., Qin, S., Luo, J., Qin, Z., Jiang, X., Liu, Y., Duan, X., et al. (2024) · Acta Pharmaceutica Sinica B · doi:10.1016/j.apsb.2023.11.003 · Cited by 47
Polyplex designs for improving the stability and safety of RNA therapeutics
Uchida, S., Lau, C. Y. J., Oba, M., & Miyata, K. (2023) · Advanced Drug Delivery Reviews · doi:10.1016/j.addr.2023.114972 · Cited by 46
Therapeutic synthetic and natural materials for immunoengineering
Slezak, A., Chang, K., Hossainy, S., Mansurov, A., Rowan, S. J., Hubbell, J. A., & Guler, M. O. (2024) · Chemical Society Reviews · doi:10.1039/d3cs00805c · Cited by 44
Learning the Regulatory Code of Gene Expression
Zrimec, J., Buric, F., Kokina, M., García, V., & Zelezniak, A. (2021) · Frontiers in Molecular Biosciences · doi:10.3389/fmolb.2021.673363 · Cited by 45
Metadata
- Authors
- Sample, P. J., Wang, B., Reid, D. W., Presnyak, V., McFadyen, I. J., Morris, D. R., & Seelig, G.
- Affiliations
- University of Washington (US)
- Moderna Therapeutics (United States) (US)
- Corresponding author
- Seelig, Georg
- Venue
- Nature Biotechnology
- Published
- 2019-07
- Volume
- 37
- Issue
- 7
- Pages
- 803-809
- ISSN
- 0733-222X (ISSN-L)
- 1087-0156
- 1546-1696
- PMC
- PMC7100133
- OpenAlex
- W2955231772
- Cited by
- 442
- References
- 52
- Citations by year
- 2026: 71 · 2025: 92 · 2024: 96 · 2023: 58 · 2022: 41 · 2021: 47 · 2020: 21 · 2019: 14
- FWCI
- 10.18
- Type
- journal article
- Language
- en
- Open access
- green
- doi.org
- Licence
- Locations
- Nature Biotechnology (publishedVersion)
- bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nc-nd)
- PubMed (publishedVersion)
- PubMed Central (acceptedVersion)
- Zenodo (CERN European Organization for Nuclear Research) (submittedVersion, cc-by)
- Funders
- U.S. Department of Health & Human Services | NIH | National Human Genome Research Institute (R01CA207029)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › RNA and protein synthesis mechanisms
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › RNA Research and Splicing
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › RNA modifications and cancer
- Keywords
- 5' untranslated region, translation initiation, ribosome loading, cis-regulatory sequences, variant effect prediction, single nucleotide variants, RNA modification, mRNA therapeutics, synthetic biology, deep learning, genetic algorithm, sequence design, polysome profiling, disease-associated variants
- MeSH terms
- Base Sequence, Gene Expression Regulation, Humans, Models, Genetic, Pseudouridine, Ribosomes, RNA, Messenger, Protein Biosynthesis, Reproducibility of Results, 5' Untranslated Regions
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 2026-11-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-11-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
