Comparison of RNA-seq and microarray-based models for clinical endpoint prediction
Open accessAbstract
Background: Gene expression profiling is being widely applied in cancer research to identify biomarkers for clinical endpoint prediction. Since RNA-seq provides a powerful tool for transcriptome-based applications beyond the limitations of microarrays, we sought to systematically evaluate the performance of RNA-seq-based and microarray-based classifiers in this MAQC-III/SEQC study for clinical endpoint prediction using neuroblastoma as a model. Results: We generate gene expression profiles from 498 primary neuroblastomas using both RNA-seq and 44 k microarrays. Characterization of the neuroblastoma transcriptome by RNA-seq reveals that more than 48,000 genes and 200,000 transcripts are being expressed in this malignancy. We also find that RNA-seq provides much more detailed information on specific transcript expression patterns in clinico-genetic neuroblastoma subgroups than microarrays. To systematically compare the power of RNA-seq and microarray-based models in predicting clinical endpoints, we divide the cohort randomly into training and validation sets and develop 360 predictive models on six clinical endpoints of varying predictability. Evaluation of factors potentially affecting model performances reveals that prediction accuracies are most strongly influenced by the nature of the clinical endpoint, whereas technological platforms (RNA-seq vs. microarrays), RNA-seq data analysis pipelines, and feature levels (gene vs. transcript vs. exon-junction level) do not significantly affect performances of the models. Conclusions: We demonstrate that RNA-seq outperforms microarrays in determining the transcriptomic characteristics of cancer, while RNA-seq and microarray-based models perform similarly in clinical endpoint prediction. Our findings may be valuable to guide future studies on the development of gene expression-based predictive models and their implementation in clinical practice.
AI snapshot
Written from the paper’s full text.
TL;DR
In 498 primary neuroblastomas, RNA-seq provided more detailed transcriptome characterization than microarrays, but RNA-seq- and microarray-based classifiers performed similarly across six clinical endpoints. The nature of the clinical endpoint was the dominant factor affecting prediction accuracy, with RNA-seq significantly better only for EFS ALL (P = 0.043).
Key findings
- RNA-seq detected 48,415 expressed genes and 204,352 transcripts in the entire neuroblastoma cohort, including >39,000 novel exons; the 44 k microarray represented 21,101 AceView genes.
- In four clinico-genetic subgroups, RNA-seq identified 21,315 DEGs versus 11,688 DEGs by microarrays; RNA-seq captured 80.1% of microarray DEGs, and 65.9% of RNA-seq DEGs showed discordant transcript expression patterns.
- Across 360 predictive models for six endpoints, endpoint effects explained 95% of variation in MCC; platform, RNA-seq pipeline, and feature level did not significantly affect performance, except RNA-seq models performed better for EFS ALL (P = 0.043).
- Models with 100 to 1,000 features performed best on average, while easy endpoints SEX and CLASS LABEL required only 1 to 10 features; internal-external validation correlations were similar for microarray (r = 0.716) and RNA-seq (r = 0.723).
Methods
- The study used 498 primary neuroblastoma samples from seven countries, with gene expression profiles generated by RNA-seq and Agilent 44 k oligonucleotide microarrays.
- Samples were randomly divided into training and validation sets; six data analysis teams developed models for six clinical endpoints (SEX, CLASS LABEL, EFS ALL, OS ALL, EFS HR, OS HR), yielding 360 models across 10 expression profiles per sample (nine RNA-seq profiles from three pipelines and three feature levels, plus one microarray profile).
- RNA-seq pipelines were Magic-AceView, TopHat-AceView, and TopHat-UCSC; feature levels were gene, transcript, and exon-junction. Performance was evaluated by MCC, with variance component analysis using mixed modeling and survival analyses by Cox regression and Kaplan-Meier estimates.
Limitations
- The paper states that all analyses are regarded as explorative, that prognostic value of multigene signatures needs validation in independent high-risk neuroblastoma cohorts, and that models for EFS HR and OS HR tended to be biased toward internal validation, indicating overfitting.
- Evident from the text: validation used a random split of the same 498-sample neuroblastoma cohort rather than an independent external cohort, and findings are limited to neuroblastoma and the six selected endpoints, so generalizability to other cancers or clinical settings is not established.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
0 of 30 cited sources checked · 0 supported · 0 flagged
- Findingown data
RNA-seq supported 204,352 transcripts annotated in AceView in the neuroblastoma cohort.
“Furthermore, the mapped reads supported a total of 204,352 transcripts annotated in AceView, comprising 319,231 annotated exon-junctions (Fig. 1b; mapping results using the RefSeq and Gencode databases as references are given in Additional file 1: Figure S4).”
Quote in the paperNot a substantial claim - Findingown data
RNA-seq identified 48,415 genes expressed above the background threshold in the entire neuroblastoma cohort using AceView as a reference database.
“In the entire neuroblastoma cohort, we identified 48,415 genes expressed above the background threshold when using AceView as a reference database (Fig. 1b), corresponding to an average of 28,490 (±1,399) expressed genes per sample (Additional file 1: Figure S3).”
Quote in the paperNot a substantial claim - Findingown data
Prediction accuracies are most strongly influenced by the nature of the clinical endpoint.
“Evaluation of factors potentially affecting model performances reveals that prediction accuracies are most strongly influenced by the nature of the clinical endpoint, whereas technological platforms (RNA-seq vs. microarrays), RNA-seq data analysis pipelines, and feature levels (gene vs. transcript vs. exon-junction level) do not significantly affect performances of the models.”
- Also stated in: § Clinical endpoint prediction by RNA-seq- and microarray-based models · boosted
“In line with our findings from the MAQC II study [7], we noticed that prediction performances were strongly influenced by the respective clinical endpoint.”
Quote in the paperSubstantial claim - Also stated in: § Clinical endpoint prediction by RNA-seq- and microarray-based models · boosted
- Findingown data
Technological platforms (RNA-seq vs. microarrays) do not significantly affect performances of the models.
“Evaluation of factors potentially affecting model performances reveals that prediction accuracies are most strongly influenced by the nature of the clinical endpoint, whereas technological platforms (RNA-seq vs. microarrays), RNA-seq data analysis pipelines, and feature levels (gene vs. transcript vs. exon-junction level) do not significantly affect performances of the models.”
Quote in the paperSubstantial claim - Findingown data
RNA-seq data analysis pipelines do not significantly affect performances of the models.
“Evaluation of factors potentially affecting model performances reveals that prediction accuracies are most strongly influenced by the nature of the clinical endpoint, whereas technological platforms (RNA-seq vs. microarrays), RNA-seq data analysis pipelines, and feature levels (gene vs. transcript vs. exon-junction level) do not significantly affect performances of the models.”
- Also stated in: § Variables affecting prediction performances of RNA-seq-based models · plain
“In general, models derived from distinct processing pipelines and distinct feature levels did not differ significantly in their prediction performances with few exceptions (Fig. 3a and b).”
Quote in the paperSubstantial claim - Also stated in: § Variables affecting prediction performances of RNA-seq-based models · plain
- Findingown data
RNA-seq and microarray-based models perform similarly in clinical endpoint prediction.
“We demonstrate that RNA-seq outperforms microarrays in determining the transcriptomic characteristics of cancer, while RNA-seq and microarray-based models perform similarly in clinical endpoint prediction.”
- Also stated in: § Variables affecting prediction performances of RNA-seq-based models · boosted
“Taken together, our data demonstrate that microarray and RNA-seq models perform similar in clinical endpoint prediction, and that the RNA-seq processing pipeline and the feature level do not influence prediction performances systematically.”
- Also stated in: § Discussion · boosted
“Collectively, our data demonstrate that RNA-seq and microarray-based models perform similarly in clinical endpoint prediction.”
- Also stated in: § Conclusions · boosted
“Our study demonstrates that RNA-seq based models are suitable for clinical endpoint prediction, and that prediction performances are similar to those of microarray-based models.”
Quote in the paperSubstantial claim - Also stated in: § Variables affecting prediction performances of RNA-seq-based models · boosted
- Findingown data
Feature levels (gene vs. transcript vs. exon-junction level) do not significantly affect performances of the models.
“Evaluation of factors potentially affecting model performances reveals that prediction accuracies are most strongly influenced by the nature of the clinical endpoint, whereas technological platforms (RNA-seq vs. microarrays), RNA-seq data analysis pipelines, and feature levels (gene vs. transcript vs. exon-junction level) do not significantly affect performances of the models.”
- Also stated in: § Variables affecting prediction performances of RNA-seq-based models · plain
“In general, models derived from distinct processing pipelines and distinct feature levels did not differ significantly in their prediction performances with few exceptions (Fig. 3a and b).”
Quote in the paperSubstantial claim - Also stated in: § Variables affecting prediction performances of RNA-seq-based models · plain
- Findingown data
In the high-risk cohort, the best performing models were able to predict patient outcome, in contrast to the prognostic markers age, stage and MYCN status.
“In the high-risk cohort, the best performing models were also able to predict patient outcome, which was in contrast to the prognostic markers age, stage and MYCN status (Additional file 1: Figure S9e and f, Fig. S10, and Table S5).”
Quote in the paperSubstantial claim - Findingown data
RNA-seq detected 54,164 differentially expressed transcripts corresponding to 21,315 DEGs.
“In total, we detected 54,164 differentially expressed transcripts corresponding to 21,315 DEGs using RNA-seq (RefSeq, n = 14,251; AceView only, n = 7,064; Additional file 3).”
Quote in the paperSubstantial claim - Findingown data
The size of the model was the only factor that significantly affected prediction performances independent of the endpoint.
“The size of the model (that is, the number of features included in the model) was the only factor that significantly affected prediction performances independent of the endpoint.”
Quote in the paperNot a substantial claim - Findingown data
Models derived from distinct processing pipelines and distinct feature levels did not differ significantly in their prediction performances with few exceptions.
“In general, models derived from distinct processing pipelines and distinct feature levels did not differ significantly in their prediction performances with few exceptions (Fig. 3a and b).”
Quote in the paperNot a substantial claim - Findingown data
RNA-seq detected 5,488 (69.2%) of the DEGs identified by microarrays using the MAQC-I method.
“On the platform level, we found that RNA-seq detected 5,488 (69.2 %) and 7,827 (67.0 %) of the DEGs identified by microarrays using the MAQC-I and Magic methods, respectively (Additional file 1: Figure S5).”
Quote in the paperSubstantial claim - Findingown data
RNA-seq identified many more genes than microarrays.
“Notably, RNA-seq analysis identified 80.1 % of the DEGs detected by microarrays (Fig. 1c), but in addition many more genes (annotated both in the RefSeq database and in the AceView database only) than microarrays (Additional file 1: Figure S6a).”
Quote in the paperSubstantial claim - Findingown data
Models based on the two platforms performed similarly well in predicting the remaining five endpoints.
“RNA-seq-based models performed significantly better than microarray-based models in predicting endpoint EFS ALL (P = 0.043), while models based on the two platforms performed similarly well in predicting the remaining five endpoints (Fig. 2b).”
Quote in the paperSubstantial claim - Findingown data
More than 39,000 novel exons are expressed in neuroblastoma.
“We also identified >39,000 novel exons to be expressed in neuroblastoma, providing further evidence that the human transcriptome still exceeds the complexity reflected by current reference databases such as RefSeq, Gencode, and AceView.”
Quote in the paperNot a substantial claim - Findingown data
6,939 DEGs identified by RNA-seq were not represented by a probe on the microarray.
“One reason for the discrepant numbers received by RNA-seq and microarrays derives from the fact that 6,939 DEGs identified by RNA-seq were not represented by a probe on the microarray.”
Quote in the paperSubstantial claim - Findingown data
Different analysis teams and different modeling methods had varying performances across the endpoints.
“In addition, different analysis teams and different modeling methods had varying performances across the endpoints (Additional file 1: Figure S11b and c).”
Quote in the paperNot a substantial claim - Findingown data
Correlations of internal and external validation performances for microarray- and RNA-seq-based models were almost identical (r = 0.716 and r = 0.723, respectively).
“Strikingly, correlations of internal and external validation performances for microarray- and RNA-seq-based models were almost identical (r = 0.716 and r = 0.723, respectively), indicating that the technical platform does not affect the reliability of performance estimates obtained during model training.”
Quote in the paperSubstantial claim - Findingown data
RNA-seq identified almost twice as many DEGs as microarrays when comparing gene expression profiles of four major clinico-genetic subgroups.
“The comparison of gene expression profiles of four major clinico-genetic subgroups revealed that RNA-seq identified almost twice as many DEGs as microarrays.”
Quote in the paperNot a substantial claim - Findingown data
RNA-seq analysis identified 80.1% of the DEGs detected by microarrays.
“Notably, RNA-seq analysis identified 80.1 % of the DEGs detected by microarrays (Fig. 1c), but in addition many more genes (annotated both in the RefSeq database and in the AceView database only) than microarrays (Additional file 1: Figure S6a).”
Quote in the paperSubstantial claim - Findingown data
The endpoint represents the most relevant factor affecting model performances.
“Evaluation of the prediction performances in the validation set revealed that the endpoint represents the most relevant factor affecting model performances, which is well in line with the findings of the MAQC-II study [7].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [7] Shi L, Campbell G, Jones WD, Campagne F, Wen Z, Walker SJ, et al. The MicroArray Quality Control (MAQC)-II study of common practices for the development and validation of microarray-based predictive models. Nat Biotechnol. 2010;28:827–38. source
Not checked in the free run.
- Findingown data
DEGs determined by RNA-seq comprised 80.1% of the DEGs detected by microarrays.
“Of note, DEGs determined by RNA-seq comprised 80.1 % of the DEGs detected by microarrays, pointing towards the reliability of identifying DEGs by either method.”
Quote in the paperSubstantial claim - Findingown data
Magic detected 7,423 (93.5%) of the DEGs identified by the MAQC-I method using microarray data.
“On the analytical method level, Magic detected 7,423 (93.5 %) and 6,728 (80.4 %) of the DEGs identified by the MAQC-I method using microarray and RNA-seq data, respectively (Additional file 1: Figure S5).”
Quote in the paperNot a substantial claim - Findingown data
The remaining 3.5% were residual variance not explained by any of the factors under investigation.
“The remaining 3.5 % were residual variance not explained by any of the factors under investigation (Fig. 4a).”
Quote in the paperNot a substantial claim - Findingown data
Microarray probes found 16,245 differentially represented probes corresponding to 11,688 DEGs.
“By contrast, the 16,245 microarray probes found to be differentially represented in the four subgroups correspond to 11,688 DEGs (RefSeq, n = 10,308; AceView only, n = 1,380).”
Quote in the paperSubstantial claim - Findingreasoning
4,776 DEGs were not detected by microarrays although the genes were represented by a probe.
“In addition, 4,776 DEGs were not detected by microarrays although the genes were represented by a probe, which may be at least partly attributed to our analytical approach which was taking expression profiles at the transcript level into account.”
Quote in the paperSubstantial claim - Findingown data
Magic detected 6,728 (80.4%) of the DEGs identified by the MAQC-I method using RNA-seq data.
“On the analytical method level, Magic detected 7,423 (93.5 %) and 6,728 (80.4 %) of the DEGs identified by the MAQC-I method using microarray and RNA-seq data, respectively (Additional file 1: Figure S5).”
Quote in the paperNot a substantial claim - Findingown data
On average, models comprising 100 to 1,000 features gave the best prediction results.
“On average, models comprising 100 to 1,000 features gave the best prediction results (Fig. 4b).”
Quote in the paperNot a substantial claim - Findingown data
The results do not support the hypothesis that the more extensive transcriptomic information provided by RNA-seq in comparison to microarrays may improve gene expression-based prediction performances in general.
“Our results do not support the hypothesis that the more extensive transcriptomic information provided by RNA-seq in comparison to microarrays may improve gene expression-based prediction performances in general.”
Quote in the paperSubstantial claim - Findingown data
Performances were substantially inferior for predicting EFS and OS in the entire cohort.
“While patient’s SEX and CLASS LABEL were predicted accurately, performances were substantially inferior for predicting EFS and OS in the entire cohort, and even worse in the high-risk subgroup by both RNA-seq and microarray-based models (Fig. 2b).”
Quote in the paperSubstantial claim - Findingown data
The proportion of RefSeq-annotated features in the prediction models was in the range of, or even above, their proportion in the AceView database for most endpoints.
“An alternative explanation for our results may be inferred from the observation that the proportion of RefSeq-annotated features in the prediction models was in the range of, or even above their proportion in the AceView database for most endpoints.”
Quote in the paperSubstantial claim - Findingown data
Models with 100 to 1,000 features perform better than those with fewer or more features.
“BLUPs for Log10(Model Size) indicate that models with 100 to 1,000 features perform better than those with fewer or more features”
Quote in the paperSubstantial claim - Findingown data
RNA-seq detected 7,827 (67.0%) of the DEGs identified by microarrays using the Magic method.
“On the platform level, we found that RNA-seq detected 5,488 (69.2 %) and 7,827 (67.0 %) of the DEGs identified by microarrays using the MAQC-I and Magic methods, respectively (Additional file 1: Figure S5).”
Quote in the paperSubstantial claim - Findingown data
Performances were even worse in the high-risk subgroup by both RNA-seq and microarray-based models.
“While patient’s SEX and CLASS LABEL were predicted accurately, performances were substantially inferior for predicting EFS and OS in the entire cohort, and even worse in the high-risk subgroup by both RNA-seq and microarray-based models (Fig. 2b).”
Quote in the paperNot a substantial claim - Findingown data
Of the remaining 5% of variability, 1.5% was explained by four statistically significant effects.
“Of the remaining 5 % of variability, 1.5 % was explained by four statistically significant effects.”
Quote in the paperNot a substantial claim - Findingcitation
Both gene expression-based models derived from RNA-seq and microarray analyses were capable of predicting patient outcome in the entire neuroblastoma cohort accurately.
“Both gene expression-based models derived from RNA-seq and microarray analyses were capable of predicting patient outcome in the entire neuroblastoma cohort accurately, thereby validating results from previous studies and underscoring their potential clinical utility for risk estimation in neuroblastoma [16–18, 20].”
Quote in the paperSubstantial claim0 of 4 sources checked- cites [16] Oberthuer A, Berthold F, Warnat P, Hero B, Kahlert Y, Spitz R, et al. Customized oligonucleotide microarray gene expression-based classification of neuroblastoma patients outperforms current clinical risk stratification. J Clin Oncol. 2006;24:5070–8. source
Not checked in the free run.
- cites [17] Ohira M, Oba S, Nakamura Y, Isogai E, Kaneko S, Nakagawa A, et al. Expression profiling using a tumor-specific cDNA microarray predicts the prognosis of intermediate risk neuroblastomas. Cancer Cell. 2005;7:337–50. source
Not checked in the free run.
- cites [18] Oberthuer A, Hero B, Berthold F, Juraeva D, Faldum A, Kahlert Y, et al. Prognostic impact of gene expression-based classification for neuroblastoma. J Clin Oncol. 2010;28:3506–15. source
Not checked in the free run.
- cites [20] Vermeulen J, De Preter K, Naranjo A, Vercruysse L, Van Roy N, Hellemans J, et al. Predicting outcomes for children with neuroblastoma using a multigene-expression signature: a retrospective SIOPEN/COG/GPOH study. Lancet Oncol. 2009;10:663–71. source
Not checked in the free run.
- Findingown data
Models based on different feature levels predicted clinical endpoints with comparable accuracies.
“We also noticed that models based on different feature levels predicted clinical endpoints with comparable accuracies.”
Quote in the paperSubstantial claim - Findingown data
65.9% of the 21,315 DEGs identified by RNA-seq showed discordant expression patterns.
“We noted that 65.9 % of the 21,315 DEGs identified by RNA-seq showed such discordant expression patterns.”
Quote in the paperSubstantial claim - Findingown data
95% of the variation across MCC values was caused by effects of the endpoints.
“We found that 95 % of the variation across MCC values was caused by effects of the endpoints, with lowest variances observed for endpoints predicted most accurately (Fig. 4a; Additional file 1: Table S7).”
Quote in the paperNot a substantial claim - Findingown data
Models containing 100 to 1,000 features on average performed better than models containing fewer features.
“Notably, we observed that models containing 100 to 1,000 features on average performed better than models containing fewer features.”
Quote in the paperSubstantial claim - Findingown data
RNA-seq-based models performed significantly better than microarray-based models in predicting endpoint EFS ALL (P = 0.043).
“RNA-seq-based models performed significantly better than microarray-based models in predicting endpoint EFS ALL (P = 0.043), while models based on the two platforms performed similarly well in predicting the remaining five endpoints (Fig. 2b).”
Quote in the paperSubstantial claim - Findingown data
Performances of both RNA-seq- and microarray-based models for endpoints EFS HR and OS HR tended to be biased strongly towards internal validation.
“It has to be noted, though, that performances of both RNA-seq- and microarray-based models for endpoints EFS HR and OS HR tended to be biased strongly towards internal validation, which is indicative of model overfitting on the training set.”
Quote in the paperSubstantial claim - Findingown data
Complex expression patterns were observed for 26 of the current 513 cancer census genes.
“Focusing on cancer census genes [28], we observed such complex expression patterns for 26 of the current 513 cancer census genes (Additional file 1: Table S2).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [28] Futreal PA, Coin L, Marshall M, Down T, Hubbard T, Wooster R, et al. A census of human cancer genes. Nat Rev Cancer. 2004;4:177–83. source
Not checked in the free run.
- Findingown data
The correlation of internal and external validation performances was almost identical for RNA-seq and microarray-based models.
“We noted, however, that the correlation of internal and external validation performances was almost identical for RNA-seq and microarray-based models, indicating that over-fitting effects are independent of the technological platform.”
Quote in the paperSubstantial claim - Findingown data
The proportion of RefSeq-annotated features in the models was in the range of their proportion in the AceView database.
“Similarly, the proportion of RefSeq-annotated features in the models was in the range of their proportion in the AceView database.”
Quote in the paperNot a substantial claim - Findingown data
The proportion of coding and spliced features in the models largely reflected the proportion of coding and spliced features in the AceView database.
“We found that the proportion of coding and spliced features in the models largely reflected the proportion of coding and spliced features in the AceView database (Additional file 1: Figure S12 and Table S8).”
Quote in the paperNot a substantial claim - Findingown data
Model size is the single factor contributing significantly to prediction variability independent of the endpoint.
“Best linear unbiased predictor (BLUP) estimates for the log10(model size) as the single factor contributing significantly to the prediction variability independent of the endpoint.”
Quote in the paperSubstantial claim - Findingown data
Prediction accuracies were significantly correlated with the fraction of RefSeq features, coding features, and spliced features in the models both on the gene and the transcript level.
“Taking all endpoints into account, however, we observed that prediction accuracies were significantly correlated with the fraction of RefSeq features, coding features, and spliced features in the models both on the gene and the transcript level (Fig. 5, Additional file 1: Figure S13).”
Quote in the paperSubstantial claim - Premisecitation
In the MAQC-II study, the performance of gene expression models in predicting clinical outcome was limited.
“Within the MicroArray Quality Control (MAQC)-II study [7], we observed, however, that the performance of gene expression models in predicting clinical outcome was limited and largely dependent on the respective clinical endpoint.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [7] Shi L, Campbell G, Jones WD, Campagne F, Wen Z, Walker SJ, et al. The MicroArray Quality Control (MAQC)-II study of common practices for the development and validation of microarray-based predictive models. Nat Biotechnol. 2010;28:827–38. source
Not checked in the free run.
- Interpretationown data
RNA-seq provides much more detailed information on specific transcript expression patterns in clinico-genetic neuroblastoma subgroups than microarrays.
“We also find that RNA-seq provides much more detailed information on specific transcript expression patterns in clinico-genetic neuroblastoma subgroups than microarrays.”
Quote in the paperNot a substantial claim - Premisecitation
In the MAQC-II study, performance of gene expression models was largely dependent on the respective clinical endpoint.
“Within the MicroArray Quality Control (MAQC)-II study [7], we observed, however, that the performance of gene expression models in predicting clinical outcome was limited and largely dependent on the respective clinical endpoint.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [7] Shi L, Campbell G, Jones WD, Campagne F, Wen Z, Walker SJ, et al. The MicroArray Quality Control (MAQC)-II study of common practices for the development and validation of microarray-based predictive models. Nat Biotechnol. 2010;28:827–38. source
Not checked in the free run.
- Premisecitation
Prediction of high-risk neuroblastoma patient outcome has remained challenging.
“However, while predictive models were highly accurate in risk assessment of current low- and intermediate-risk patients [18], the prediction of high-risk patient outcome has remained challenging [18–20].”
Quote in the paperSubstantial claim0 of 3 sources checked- cites [18] Oberthuer A, Hero B, Berthold F, Juraeva D, Faldum A, Kahlert Y, et al. Prognostic impact of gene expression-based classification for neuroblastoma. J Clin Oncol. 2010;28:3506–15. source
Not checked in the free run.
- cites [19] Asgharzadeh S, Pique-Regi R, Sposto R, Wang H, Yang Y, Shimada H, et al. Prognostic significance of gene expression profiles of metastatic neuroblastomas lacking MYCN gene amplification. J Natl Cancer Inst. 2006;98:1193–203. source
Not checked in the free run.
- cites [20] Vermeulen J, De Preter K, Naranjo A, Vercruysse L, Van Roy N, Hellemans J, et al. Predicting outcomes for children with neuroblastoma using a multigene-expression signature: a retrospective SIOPEN/COG/GPOH study. Lancet Oncol. 2009;10:663–71. source
Not checked in the free run.
- Premisecitation
RNA-seq provides a powerful tool to decipher global gene expression patterns far beyond the limitations of microarrays.
“RNA deep-sequencing (RNA-seq) provides a powerful tool to decipher global gene expression patterns far beyond the limitations of microarrays, including an unprecedented capability to discover novel genes, alternative transcript variants, chimeric transcripts, and expressed sequence variants as well as allele-specific expression [8–12].”
Quote in the paperSubstantial claim0 of 5 sources checked- cites [8] Su Z, Łabaj PP, Li S, Thierry-Mieg J, Thierry-Mieg D, Shi W, et al. A comprehensive assessment of RNA-seq accuracy, reproducibility and information content by the Sequencing Quality Control Consortium. Nat Biotech. 2014;32:903–14. source
Not checked in the free run.
- cites [9] Djebali S, Davis CA, Merkel A, Dobin A, Lassmann T, Mortazavi A, et al. Landscape of transcription in human cells. Nature. 2012;489:101–8. source
Not checked in the free run.
- cites [10] Ozsolak F, Milos PM. RNA sequencing: advances, challenges and opportunities. Nat Rev Genet. 2011;12:87–98. source
Not checked in the free run.
- cites [11] Sultan M, Schulz MH, Richard H, Magen A, Klingenhoff A, Scherf M, et al. A global view of gene activity and alternative splicing by deep sequencing of the human transcriptome. Science. 2008;321:956–60. source
Not checked in the free run.
- cites [12] Ferreira PG, Jares P, Rico D, Gomez-Lopez G, Martinez-Trillos A, Villamor N, et al. Transcriptome characterization by RNA sequencing identifies a major molecular and clinical subdivision in chronic lymphocytic leukemia. Genome Res. 2014;24:212–26. source
Not checked in the free run.
- Interpretationown data
RNA-seq is a powerful tool to determine the complex transcriptomic characteristics of cancer.
“Together, our findings substantiate that RNA-seq is a powerful tool to determine the complex transcriptomic characteristics of cancer.”
Quote in the paperSubstantial claim - Interpretationreasoning
The endpoint itself influences the reliability of performance estimates derived from training cohorts.
“These results suggest that the endpoint itself influences the reliability of performance estimates derived from training cohorts.”
Quote in the paperSubstantial claim - Interpretationown data
RNA-seq is capable of providing much more detailed insights into the transcriptomic characteristics of neuroblastoma than microarrays.
“Taken together, our study substantiates that RNA-seq is capable of providing much more detailed insights into the transcriptomic characteristics of neuroblastoma than microarrays.”
Quote in the paperSubstantial claim - Interpretationown data
The technical platform does not affect the reliability of performance estimates obtained during model training.
“Strikingly, correlations of internal and external validation performances for microarray- and RNA-seq-based models were almost identical (r = 0.716 and r = 0.723, respectively), indicating that the technical platform does not affect the reliability of performance estimates obtained during model training.”
Quote in the paperSubstantial claim - Interpretationown data
The human transcriptome still exceeds the complexity reflected by current reference databases such as RefSeq, Gencode, and AceView.
“We also identified >39,000 novel exons to be expressed in neuroblastoma, providing further evidence that the human transcriptome still exceeds the complexity reflected by current reference databases such as RefSeq, Gencode, and AceView.”
Quote in the paperNot a substantial claim - Interpretationreasoning
The bias towards internal validation for EFS HR and OS HR is indicative of model overfitting on the training set.
“It has to be noted, though, that performances of both RNA-seq- and microarray-based models for endpoints EFS HR and OS HR tended to be biased strongly towards internal validation, which is indicative of model overfitting on the training set.”
Quote in the paperSubstantial claim - Interpretationown data
The methods used in the study are valid.
“Together, these results demonstrate that both the different platforms and the different analytical methods provide largely comparable results, indicating the validity of the methods used in our study.”
Quote in the paperSubstantial claim - Interpretationreasoning
Models based on exon-junction levels perform equally well as models based on gene levels.
“In turn, this result implies that models based on exon-junction levels perform equally well as models based on gene levels.”
Quote in the paperSubstantial claim - Interpretationown data
Over-fitting effects are independent of the technological platform.
“We noted, however, that the correlation of internal and external validation performances was almost identical for RNA-seq and microarray-based models, indicating that over-fitting effects are independent of the technological platform.”
Quote in the paperSubstantial claim - Interpretationreasoning
The 4,776 undetected DEGs may be at least partly attributed to the analytical approach taking expression profiles at the transcript level into account.
“In addition, 4,776 DEGs were not detected by microarrays although the genes were represented by a probe, which may be at least partly attributed to our analytical approach which was taking expression profiles at the transcript level into account.”
Quote in the paperSubstantial claim - Study design and conductcitation
Endpoints were included that cover a broad range of predictive difficulties.
“Considering the fact that the nature of the clinical endpoint strongly affects a classifier’s performance [7, 18], we included endpoints that are known to cover a broad range of predictive difficulties (low difficulty, SEX, and CLASS LABEL; intermediate difficulty, EFS ALL, and OS ALL; high difficulty, EFS HR, and OS HR).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [7] Shi L, Campbell G, Jones WD, Campagne F, Wen Z, Walker SJ, et al. The MicroArray Quality Control (MAQC)-II study of common practices for the development and validation of microarray-based predictive models. Nat Biotechnol. 2010;28:827–38. source
Not checked in the free run.
- cites [18] Oberthuer A, Hero B, Berthold F, Juraeva D, Faldum A, Kahlert Y, et al. Prognostic impact of gene expression-based classification for neuroblastoma. J Clin Oncol. 2010;28:3506–15. source
Not checked in the free run.
- Interpretationown data
The high overlap (80.1%) points towards the reliability of identifying DEGs by either method.
“Of note, DEGs determined by RNA-seq comprised 80.1 % of the DEGs detected by microarrays, pointing towards the reliability of identifying DEGs by either method.”
Quote in the paperSubstantial claim - Interpretationreasoning
Exon-junctions identified by RNA-seq provide an unambiguous source of expression information for developing specific diagnostic tests.
“While assays based on gene expression levels may lack specificity due to uncertainties on the underlying relevant transcript variants, exon-junctions identified by RNA-seq provide an unambiguous source of expression information for developing specific diagnostic tests.”
Quote in the paperSubstantial claim - Study design and conductnone given
A two-step modeling strategy was implemented to ensure unbiased validation: models were developed on a training set and frozen, then the validation set was released for evaluation of the frozen models.
“We implemented a two-step modeling strategy to ensure an unbiased validation: first, the models were developed based on a training set and frozen; then, the validation set was released for evaluation of the frozen models.”
Quote in the paperSubstantial claim - Interpretationown data
The composition of prediction models depends on the predictability of the clinical endpoints in general.
“These data indicate that the composition of prediction models depends on the predictability of the clinical endpoints in general.”
Quote in the paperSubstantial claim - Interpretationown data
RNA-seq based models are suitable for clinical endpoint prediction.
“Our study demonstrates that RNA-seq based models are suitable for clinical endpoint prediction, and that prediction performances are similar to those of microarray-based models.”
Quote in the paperSubstantial claim
- May combine two claims. 9
- Classified as probably not a substantial claim. 9
May combine two claims.
Findingown dataThe best performing models were able to predict outcome of high-risk patients with a similar precision as previously published multigene signatures, and independently from current prognostic markers.
“In addition, we found that the best performing models were able to predict outcome of high-risk patients with a similar precision as previously published multigene signatures [18, 20, 33], and independently from current prognostic markers.”
Quote in the paperSubstantial claim0 of 3 sources checked- cites [18] Oberthuer A, Hero B, Berthold F, Juraeva D, Faldum A, Kahlert Y, et al. Prognostic impact of gene expression-based classification for neuroblastoma. J Clin Oncol. 2010;28:3506–15. source
Not checked in the free run.
- cites [20] Vermeulen J, De Preter K, Naranjo A, Vercruysse L, Van Roy N, Hellemans J, et al. Predicting outcomes for children with neuroblastoma using a multigene-expression signature: a retrospective SIOPEN/COG/GPOH study. Lancet Oncol. 2009;10:663–71. source
Not checked in the free run.
- cites [33] Stricker TP. Morales La Madrid A, Chlenski A, Guerrero L, Salwen HR, Gosiengfiao Y, et al. Validation of a prognostic multi-gene signature in high-risk neuroblastoma using the high throughput digital NanoString nCounter system. Mol Oncol. 2014;8:669–78. source
Not checked in the free run.
May combine two claims.
Findingown dataIn the entire neuroblastoma cohort, 48,415 genes and 204,352 transcripts were expressed, comprising 86.7% and 77.3% of all features annotated in the AceView database, respectively.
“In the entire neuroblastoma cohort, we found 48,415 genes and 204,352 transcripts to be expressed, comprising 86.7 % and 77.3 % of all features annotated in the AceView database, respectively.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataThe best performing models significantly discriminated patients with favorable and unfavorable outcome in the entire cohort and in the class labeled cohort.
“The models significantly discriminated patients with favorable and unfavorable outcome in the entire cohort and in the class labeled cohort (Additional file 1: Figure S9a-d, Table S3 and S4), which is consistent with previous observations made by us and others [16–18, 20].”
Quote in the paperSubstantial claim0 of 4 sources checked- cites [16] Oberthuer A, Berthold F, Warnat P, Hero B, Kahlert Y, Spitz R, et al. Customized oligonucleotide microarray gene expression-based classification of neuroblastoma patients outperforms current clinical risk stratification. J Clin Oncol. 2006;24:5070–8. source
Not checked in the free run.
- cites [17] Ohira M, Oba S, Nakamura Y, Isogai E, Kaneko S, Nakagawa A, et al. Expression profiling using a tumor-specific cDNA microarray predicts the prognosis of intermediate risk neuroblastomas. Cancer Cell. 2005;7:337–50. source
Not checked in the free run.
- cites [18] Oberthuer A, Hero B, Berthold F, Juraeva D, Faldum A, Kahlert Y, et al. Prognostic impact of gene expression-based classification for neuroblastoma. J Clin Oncol. 2010;28:3506–15. source
Not checked in the free run.
- cites [20] Vermeulen J, De Preter K, Naranjo A, Vercruysse L, Van Roy N, Hellemans J, et al. Predicting outcomes for children with neuroblastoma using a multigene-expression signature: a retrospective SIOPEN/COG/GPOH study. Lancet Oncol. 2009;10:663–71. source
Not checked in the free run.
May combine two claims.
Findingown dataRNA-seq identified 1,073 DEGs with at least one upregulated and one downregulated transcript variant, as opposed to 129 by microarrays.
“In this category, we identified 1,073 DEGs by RNA-seq, as opposed to 129 DEGs by microarrays (Additional file 1: Figure S6b).”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataFor prediction of SEX and CLASS LABEL, only one to 10 features were required, while more complex models improved prediction results for the remaining endpoints.
“The model size effect, however, differed between the endpoints analyzed: For prediction of the two endpoints that were easy to predict (SEX and CLASS LABEL), only one to 10 features were required, while more complex models improved prediction results for the remaining endpoints (Additional file 1: Figure S11a).”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataPatient's SEX and CLASS LABEL were predicted accurately.
“While patient’s SEX and CLASS LABEL were predicted accurately, performances were substantially inferior for predicting EFS and OS in the entire cohort, and even worse in the high-risk subgroup by both RNA-seq and microarray-based models (Fig. 2b).”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataNeither the technical platform (RNA-seq vs. microarrays) nor the RNA-seq data annotation pipeline significantly affected the variability of prediction performances.
“By contrast, neither the technical platform (that is, RNA-seq vs. microarrays) nor the RNA-seq data annotation pipeline significantly affected the variability of prediction performances.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationA number of microarray-based gene expression models have been proposed to predict neuroblastoma patient outcome.
“In addition, a number of microarray-based gene expression models have been proposed to predict neuroblastoma patient outcome [16, 17].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [16] Oberthuer A, Berthold F, Warnat P, Hero B, Kahlert Y, Spitz R, et al. Customized oligonucleotide microarray gene expression-based classification of neuroblastoma patients outperforms current clinical risk stratification. J Clin Oncol. 2006;24:5070–8. source
Not checked in the free run.
- cites [17] Ohira M, Oba S, Nakamura Y, Isogai E, Kaneko S, Nakagawa A, et al. Expression profiling using a tumor-specific cDNA microarray predicts the prognosis of intermediate risk neuroblastomas. Cancer Cell. 2005;7:337–50. source
Not checked in the free run.
Classified as probably not a substantial claim.
PremisecitationPredictive models were highly accurate in risk assessment of current low- and intermediate-risk neuroblastoma patients.
“However, while predictive models were highly accurate in risk assessment of current low- and intermediate-risk patients [18], the prediction of high-risk patient outcome has remained challenging [18–20].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [18] Oberthuer A, Hero B, Berthold F, Juraeva D, Faldum A, Kahlert Y, et al. Prognostic impact of gene expression-based classification for neuroblastoma. J Clin Oncol. 2010;28:3506–15. source
Not checked in the free run.
Classified as probably not a substantial claim.
PremisecitationRNA-seq data have been used to develop gene expression-based predictive models in cancer research.
“RNA-seq data have also been used to develop gene expression-based predictive models in cancer research [13, 14].”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [13] Cancer Genome Atlas Research Network. Comprehensive molecular characterization of clear cell renal cell carcinoma. Nature. 2013;499:43–9. source
Not checked in the free run.
- cites [14] Volinia S, Croce CM. Prognostic microRNA/mRNA signature from the integrated analysis of patients with invasive breast cancer. Proc Natl Acad Sci U S A. 2013;110:7413–7. source
Not checked in the free run.
May combine two claims.
Interpretationown dataThe predictive information of RefSeq-annotated genes represented by standard microarrays is saturated, and predictive information of more complex transcriptomic data provided by RNA-seq is largely redundant.
“This finding may suggest that the predictive information of RefSeq-annotated genes represented by standard microarrays is saturated, and that predictive information of more complex transcriptomic data provided by RNA-seq is largely redundant.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Study design and conductnone givenNeuroblastoma samples were randomly divided into training and validation sets.
“Neuroblastoma samples were randomly divided into training and validation sets (Table 2).”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Study design and conductreasoningSix data analysis teams and various classification methods were involved to account for potential impact of data analysis teams and classifier choice on prediction results.
“We involved six data analysis teams and applied various classification methods to take into account that the proficiency of data analysis teams and the choice of the classifier may impact the prediction results.”
Quote in the paperSubstantial claim- Rests on an assumption: The proficiency of data analysis teams and the choice of the classifier may impact prediction results.
May combine two claims.
Interpretationown dataBoth the different platforms and the different analytical methods provide largely comparable results.
“Together, these results demonstrate that both the different platforms and the different analytical methods provide largely comparable results, indicating the validity of the methods used in our study.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Study design and conductnone givenModels with highest average performance metrics were selected and submitted for testing on adequate blinded validation sets for each endpoint.
“Models with highest average performance metrics were selected and submitted for testing on adequate blinded validation sets for each endpoint.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S4 (based).
Interpretationnone givenThe key advantage of RNA deep-sequencing resides in its ability to characterize transcriptomes at an unprecedented level of detail.
“The key advantage of RNA deep-sequencing, however, resides in its ability to characterize transcriptomes at an unprecedented level of detail, which may lead to new insights into the molecular mechanisms of disease, thereby providing starting points for the development of rational targeted therapeutic strategies.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
Study design and conductnone givenData analysts were asked to include a five-fold cross validation for 10 iterations to assess model performances in the training datasets.
“Data analysts were asked to include a five-fold cross validation for 10 iterations to assess model performances in the training datasets, but were otherwise free in their choice of modeling algorithms and statistical tests to generate and select suitable prediction models.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Study design and conductnone givenAll analyses are regarded as explorative.
“All analyses are regarded as explorative.”
Quote in the paperNot a substantial claim
The paper relies on this without support.
A comprehensive comparison of RNA-seq and microarray-based predictive models is lacking to date.
“A comprehensive comparison of RNA-seq and microarray-based predictive models, however, is lacking to date.”
The paper relies on this without support.
This is the first study evaluating predictive models using RNA-seq in comparison to microarrays.
“To our knowledge, we present the first study on the evaluation of predictive models using RNA-seq in comparison to microarrays, which may provide valuable information for designing future experiments on gene expression-based classifiers using high-throughput technologies.”
The paper relies on this without support.
Gene expression-based prediction of clinical endpoints depends on differing mRNA levels in clinically relevant disease subgroups.
“Since gene expression-based prediction of clinical endpoints depends on differing mRNA levels in clinically relevant disease subgroups, we evaluated how analysis of RNA-seq data compares to microarrays in determining differential gene expression in four major clinico-genetic subtypes of neuroblastoma (Additional file 1: Table S1).”
The paper relies on this without support.
The proficiency of data analysis teams and the choice of the classifier may impact prediction results.
“We involved six data analysis teams and applied various classification methods to take into account that the proficiency of data analysis teams and the choice of the classifier may impact the prediction results.”
The paper relies on this without support.
RNA-seq data processing may influence prediction performances.
“We extracted gene expression profiles from raw RNA-seq data using three different processing pipelines for mapping and quantifying sequence reads to also take the potential influence of RNA-seq data processing on prediction performances into account (Fig. 2a):”
The paper relies on this without support.
The six endpoints cover a broad range of prediction difficulties.
“We generated 360 predictive models for six endpoints covering a broad range of prediction difficulties.”
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?
Most transcriptomic clock studies have been conducted on small sample sizes and not tested on independent cohorts, or used older microarray technology, which is less accurate and reproducible than modern RNA sequencing.
“Most studies have been conducted on small sample sizes and not tested on independent cohorts, or used older microarray technology, which is less accurate and reproducible than modern RNA sequencing85,86.”
From Measuring biological age using omics data (2022) · cites this paper as [86]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Sex counts in the prediction cohort conflict with patient characteristics
MinorTable 2's SEX endpoint totals 211 female and 287 male patients across the training and validation sets. Table 1 instead reports 205 female, 278 male, and 15 patients with sex unavailable, so the stated endpoint labels cannot describe the same 498 patients.
“| All patients (498) | SEX | 249 | 103 | 146 | 249 | 108 | 141 |”
“| Sex | | | | Male | 278 | 55.8 % | | Female | 205 | 41.2 % | | N.A. | 15 | 3.0 % |”
Cites
Trapnell, C., Williams, B. A., Pertea, G., Mortazavi, A., Kwan, G., van Baren, M. J., Salzberg, S. L., Wold, B. J., & Pachter, L. (2010) · Nature Biotechnology · doi:10.1038/nbt.1621 · Cited by 15,556 · see citing works
TopHat: discovering splice junctions with RNA-Seq
Trapnell, C., Pachter, L. S., & Salzberg, S. L. (2009) · Bioinformatics · doi:10.1093/bioinformatics/btp120 · Cited by 11,477 · see citing works
Littell, R. C. (1996) · Medical Entomology and Zoology · Cited by 10,345 · see citing works
Gene expression profiling predicts clinical outcome of breast cancer
van 't Veer, L. J., Dai, H., van de Vijver, M. J., He, Y. D., Hart, A. A. M., Mao, M., Peterse, H. L., van der Kooy, K., Marton, M. J., Witteveen, A. T., et al. (2002) · Nature · doi:10.1038/415530a · Cited by 9,536 · see citing works
Landscape of transcription in human cells
Djebali, S., Davis, C. A., Merkel, A., Dobin, A., Lassmann, T., Mortazavi, A., Tanzer, A., Lagarde, J., Lin, W., Schlesinger, F., et al. (2012) · Nature · doi:10.1038/nature11233 · Cited by 5,187 · see citing works
Self-citationGENCODE: The reference human genome annotation for The ENCODE Project
Harrow, J., Frankish, A., Gonzalez, J. M., Tapanari, E., Diekhans, M., Kokocinski, F., Aken, B. L., Barrell, D., Zadissa, A., Searle, S., et al. (2012) · Genome Research · doi:10.1101/gr.135350.111 · Cited by 4,585 · see citing works
Comprehensive molecular characterization of clear cell renal cell carcinoma
The Cancer Genome Atlas Research Network (2013) · Nature · doi:10.1038/nature12222 · Cited by 3,485 · see citing works
A census of human cancer genes
Futreal, P. A., Coin, L., Marshall, M., Down, T., Hubbard, T., Wooster, R., Rahman, N., & Stratton, M. R. (2004) · Nature Reviews Cancer · doi:10.1038/nrc1299 · Cited by 3,184 · see citing works
Prediction of central nervous system embryonal tumour outcome based on gene expression
Pomeroy, S. L., Tamayo, P., Gaasenbeek, M., Sturla, L. M., Angelo, M., McLaughlin, M. E., Kim, J. Y. H., Goumnerova, L. C., Black, P. M., Lau, C. C., et al. (2002) · Nature · doi:10.1038/415436a · Cited by 2,369 · see citing works
Brodeur, G. M., Pritchard, J., Berthold, F., Carlsen, N., Castel, V., Castelberry, R. P., Bernardi, B. D., Evans, A. E., Favrot, M., & Hedborg, F. (1993) · Journal of Clinical Oncology · doi:10.1200/jco.1993.11.8.1466 · Cited by 2,287 · see citing works
Self-citationRNA sequencing: advances, challenges and opportunities
Ozsolak, F., & Milos, P. M. (2011) · Nature Reviews Genetics · doi:10.1038/nrg2934 · Cited by 2,135 · cited 1× · see citing works
“RNA deep-sequencing (RNA-seq) provides a powerful tool to decipher global gene expression patterns far beyond the limitations of microarrays, including an unprecedented capability to discover novel genes, alternative transcript variants, chimeric transcripts, and expressed sequence variants as well as allele-specific…”
BackgroundMAQC Consortium, Shi, L., Shi, L., Reid, L. H., Jones, W. D., Shippy, R., Warrington, J. A., Baker, S. C., Collins, P. J., de Longueville, F., et al. (2006) · Nature Biotechnology · doi:10.1038/nbt1239 · Cited by 2,126 · see citing works
Self-citationMaris, J. M., Hogarty, M. D., Bagatell, R., & Cohn, S. L. (2007) · The Lancet · doi:10.1016/s0140-6736(07)60983-0 · Cited by 2,029 · see citing works
Gene-expression profiles predict survival of patients with lung adenocarcinoma
Beer, D. G., Kardia, S. L. R., Huang, C.-C., Giordano, T. J., Levin, A. M., Misek, D. E., Lin, L., Chen, G., Gharib, T. G., Thomas, D. G., et al. (2002) · Nature Medicine · doi:10.1038/nm733 · Cited by 2,025 · see citing works
Sultan, M., Schulz, M. H., Richard, H., Magen, A., Klingenhoff, A., Scherf, M., Seifert, M., Borodina, T., Soldatov, A., Parkhomchuk, D., et al. (2008) · Science · doi:10.1126/science.1160342 · Cited by 1,293 · see citing works
Prensner, J. R., Iyer, M. K., Balbin, O. A., Dhanasekaran, S. M., Cao, Q., Brenner, J. C., Laxman, B., Asangani, I. A., Grasso, C. S., Kominsky, H. D., et al. (2011) · Nature Biotechnology · doi:10.1038/nbt.1914 · Cited by 1,026 · see citing works
SEQC/MAQC-III Consortium (2014) · Nature Biotechnology · doi:10.1038/nbt.2957 · Cited by 983 · cited 1× · see citing works
“…powerful tool to decipher global gene expression patterns far beyond the limitations of microarrays, including an unprecedented capability to discover novel genes, alternative transcript variants, chimeric transcripts, and expressed sequence variants as well as allele-specific expression [8 – 12].”
BackgroundSelf-citationRefSeq: an update on mammalian reference sequences
Pruitt, K. D., Brown, G., Hiatt, S. M., Thibaud‐Nissen, F., Astashyn, A., Ermolaeva, O. D., Farrell, C. M., Hart, J. A., Landrum, M., McGarvey, K. M., et al. (2013) · Nucleic Acids Research · doi:10.1093/nar/gkt1114 · Cited by 936 · see citing works
MAQC Consortium (2010) · Nature Biotechnology · doi:10.1038/nbt.1665 · Cited by 889 · see citing works
Gene expression profiling in breast cancer: classification, prognostication, and prediction
Reis-Filho, J. S., & Pusztai, L. (2011) · The Lancet · doi:10.1016/s0140-6736(11)61539-0 · Cited by 729 · see citing works
AceView: a comprehensive cDNA-supported gene and transcripts annotation
Thierry-Mieg, D., & Thierry-Mieg, J. (2006) · Genome Biology · doi:10.1186/gb-2006-7-s1-s12 · Cited by 629 · see citing works
Self-citationGene expression profiling predicts clinical outcome of prostate cancer
Glinsky, G. V., Glinskii, A. B., Stephenson, A. J., Hoffman, R. M., & Gerald, W. L. (2004) · Journal of Clinical Investigation · doi:10.1172/jci20032 · Cited by 478 · see citing works
Rat toxicogenomic study reveals analytical consistency across microarray platforms
Guo, L., Lobenhofer, E. K., Wang, C., Shippy, R., Harris, S., Zhang, L., Mei, N., Chen, T., Herman, D., Goodsaid, F. M., et al. (2006) · Nature Biotechnology · doi:10.1038/nbt1238 · Cited by 406 · see citing works
Self-citationGm, B., Pritchard, J., Berthold, F., Nl, C., Castel, V., Castelberry, R. P., De Bernardi, B., Evans, A. E., Favrot, M., & Hedborg, F. (1994) · PubMed · Cited by 392 · see citing works
Self-citationRoadmap for Developing and Validating Therapeutically Relevant Genomic Classifiers
Simon, R. M. (2005) · Journal of Clinical Oncology · doi:10.1200/jco.2005.02.8712 · Cited by 382 · see citing works
Oberthuer, A., Berthold, F., Warnat, P., Hero, B., Kahlert, Y., Spitz, R., Ernestus, K., König, R., Haas, S., Eils, R., et al. (2006) · Journal of Clinical Oncology · doi:10.1200/jco.2006.06.1879 · Cited by 291 · see citing works
Self-citationAsgharzadeh, S., Pique-Regi, R., Sposto, R., Wang, H., Yang, Y., Shimada, H., Matthay, K., Buckley, J., Ortega, A., & Seeger, R. C. (2006) · JNCI: Journal of the National Cancer Institute · doi:10.1093/jnci/djj330 · Cited by 237 · see citing works
Self-citationGlas, A. M. (2004) · Blood · doi:10.1182/blood-2004-06-2298 · Cited by 226 · see citing works
Ferreira, P. G., Jares, P., Rico, D., Gómez-López, G., Martínez-Trillos, A., Villamor, N., Ecker, S., González-Pérez, A., Knowles, D. G., Monlong, J., et al. (2014) · Genome Research · doi:10.1101/gr.152132.112 · Cited by 214 · see citing works
Vermeulen, J., De Preter, K., Naranjo, A., Vercruysse, L., Van Roy, N., Hellemans, J., Swerts, K., Bravo, S., Scaruffi, P., Tonini, G. P., et al. (2009) · The Lancet Oncology · doi:10.1016/s1470-2045(09)70154-8 · Cited by 192 · see citing works
Self-citationOhira, M., Oba, S., Nakamura, Y., Isogai, E., Kaneko, S., Nakagawa, A., Hirata, T., Kubo, H., Goto, T., Yamada, S., et al. (2005) · Cancer Cell · doi:10.1016/j.ccr.2005.03.019 · Cited by 155 · see citing works
Volinia, S., & Croce, C. M. (2013) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1304977110 · Cited by 149 · see citing works
Prognostic Impact of Gene Expression–Based Classification for Neuroblastoma
Oberthuer, A., Hero, B., Berthold, F., Juraeva, D., Faldum, A., Kahlert, Y., Asgharzadeh, S., Seeger, R., Scaruffi, P., Tonini, G. P., et al. (2010) · Journal of Clinical Oncology · doi:10.1200/jco.2009.27.3367 · Cited by 135 · see citing works
Self-citationNishi, T., Lee, P. S. Y., Oka, K., Levin, V. A., Tanase, S., Morino, Y., & Saya, H. (1991) · PubMed · Cited by 118 · see citing works
Intron-centric estimation of alternative splicing from RNA-seq data
Pervouchine, D. D., Knowles, D. G., & Guigó, R. (2012) · Bioinformatics · doi:10.1093/bioinformatics/bts678 · Cited by 97 · see citing works
Oberthuer, A., Juraeva, D., Li, L., Kahlert, Y., Westermann, F., Eils, R., Berthold, F., Shi, L. M., Wolfinger, R. D., Fischer, M., et al. (2010) · The Pharmacogenomics Journal · doi:10.1038/tpj.2010.53 · Cited by 67 · see citing works
Self-citationLenos, K., Grawenda, A. M., Lodder, K., Kuijjer, M. L., Teunisse, A. F. A. S., Repapi, E., Grochola, L. F., Bartel, F., Hogendoorn, P. C. W., Wuerl, P., et al. (2012) · Cancer Research · doi:10.1158/0008-5472.can-12-0215 · Cited by 65 · see citing works
A Three-Gene Expression Signature Model for Risk Stratification of Patients with Neuroblastoma
García, I., Mayol, G., Ríos, J., Doménech, G., Cheung, N. V., Oberthuer, A., Fischer, M., Maris, J. M., Brodeur, G. M., Hero, B., et al. (2012) · Clinical Cancer Research · doi:10.1158/1078-0432.ccr-11-2483 · Cited by 63 · see citing works
Self-citationStricker, T. P., Morales La Madrid, A., Chlenski, A., Guerrero, L., Salwen, H. R., Gosiengfiao, Y., Perlman, E. J., Furman, W., Bahrami, A., Shohet, J. M., et al. (2014) · Molecular Oncology · doi:10.1016/j.molonc.2014.01.010 · Cited by 36 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Measuring biological age using omics data
Rutledge, J., Oh, H., & Wyss-Coray, T. (2022) · Nature Reviews Genetics · doi:10.1038/s41576-022-00511-7 · Cited by 570 · see citing works
Zhang, L., Lv, C., Jin, Y., Cheng, G., Fu, Y., Yuan, D., Tao, Y., Guo, Y., Ni, X., & Shi, T. (2018) · Frontiers in Genetics · doi:10.3389/fgene.2018.00477 · Cited by 203 · see citing works
Self-citationLightbody, G., Haberland, V., Browne, F., Taggart, L., Zheng, H., Parkes, E., & Blayney, J. K. (2019) · Briefings in Bioinformatics · doi:10.1093/bib/bby051 · Cited by 173 · see citing works
- Authors
- Zhang, W., Yu, Y., Hertwig, F., Thierry-Mieg, J., Zhang, W., Thierry-Mieg, D., Wang, J., Furlanello, C., Devanarayan, V., Cheng, J., Deng, Y., Hero, B., Hong, H., Jia, M., Li, L., Lin, S. M., Nikolsky, Y., Oberthuer, A., Qing, T., Su, Z., Volland, R., Wang, C., Wang, M. D., Ai, J., Albanese, D., Asgharzadeh, S., Avigad, S., Bao, W., Bessarabova, M., Brilliant, M. H., Brors, B., Chierici, M., Chu, T.-M., Zhang, J., Grundy, R. G., He, M. M., Hebbring, S., Kaufman, H. L., Lababidi, S., Lancashire, L. J., Li, Y., Lu, X. X., Luo, H., Ma, X., Ning, B., Noguera, R., Peifer, M., Phan, J. H., Roels, F., Rosswog, C., Shao, S., Shen, J., Theissen, J., Tonini, G. P., Vandesompele, J., Wu, P.-Y., Xiao, W., Xu, J., Xu, W., Xuan, J., Yang, Y., Ye, Z., Dong, Z., Zhang, K. K., Yin, Y., Zhao, C., Zheng, Y., Wolfinger, R. D., Shi, T., Malkas, L. H., Berthold, F., Wang, J., Tong, W., Shi, L., Peng, Z., & Fischer, M.
- Affiliations
- BGI Group (China) (CN)
- Fudan University (CN)
- University of Cologne (DE)
- Fondazione Bruno Kessler (IT)
- AbbVie (United States) (US)
- GlaxoSmithKline (United States) (US)
- Rush University Medical Center (US)
- National Center for Toxicological Research (US)
- SAS Institute (United States) (US)
- Marshfield Clinic Research Foundation (US)
- Thomson Reuters (United States) (US)
- Loma Linda University (US)
- Georgia Institute of Technology (US)
- Fondazione Edmund Mach (IT)
- Children's Hospital Los Angeles (US)
- Schneider Children's Medical Center (IL)
- German Cancer Research Center (DE)
- University of Nottingham (GB)
- Center for Biologics Evaluation and Research (US)
- University of Arkansas at Little Rock (US)
- Eli Lilly (United States) (US)
- Universitat de València (ES)
- University of Padua (IT)
- Ghent University (BE)
- Harvard University (US)
- Stanford University (US)
- University of North Dakota (US)
- East China Normal University (CN)
- City Of Hope National Medical Center (US)
- Corresponding author
- Shi, Leming
- Peng, Zhiyu
- Fischer, Matthias
- ORCID
- Yu, Ying
- Hertwig, Falk
- Zhang, Wenwei
- Wang, Jian
- Furlanello, Cesare
- Devanarayan, Viswanath
- Cheng, Jie
- Deng, Youping
- Hero, Barbara
- Hong, Huixiao
- Lin, Simon M
- Oberthuer, André
- Qing, Tao
- Su, Zhenqiang
- Wang, Charles
- Wang, May D.
- Albanese, Davide
- Asgharzadeh, Shahab
- Avigad, Smadar
- Bao, Wenjun
- Brilliant, Murray H.
- Brors, Benedikt
- Chierici, Marco
- Zhang, Jibin
- Grundy, Richard G.
- He, Min Max
- Hebbring, Scott
- Kaufman, Howard L.
- Lababidi, Samir
- Lancashire, Lee J.
- Li, Yan
- Luo, Heng
- Ma, Xiwen
- Ning, Baitang
- Noguera, Rosa
- Peifer, Martin
- Phan, John H.
- Rosswog, Carolina
- Shen, Jie
- Theissen, Jessica
- Tonini, Gian Paolo
- Vandesompele, Jo
- Wu, Po-Yen
- Xiao, Wenzhong
- Xu, Joshua
- Xu, Weihong
- Dong, Zirui
- Zhang, Ke K.
- Yin, Ye
- Zhao, Chen
- Zheng, Yuanting
- Wolfinger, Russell D.
- Shi, Tieliu
- Malkas, Linda H.
- Berthold, Frank
- Wang, Jun
- Tong, Weida
- Shi, Leming
- Peng, Zhiyu
- Fischer, Matthias
- Venue
- Genome Biology
- Published
- 2015-06-25
- Volume
- 16
- Issue
- 1
- Pages
- 133-133
- ISSN
- 1465-6906 (ISSN-L)
- PMC
- PMC4506430
- OpenAlex
- W2124140824
- Cited by
- 393
- References
- 39
- Citations by year
- 2026: 13 · 2025: 25 · 2024: 34 · 2023: 44 · 2022: 47 · 2021: 41 · 2020: 55 · 2019: 38 · 2018: 39 · 2017: 36 · 2016: 16 · 2015: 3 · 2012: 1
- FWCI
- 13.59
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- Genome biology (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- Kölner Universitäts PublikationsServer (Universität zu Köln) (publishedVersion)
- Research at the University of Copenhagen (University of Copenhagen) (publishedVersion, cc-by)
- Ghent University Academic Bibliography (Ghent University) (submittedVersion, other-oa)
- CINECA IRIS Institutional Research Information System (Fondazione Edmund Mach) (submittedVersion, cc-by-nc-sa)
- Digital Access to Scholarship at Harvard (DASH) (Harvard University) (submittedVersion, cc-by)
- Funders
- Georgia Cancer Coalition
- National Natural Science Foundation of China (863 Program, R01CA163256, 31471239, 2015AA020104)
- Bundesministerium für Bildung und Forschung (01GS0895)
- Deutsche Krebshilfe (110122)
- National Institutes of Health
- National High-tech Research and Development Program (2012AA02A201, 2015AA020104)
- National Institute on Alcohol Abuse and Alcoholism (R21AA020104)
- National Human Genome Research Institute (P01HG000205)
- National Cancer Institute (RC2CA148265, U54CA119338, R01CA163256)
- U.S. National Library of Medicine
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Gene expression and cancer classification
- Health Sciences › Medicine › Neurology › Neuroblastoma Research and Treatments
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Cancer Research › Breast Cancer Treatment Studies
- Keywords
- RNA-seq, microarray, neuroblastoma, gene expression profiling, prediction model, transcriptome, exon-exon junctions, biomarkers, model performance comparison
- MeSH terms
- Adolescent, Adult, Child, Child, Preschool, Female, Humans, Infant, Infant, Newborn, Male, Models, Genetic, Neuroblastoma, Tumor Cells, Cultured, Sequence Analysis, RNA, Oligonucleotide Array Sequence Analysis, Gene Expression Profiling, Endpoint Determination, Young Adult
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.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-10.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-10.
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
