Normalization, testing, and false discovery rate estimation for RNA-sequencing data

Li, Jun; Witten, Daniela; Johnstone, Iain M.; Tibshirani, Robert

doi:10.1093/biostatistics/kxr031

Cited by 314 publications

(341 citation statements)

References 35 publications

Supporting

Mentioning

338

Contrasting

Order By: Relevance

“…37,39 Moreover, several normalization methods for RNAseq assume that counting of the total, or some portion of the transcripts is equivalent across the samples. [40][41][42] However, the traditional normalization methods could overestimate the stability of the cellular systems. For example, the RNA/DNA ratio of human keratinocytes changed sixfold, depending on the cell-cycle phase, cell size, or the passages.…”

Section: From Microarrays To Sequencingmentioning

confidence: 99%

Human Wound-Healing Research: Issues and Perspectives for Studies Using Wide-Scale Analytic Platforms

Nuutila

Katayama

Vuola

et al. 2014

Advances in Wound Care

View full text Add to dashboard Cite

Section: From Microarrays To Sequencingmentioning

confidence: 99%

Human Wound-Healing Research: Issues and Perspectives for Studies Using Wide-Scale Analytic Platforms

Nuutila

Katayama

Vuola

et al. 2014

Advances in Wound Care

View full text Add to dashboard Cite

“…The permutation principle is a well-established method 23,[31][32][33] , originally introduced by Fisher. 34 First, the original score is calculated for each protein by the scoring method.…”

Section: Postprocessing: Permutation Framework and Adjustment Proceduresmentioning

confidence: 99%

Pre- and Post-Processing Workflow for Affinity Purification Mass Spectrometry Data

et al. 2014

View full text Add to dashboard Cite

The reliable detection of protein-protein interactions by affinity purification mass spectrometry (AP-MS) is crucial for the understanding of biological processes. Quantitative information can be used to separate truly interacting proteins from false positives by contrasting counts of proteins binding to specific baits with counts of negative controls.Several approaches have been proposed for computing scores for potential interaction proteins, e.g. the commonly used SAINT software. However, it remains a subjective decision where to set the cutoff score for candidate selection and, further, no precise control for the expected number of false positives is provided.In related fields, successful data analysis strongly relies on statistical pre-and postprocessing steps which, so far, have only played a minor role in AP-MS data analysis. We introduce a complete workflow, embedding either the scoring method SAINT or alternatively a two-stage-poisson model into a pre-and postprocessing framework. To this end, we investigate different normalization methods and apply a statistical filter adjusted to AP-MS data. Further, we propose permutation and adjustment procedures, which allow the replacement of scores by statistical p-values.

show abstract

“…Several R packages have been developed for statistical testing for DE using RNA-Seq data [38] . Those include edgeR [39] , DESeq [40] , DEGSeq [41] , baySeq [42] , BBSeq [43] , TSMP [44] , NBPSeq [45] and PoissonSeq [46] . Additionally, databases like SEQC (SEquencing Quality Control) have been established to assess the performance of the NGS technologies.…”

Section: Advantages and Challengesmentioning

confidence: 99%

“…Li et al take a different approach by seeking a power transformation of the over-dispersed data [46] . They first select a conservative set S of transcripts, where all the transcripts in S are believed to be not differentially expressed.…”

Section: Power Transformationmentioning

confidence: 99%

“…In fact, the requirement that θ is the same for all transcripts is too restrictive, so the data is divided into multiple groups with θ estimated for each group. Their power transformation approach is implemented in the R package PoissonSeq [46] . In addition to modeling the read count data, researchers must normalize the RNA-Seq data to account for systematic variation.…”

Section: Power Transformationmentioning

confidence: 99%

See 1 more Smart Citation

Statistical modeling for differential transcriptome analysis using RNA-Seq technology

Miecznikowski

Liu

Ren

2012

JST

View full text Add to dashboard Cite

RNA-Seq is a recently developed technology for transcriptome profiling. Numerous advantages of RNA-Seq suggest that it will be the platform of choice for genome-wide expression studies. RNA-Seq generates large volumes of data which require statistical methods for data processing and accurate inference. This article reviews the RNA-Seq technologies followed by a detailed discussion of current statistical methods for normalization and differential expression analysis. Key wordsRNA-Seq, Normalization, Biological counts, Differential expression Transcriptome profilingOver 99.9% of genome sequences are the same in all humans [1,2] , yet individuals show great distinction from each other.In a multicellular organism nearly all cells contain the same genome, but they develop into different tissues. A major source for many of these variations is the different gene expression patterns [3] . In control of gene expression, the transcriptome is the complete set of ribonucleic acid (RNA) transcripts, including messenger RNA (mRNA), ribosomal RNA (rRNA), transfer RNA (tRNA), and other non-coding RNA in a given cell type. It is the connection between genes and phenotype. The constitution of the transcriptome for an organism varies at different developmental stages or under different physiological conditions. As a quantitatively cataloged transcriptome provides us information on the underlying genetic mechanisms, the transcriptome is studied in relation to many diseases, e.g., cancers. One of the main goals in a cancer transcriptome study is quantifying the changes in expression levels of all the transcripts in tumor cells. In this manuscript, we discuss the cutting edge methods for quantifying the transcriptome and how the resulting data is used to determine significantly differentially expressed transcripts. MicroarraysMicroarrays have been the primary technology for quantitative transcriptome analysis since the mid-1990s [4] , and they have discovered many results in cancer research [5][6][7][8] . Although expression studies by microarrays have been very successful in the last decade, there are at least three intrinsic limitations to this hybridization-based technology. First, the background noise in microarrays is large due to cross-hybridization of closely related genes. Second, the microarray signal often reaches a limit of detection or saturation, therefore microarrays have a limited dynamic range (a few hundredfold) [9] .

show abstract

Normalization, testing, and false discovery rate estimation for RNA-sequencing data

Cited by 314 publications

References 35 publications

Human Wound-Healing Research: Issues and Perspectives for Studies Using Wide-Scale Analytic Platforms

Human Wound-Healing Research: Issues and Perspectives for Studies Using Wide-Scale Analytic Platforms

Pre- and Post-Processing Workflow for Affinity Purification Mass Spectrometry Data

Statistical modeling for differential transcriptome analysis using RNA-Seq technology

Contact Info

Product

Resources

About