Question: missing coeeficients in resultsNames(dds)
gravatar for Assa Yeroslaviz
8 months ago by
Assa Yeroslaviz1.4k
Assa Yeroslaviz1.4k wrote:

I have the following design matrix

> colData(ddsMat)
DataFrame with 8 rows and 9 columns
sample  names   day condition   differentiated  rep ReadSums    design  sizeFactor
<character> <character> <factor>    <factor>    <factor>    <integer>   <character> <factor>    <numeric>
D0_MAEAKO_High_1    1   D0_MAEAKO_High_1    D0  KO  high    1   71041314    D0_KO_high  0.80109874
D0_MAEAKO_High_2    2   D0_MAEAKO_High_2    D0  KO  high    2   62294732    D0_KO_high  0.688588086
D0_MAEAKO_Low_1 3   D0_MAEAKO_Low_1 D0  KO  low 1   52015191    D0_KO_low   0.703319202
D0_MAEAKO_Low_2 4   D0_MAEAKO_Low_2 D0  KO  low 2   61838744    D0_KO_low   0.789337824
D0_WT_High_1    5   D0_WT_High_1    D0  WT  high    1   102380876   D0_WT_high  1.351169557
D0_WT_High_2    6   D0_WT_High_2    D0  WT  high    2   83154046    D0_WT_high  1.094416169
D0_WT_Low_1 7   D0_WT_Low_1 D0  WT  low 1   126036956   D0_WT_low   1.594154982
D0_WT_Low_2 8   D0_WT_Low_2 D0  WT  low 2   120620158   D0_WT_low   1.502663581

To run the DGE analysis I created the DESeq2 object

ddsMat <- DESeqDataSetFromMatrix(countData = counts,
                             colData = coldata,
                             design = ~design)

But now when I check for the resultsNames I see only these:

[1] "Intercept"   "design_D0_KO_low_vs_D0_KO_high"  "design_D0_WT_high_vs_D0_KO_high" "design_D0_WT_low_vs_D0_KO_high"

I would like to analyze multiple possibilities, e.g. "KO-low vs. KO-high" (coeff=2) or "WT-high vs. KO-high" (coeff=3), but also such as "WT-low vs. WT-high" or "KO-low vs. WT-low".

IS there any reason why I don't see them in the list of resultsNames? Can they be coerced into the list? I would to use the lfcShrink command which can't use the contrast option as opposed to the results command.

any ideas?

thanks Assa

design deseq2 • 353 views
ADD COMMENTlink modified 8 months ago • written 8 months ago by Assa Yeroslaviz1.4k
gravatar for RamRS
8 months ago by
Baylor College of Medicine, Houston, TX
RamRS30k wrote:

I think you can always use contrast() to generate custom results. resultsNames are a few examples that DESeq2 generates - it is not an exhaustive list.

ADD COMMENTlink written 8 months ago by RamRS30k

This is not true.

when I run this:

> resMA <- lfcShrink(ddsMat, coef="design_D0_KO_low_vs_D0_KO_high", type="apeglm")
using 'apeglm' for LFC shrinkage. If used in published research, please cite:
    Zhu, A., Ibrahim, J.G., Love, M.I. (2018) Heavy-tailed prior distributions for
    sequence count data: removing the noise and preserving large differences.

it works.

But this command returns an error:

> resMA <- lfcShrink(ddsMat, contrast = c("design", "D0_KO_low", "D0_KO_high"), type="apeglm")
Error in lfcShrink(ddsMat, contrast = c("design", "D0_KO_low", "D0_KO_high"),  : 
  type='apeglm' shrinkage only for use with 'coef'

which surprisingly clear.

ADD REPLYlink modified 8 months ago • written 8 months ago by Assa Yeroslaviz1.4k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 727 users visited in the last hour