Question: DEXSeq without replicates?
0
7.8 years ago by
Duke210
Duke210 wrote:
Hi folks, I am testing the DEXSeq package with a public RNA-Seq data. Unfortunately this data set does not have replicates, only two set of data with two conditions. I tried DEXSeq but got error. I checked the estimateDispersion function but there is no similar option like in DESeq where we can use for non replicate data. Is there any way to overcome this disadvantage of the data and finish the DEXseq analysis for it, or it is simply a no-go? By the way, my sessionInfo() if that helps: > sessionInfo() R Under development (unstable) (2011-12-12 r57875) Platform: x86_64-unknown-linux-gnu (64-bit) locale: [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C [3] LC_TIME=en_US.UTF-8 LC_COLLATE=C [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8 [7] LC_PAPER=C LC_NAME=C [9] LC_ADDRESS=C LC_TELEPHONE=C [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] DEXSeq_1.1.3 Biobase_2.15.3 BiocGenerics_0.1.3 loaded via a namespace (and not attached): [1] hwriter_1.3 plyr_1.6 statmod_1.4.14 stringr_0.6 > Thanks in advance, D.
dexseq • 1.2k views
modified 7.8 years ago by Simon Anders3.6k • written 7.8 years ago by Duke210
0
7.8 years ago by
Simon Anders3.6k
Zentrum für Molekularbiologie, Universität Heidelberg
Simon Anders3.6k wrote:
Hi On 2011-12-22 21:41, Duke wrote: > I am testing the DEXSeq package with a public RNA-Seq data. > Unfortunately this data set does not have replicates, only two set of > data with two conditions. I tried DEXSeq but got error. I checked the > estimateDispersion function but there is no similar option like in DESeq > where we can use for non replicate data. Is there any way to overcome > this disadvantage of the data and finish the DEXseq analysis for it, or > it is simply a no-go? I am starting to regret that we ever offered the "blind" mode with DESeq. The existence of this feature seemed to have misled too many users into believing that it is possible to perform a sensible analysis of RNA-Seq data without replication. It was, however, always only meant as a tool to salvage what is left from a botched experiment, and most of the time this will not be much. The analysis that DEXSeq attempt has much higher demands on data quality so that offering such a function would be even less likely to be of use. Nevertheless, it is actually possible to inject manually a dispersion estimate: simply add a column "dispersion" to the feature data slot and populate it with some wild guess of what the dispersion might be, e.g. by writing "fData(ecs)$dispersion <- .1" to set a common dispersion of 0.1. Then, one can continue with the test even without replicates. Obviously, this is not advisable except for exploratory purposes as any value one might inject will be nearly impossible to justify. Simon ADD COMMENTlink written 7.8 years ago by Simon Anders3.6k On 12/22/11 4:18 PM, Simon Anders wrote: > Hi > > On 2011-12-22 21:41, Duke wrote: >> I am testing the DEXSeq package with a public RNA-Seq data. >> Unfortunately this data set does not have replicates, only two set of >> data with two conditions. I tried DEXSeq but got error. I checked the >> estimateDispersion function but there is no similar option like in DESeq >> where we can use for non replicate data. Is there any way to overcome >> this disadvantage of the data and finish the DEXseq analysis for it, or >> it is simply a no-go? > > I am starting to regret that we ever offered the "blind" mode with > DESeq. The existence of this feature seemed to have misled too many > users into believing that it is possible to perform a sensible > analysis of RNA-Seq data without replication. It was, however, always > only meant as a tool to salvage what is left from a botched > experiment, and most of the time this will not be much. > > The analysis that DEXSeq attempt has much higher demands on data > quality so that offering such a function would be even less likely to > be of use. > > Nevertheless, it is actually possible to inject manually a dispersion > estimate: simply add a column "dispersion" to the feature data slot > and populate it with some wild guess of what the dispersion might be, > e.g. by writing "fData(ecs)$dispersion <- .1" to set a common > dispersion of 0.1. Then, one can continue with the test even without > replicates. > > Obviously, this is not advisable except for exploratory purposes as > any value one might inject will be nearly impossible to justify. Thanks Simon. I will find another dataset to start with then. Bests, D.