[IsBeth]

I'm reading your paper now. I will post my MA plot and ask more as soon I've read it. Thank you for your reply and for the link of your paper! =)

[chris_s]

I'm new in DESeq2 and i'm having problems to create my dds.

I have a matrix with eXpress results (read counts) with 2 conditions (SP, EB) and 2 replicates/condition.

This is what i'm doing:

Cele_SPvsEB = read.csv (file.choose(), header=TRUE, row.names=1)
CeleDesign <- data.frame(
  row.names = colnames(Cele_SPvsEB),
  condition = factor(c("SP", "SP", "EB", "EB")))
dds <- DESeqDataSetFromMatrix(countData = Cele_SPvsEB,
                              colData = CeleDesign,
                              design = ~ condition)
dds

And i'm getting this error message:

Error in get(name, envir = asNamespace(pkg), inherits = FALSE) : 
  object '.setDummyField' not found

Any help?
Thanks.

Hi,

I am in a fairly similar situation. I have used eXpress to generate read counts and they specifically recommend using the effective counts output for differential expression analysis. However, when I try to create the DESeqDataSet from the count matrix I get an error saying that :

Error in DESeqDataSet(se, design = design, ignoreRank) : 
  some values in assay are not integers

I know that DESeq2 normally takes in raw read counts, which probably explains why it is looking for integers, but since eXpress advises to use effective read counts which are not integers (see below): "The estimated number of fragments generated from this target in the sequencing experiment, adjusted for fragment and length biases. In other words, this is the expected number of reads from the experiment if these biases did not exist. This is the value recommended for input to count-biased differential expression tools."

I am keen to use the effective reads as this is applies a correction to an important potential bias.

Here is a subset from my countTable:

          A1_effcounts A3_effcounts A10_effcounts
evgtrinCL81615Contig1_79790      2.583273     0.000000      0.000000
evgtrinCL11213Contig1_47858     14.839541     9.894940     19.304148
evgtrinCL34839Contig1_73815     18.764835     0.000000      0.000000
evgtrinCL90968Contig1_58906      0.000000     0.000000      6.883716
evgtrinCL83431Contig1_102122     1.358742     1.354169      1.150552
evgtrinCL34506Contig1_139871     4.719385     2.416909      0.000000

and my design file:

   treatment  time
A1_effcounts        SED   24h
A3_effcounts        SED Week2
A10_effcounts       CON     0
A11_effcounts       CON   24h
A13_effcounts       CON Week2
H1_effcounts        SED   24h
H3_effcounts        SED Week2
H10_effcounts       CON     0
H11_effcounts       CON   24h

So do you use the total counts, the unique counts or is there a work around to use the effective counts from the eXpress output (e.g. rounding them to the nearest integer)?

Thanks for the help!

Cheers,

Chris

[dpryan]

Unique counts. If you want to use the effective counts then put them through voom() and use limma.

[TomHarrop]

Changes in r-log transformation?

Hi,

Sorry to dredge up this thread again but I have a question about the r-log transformation in the current version of DESeq2 (1.6.1).

I have a set of 8 libraries. Two of them were produced from degraded RNA. Using the previous version(s) of DESeq2, when I made density plots of r-log expression values, the two libraries that were produced from "bad" RNA had distributions that were obviously different from the libraries made from "good" RNA. This was accompanied by clear separation between the "bad" libraries and their matching biological replicates on a PCA plot made from r-log expression values.

With the current version, in density plots produced with the exact same code, all the distributions look the same as each other (i.e. the "bad" libraries now look the same as the "good" ones). The differentiation on the PCA plot is still there, though.

I am wondering if this observation can be explained by changes in the r-log transformation in the most recent version(s) of DESeq2? Unfortunately I can't say exactly which version I was using before, but it would have been the current Bioconductor release around the start of September.

Please let me know if any more information is required.

Thanks a lot for reading!

Tom

[Michael Love]

The estimator for the amount of shrinkage changed (this and all changes are noted in NEWS). Note that rlog is just one of the possible transformations, which we have found useful in many situations, but certainly it can shrink too much or too little given the diversity of datasets.

I'd try out the varianceStabilizingTranformation, or even log2(counts(dds, normalized=TRUE) + pc) for varying pseudocount.

[raphael123]

Hi !

I update my DEseq2 package to 1.4.5. Now my replaceOutliersWithTrimmedMean function is not replacing anything ! I have a very clear outlier that was remove previously and now any way I call replaceOutliersWithTrimmedMean, my counts(ddsClean)["LINC00624",] and my counts(dds)["LINC00624",] are axactly the same (with a basemean around 8 and an outlier around 240 ...

I didn t change the way I call the cooks outlier replacement funciton ... Any idea ?

[TomHarrop]

The estimator for the amount of shrinkage changed (this and all changes are noted in NEWS). Note that rlog is just one of the possible transformations, which we have found useful in many situations, but certainly it can shrink too much or too little given the diversity of datasets.

I'd try out the varianceStabilizingTranformation, or even log2(counts(dds, normalized=TRUE) + pc) for varying pseudocount.

Hi Michael,

Forgot to thank you for your suggestions. Simple log-transformation with pseudocount reveals the differences in distribution of expression values in the degraded libraries.

Cheers,

Tom

[Michael Love]

hi Raphael,

In version 1.4, DESeq() replaces outliers for certain sample sizes as described in the help file ?DESeq, and that we therefore don't recommend using the replaceOutliers function manually anymore. Also see ?DESeq for description of how the original and replacement counts are stored. If you have further questions, maybe best to start a new thread, and post the code you are using. Also, you might as well update to the current Bioconductor release 3.0 / DESeq2 version 1.6.

[harlequin]

Hi,

Can someone briefly explain how the "filterThreshold" attribute -which as far as I can tell is a threshold for the mean of normalized counts- of the results object is calculated in DESEq2? I've read the function manual but it's not entirely clear to me. Is it picked by an optimization algorithm so that the number of genes with adjusted P-values < 0.1 is maximum? Thanks!

[Simon Anders]

Is it picked by an optimization algorithm so that the number of genes with adjusted P-values < 0.1 is maximum?

Yes, exactly so.

[zhuya5607]

plotDispEsts() doesn't work



I can successfully follow all the commands in DESeq2 manual (updated version on December 16, 2014) except plotDispEsts(dds).

plotDispEsts(dds) gives the error message:
no slot of name "fitInfo" for this object of class "DESeqDataSet"

the following were the basic steps i ran

countData <- read.table("count_table_123vs789.txt",header=TRUE,row.names=1)
countData <- as.matrix(countData)
condition <- factor(c(rep("mirA",3),rep("mirB",3)))

coldata <- data.frame(row.names=colnames(countData),condition)
dds <- DESeqDataSetFromMatrix(countData=countData, colData=coldata, design=~condition)
dds <- DESeq(dds)
res <- results(dds)
plotDispEsts(dds)

Everything ran smoothly but the last step plotDispEsts(dds).
Can anyone please help me?

[Michael Love]

You have an older version of the older package "DESeq" masking the DESeq2 version.

We fixed the dispatch of plotDispEsts for DESeq/DESeq2/DEXSeq a year ago by creating a generic, so that all these functions can work with all libraries loaded, but if you have an outdated version of DESeq, you still experience the problem.

biocLite()

and update all. You might need to update your version of R as well. What's your

sessionInfo()

[zhuya5607]

You have an older version of the older package "DESeq" masking the DESeq2 version.

We fixed the dispatch of plotDispEsts for DESeq/DESeq2/DEXSeq a year ago by creating a generic, so that all these functions can work with all libraries loaded, but if you have an outdated version of DESeq, you still experience the problem.

biocLite()

and update all. You might need to update your version of R as well. What's your

sessionInfo()

Thanks for your reply! You are probably right about this, I did have a old version of DESeq installed before. However, i used biocLite("DESeq") to update,
the same problem is still there. Is my R version too old?

Here is my sessionInfo()
R version 3.1.0 (2014-04-10)
Platform: x86_64-apple-darwin13.1.0 (64-bit)

locale:
[1] C

attached base packages:
[1] parallel stats graphics grDevices utils datasets methods base

other attached packages:
[1] DESeq2_1.4.5 RcppArmadillo_0.4.600.0 Rcpp_0.11.3 GenomicRanges_1.16.4 GenomeInfoDb_1.0.2
[6] IRanges_1.22.10 BiocGenerics_0.10.0

loaded via a namespace (and not attached):
[1] AnnotationDbi_1.26.1 Biobase_2.24.0 DBI_0.3.1 DESeq_1.16.0 RColorBrewer_1.1-2 RSQLite_1.0.0
[7] XML_3.98-1.1 XVector_0.4.0 annotate_1.42.1 genefilter_1.46.1 geneplotter_1.42.0 grid_3.1.0
[13] lattice_0.20-29 locfit_1.5-9.1 splines_3.1.0 stats4_3.1.0 survival_2.37-7 tools_3.1.0
[19] xtable_1.7-4

[Michael Love]

I don't see why the problem should exist if you have updated BiocGenerics, DESeq, DESeq2 and DEXSeq to the same version of Bioconductor.

I recommend users stay with the current release of Bioconductor, which would be one version ahead of the one you have, by using the current release of R.

But if you don't have time to upgrade, or are in the middle of an analysis and need to save the state, a workaround is to use double colon to specify the library:

DESeq2::plotDispEsts

[zhuya5607]

I don't see why the problem should exist if you have updated BiocGenerics, DESeq, DESeq2 and DEXSeq to the same version of Bioconductor.

I recommend users stay with the current release of Bioconductor, which would be one version ahead of the one you have, by using the current release of R.

But if you don't have time to upgrade, or are in the middle of an analysis and need to save the state, a workaround is to use double colon to specify the library:

DESeq2::plotDispEsts

OK, I updated R this time, install DESeq, DESeq2 and DEXSeq again using biocLite().
R version 3.1.2 (2014-10-31)
Platform: x86_64-apple-darwin13.4.0 (64-bit)

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] parallel stats graphics grDevices utils datasets methods base

other attached packages:
[1] DESeq2_1.4.5 RcppArmadillo_0.4.600.0 Rcpp_0.11.3 GenomicRanges_1.16.4 GenomeInfoDb_1.0.2
[6] IRanges_1.22.10 BiocGenerics_0.10.0 BiocInstaller_1.14.3

loaded via a namespace (and not attached):
[1] annotate_1.42.1 AnnotationDbi_1.26.1 Biobase_2.24.0 DBI_0.3.1 DESeq_1.16.0 genefilter_1.46.1
[7] geneplotter_1.42.0 grid_3.1.2 lattice_0.20-29 locfit_1.5-9.1 RColorBrewer_1.1-2 RSQLite_1.0.0
[13] splines_3.1.2 stats4_3.1.2 survival_2.37-7 tools_3.1.2 XML_3.98-1.1 xtable_1.7-4
[19] XVector_0.4.0


However, i noticed a warning message when i load DESeq2

> library(DESeq2)
Loading required package: GenomicRanges
Loading required package: BiocGenerics
Loading required package: parallel

Attaching package: ‘BiocGenerics’

The following object is masked _by_ ‘.GlobalEnv’:

plotDispEsts

The following objects are masked from ‘package parallel’:

clusterApply, clusterApplyLB, clusterCall, clusterEvalQ, clusterExport, clusterMap, parApply, parCapply,
parLapply, parLapplyLB, parRapply, parSapply, parSapplyLB

The following object is masked from ‘package:stats’:

xtabs

The following objects are masked from ‘package:base’:

anyDuplicated, append, as.data.frame, as.vector, cbind, colnames, do.call, duplicated, eval, evalq, Filter, Find,
get, intersect, is.unsorted, lapply, Map, mapply, match, mget, order, paste, pmax, pmax.int, pmin, pmin.int,
Position, rank, rbind, Reduce, rep.int, rownames, sapply, setdiff, sort, table, tapply, union, unique, unlist

Loading required package: IRanges
Loading required package: GenomeInfoDb
Loading required package: Rcpp
Loading required package: RcppArmadillo


OK, even I tried

DESeq2::plotDispEsts

, still not working
standardGeneric for "plotDispEsts" defined from package "BiocGenerics"

function (object, ...)
standardGeneric("plotDispEsts")
<environment: 0x7fa45c1e1148>
Methods may be defined for arguments: object
Use showMethods("plotDispEsts") for currently available ones.

> showMethods("plotDispEsts")
Function: plotDispEsts (package BiocGenerics)
object="DESeqDataSet"


########Problem solved now###
Because i was loading DESeq2 in the folder where old DESeq and DEXseq were installed, the old plotDispEsts always mask the new plotDispEsts object.
I just changed a new working directory and reload DESeq2, then problem solved. Thanks for you help, Michael!