Search
Question: DEXSeq exon usage values in result table
0
gravatar for biominer
14 months ago by
biominer0
European Union
biominer0 wrote:

In the HTML output of the DEXSeqHTML() function I have figures for both expression as well as exon usage of individual exons.

I have also produced a result table using estimateExonFoldChanges() and then DEXSeqResults().

The numbers in the html report graphics and the numbers in the table output are not identical. This let me doubt what the numbers in the result table for logFC and in the "conditionA" and "conditionB" columns actually are (the headers don't tell). Strange enough also in the html output the respective values in the result-tab don't fit the plotted values in the "splicing" tab nor the "expression" tab.

Are those the exon usage values and their fold change? Why are the numbers different from the heights in the html-figures? If so, how can I get the respective expression values as well?

 

I am interested to filter my result table on differences in exon usage as well as p-value.

 

 

ADD COMMENTlink modified 14 months ago by Alejandro Reyes1.4k • written 14 months ago by biominer0
0
gravatar for Alejandro Reyes
14 months ago by
Alejandro Reyes1.4k
Germany
Alejandro Reyes1.4k wrote:

Hi @biominer, 

The values plotted when splicing=TRUE are the exon usage values and should be the same present in the columns of your DEXSeqResults object. The difference is just the scale in which these values are represented in the plotDEXSeq function (variance stabilizing transformation). You can see that from the axis labels that are not in a linear scale. The fold changes are estimated between the exon usage values. 

If you are interested in filtering based on effect sizes I would go for the fold change column directly.

Alejandro

ADD COMMENTlink written 14 months ago by Alejandro Reyes1.4k

Thanks a lot for your immediate reply! Effect size is exactly what I am aiming for, so I filter for p-value and fold change.

I found I might also have to filter for a minimal read count ... or would you say that a significant p-value should already account for sufficient read-coverage?

ADD REPLYlink written 14 months ago by biominer0
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 2.2.0
Traffic: 250 users visited in the last hour