%0 Journal Article %1 Howard2010 %A Howard, Brian E %A Heber, Steffen %D 2010 %J BMC Bioinformatics %K Plant;Genes genetics;Arabidopsis methods;ComputerSimulation;Databases genetics;BaseSequence;Chi-SquareDistribution;ComputationalBiology RNA NucleicAcid;GeneExpressionRegulation Plant;LinearModels;Models Physiological Genetic;MolecularSequenceData;ProteinIsoforms analysis/genetics;SequenceAnalysis methods;Stress Algorithms;AlternativeSplicing %P S6 %R 10.1186/1471-2105-11-S3-S6 %T Towards reliable isoform quantification using RNA-SEQ data. %U http://dx.doi.org/10.1186/1471-2105-11-S3-S6 %V 11 Suppl 3 %X BACKGROUND: In eukaryotes, alternative splicing often generates multiple splice variants from a single gene. Here we explore the use of RNA sequencing (RNA-Seq) datasets to address the isoform quantification problem. Given a set of known splice variants, the goal is to estimate the relative abundance of the individual variants. METHODS: Our method employs a linear models framework to estimate the ratios of known isoforms in a sample. A key feature of our method is that it takes into account the non-uniformity of RNA-Seq read positions along the targeted transcripts. RESULTS: Preliminary tests indicate that the model performs well on both simulated and real data. In two publicly available RNA-Seq datasets, we identified several alternatively-spliced genes with switch-like, on/off expression properties, as well as a number of other genes that varied more subtly in isoform expression. In many cases, genes exhibiting differential expression of alternatively spliced transcripts were not differentially expressed at the gene level. CONCLUSIONS: Given that changes in isoform expression level frequently involve a continuum of isoform ratios, rather than all-or-nothing expression, and that they are often independent of general gene expression changes, we anticipate that our research will contribute to revealing a so far uninvestigated layer of the transcriptome. We believe that, in the future, researchers will prioritize genes for functional analysis based not only on observed changes in gene expression levels, but also on changes in alternative splicing.

@article{Howard2010, abstract = {BACKGROUND: In eukaryotes, alternative splicing often generates multiple splice variants from a single gene. Here we explore the use of RNA sequencing (RNA-Seq) datasets to address the isoform quantification problem. Given a set of known splice variants, the goal is to estimate the relative abundance of the individual variants. METHODS: Our method employs a linear models framework to estimate the ratios of known isoforms in a sample. A key feature of our method is that it takes into account the non-uniformity of RNA-Seq read positions along the targeted transcripts. RESULTS: Preliminary tests indicate that the model performs well on both simulated and real data. In two publicly available RNA-Seq datasets, we identified several alternatively-spliced genes with switch-like, on/off expression properties, as well as a number of other genes that varied more subtly in isoform expression. In many cases, genes exhibiting differential expression of alternatively spliced transcripts were not differentially expressed at the gene level. CONCLUSIONS: Given that changes in isoform expression level frequently involve a continuum of isoform ratios, rather than all-or-nothing expression, and that they are often independent of general gene expression changes, we anticipate that our research will contribute to revealing a so far uninvestigated layer of the transcriptome. We believe that, in the future, researchers will prioritize genes for functional analysis based not only on observed changes in gene expression levels, but also on changes in alternative splicing.}, added-at = {2010-12-31T02:55:47.000+0100}, author = {Howard, Brian E and Heber, Steffen}, biburl = {https://www.bibsonomy.org/bibtex/2c1717a62eb076ba908e2f57813dfe3cb/jabreftest}, doi = {10.1186/1471-2105-11-S3-S6}, file = {Howard2010.pdf:Howard2010.pdf:PDF}, institution = {Bioinformatics Research Center, North Carolina State University, Raleigh, 27606, USA. itsbehoward@hotmail.com}, interhash = {53e24799efe8bcc1964caf995b4e7f93}, intrahash = {c1717a62eb076ba908e2f57813dfe3cb}, journal = {BMC Bioinformatics}, keywords = {Plant;Genes genetics;Arabidopsis methods;ComputerSimulation;Databases genetics;BaseSequence;Chi-SquareDistribution;ComputationalBiology RNA NucleicAcid;GeneExpressionRegulation Plant;LinearModels;Models Physiological Genetic;MolecularSequenceData;ProteinIsoforms analysis/genetics;SequenceAnalysis methods;Stress Algorithms;AlternativeSplicing}, language = {eng}, medline-pst = {epublish}, note = {Seqmap is used as alignment tool which might be slow. Arabidopsis is used.. takes into account the non-uniformity of RNA-Seq read positions along the targeted transcripts. assumed that the set of splice variants is known; the goal is to estimate the relative expression levels of these isoforms in a mixture. Implementation The algorithm described above was implemented in Java, with matrix computations by the JAMA matrix library (available at http://math.nist.gov/javanumerics/jama/ webcite). Data analyses and simulations were also performed using the R statistical programming language (http://www.r-project.org/ webcite). Real RNA-Seq datasets For each dataset, reads were mapped to the transcriptome using the SOAP v2 alignment program [15]. TAIR 8 was used to define the tested gene models. Differential splicing We first used a chi-square test of subset counts to identify genes that were differentially spliced between the two conditions. “Multireads,�? or reads that map to more than one gene, are another important problem for accurate isoform quantification.}, pages = {S6}, pii = {1471-2105-11-S3-S6}, pmid = {20438653}, timestamp = {2010-12-31T02:55:47.000+0100}, title = {Towards reliable isoform quantification using RNA-SEQ data.}, url = {http://dx.doi.org/10.1186/1471-2105-11-S3-S6}, volume = {11 Suppl 3}, year = 2010 }