Question: Warning in lfproc when using rlog or varianceStabilisingTransformation from DESeq2
0
gravatar for katja
4.1 years ago by
katja0
Sweden
katja0 wrote:

Hi,

during the analysis of sRNA sequencing data I encountered a warning that I do not understand.

From counts of sRNA sequences I built a DESeqDataSet (7 groups of sRNA sequences x 28 samples) and performed normalisation with rlog function.

rld <- rlog(dds, blind = TRUE)

After that I got a warning:

NOTE: fitType='parametric', but the dispersion trend was not well captured by the
  function: y = a/x + b, and a local regression fit was automatically substituted.
  specify fitType='local' or 'mean' to avoid this message next time.

Warning message:
In lfproc(x, y, weights = weights, cens = cens, base = base, geth = geth,  :
  Estimated rdf < 1.0; not estimating variance

When I used fitType = "mean", there was no warning, but it appeared when I chose fitType = "local":

Warning message:
In lfproc(x, y, weights = weights, cens = cens, base = base, geth = geth,  :
  Estimated rdf < 1.0; not estimating variance

The same thing also happend when I used a function varianceStabilisingTransformation. I also used blind = TRUE and tried different settings of fitType.

I read the vignette from DESeq2 package, documentation for rlog and varianceStabilisingTransformation, but could not find the answer, what the warning means. I also could not find the explanation of lfproc function or rdf. Did anyone encounter the same problem or does anyone know what it means? Could it be that there is not enough datapoints for the methods to work?

Thanks for your help,

Katja

ADD COMMENTlink modified 4.1 years ago by Michael Love23k • written 4.1 years ago by katja0
Answer: Warning in lfproc when using rlog or varianceStabilisingTransformation from DESe
0
gravatar for Michael Love
4.1 years ago by
Michael Love23k
United States
Michael Love23k wrote:
I'll take a look at writing a better warning. Yes, with only 7 rows, there are too few points to fit a smooth trend line of dispersion ~ mean. So the mean fit type is all I would use.
ADD COMMENTlink modified 4.1 years ago • written 4.1 years ago by Michael Love23k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 107 users visited in the last hour