Question: DEXSeq on time series - how to set up models
0
gravatar for MedwayC
3.6 years ago by
MedwayC0
MedwayC0 wrote:

I have three different treatment times (0h (untreated), 3h and 6h), and three samples for each time point, 9 samples in total. I want to perform a single test to identify exons where the level of expression is dependent of the 'dose' of time; 0h<3h<6h OR 0h>3h>6h .

I have read the vignette, and know I have to set up the reducedModel and fullModel:

fullModel = ~ sample + exon + time:exon

reducedModel = ~ sample + exon

I am not sure I have this right.  Would someone be able to suggest these model arguments please? I have no additional technical or biological covariates I need to correct for.

ADD COMMENTlink modified 3.6 years ago by Alejandro Reyes1.7k • written 3.6 years ago by MedwayC0
Answer: DEXSeq on time series - how to set up models
0
gravatar for Alejandro Reyes
3.6 years ago by
Alejandro Reyes1.7k
Dana-Farber Cancer Institute, Boston, USA
Alejandro Reyes1.7k wrote:

Hi MedwayC,

The models seem correct. From the question you want to answer, it seems more appropriate to treat the time variable as a quantitative variable instead of a categorical (qualitative) one. You would just need to specify the time variable in your colData object as a numerical value and the models can remain the same.

As for the effect sizes (estimateExonFoldChanges), the current implementation of this function can't handle numerical variables, but  you could specify the categorical variables and maybe use as denominator the 0h time point. 

Alejandro

ADD COMMENTlink written 3.6 years ago by Alejandro Reyes1.7k

Hi again MedwayC,

To complete the answer above, I have just added support for estimating exon fold changes using numeric variables as predictors. The changes are implemented in the latest DEXSeq development version.

Alejandro

ADD REPLYlink written 3.6 years ago by Alejandro Reyes1.7k
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: 278 users visited in the last hour