Design for DESeq2
1
0
Entering edit mode
sally_b86 • 0
@sally_b86-13975
Last seen 2.9 years ago

Hi, 

Thank you for taking my question into consideration.

I have RNA-seq dataset, a time series experiment, 3 time points with 8 replicates each. I wanna check the variation in gene expression across time. But I want to take into consideration the surgeon effect (2 surgeons were performing the surgery on mice).

I will run LRT test but I'm not really sure of what the reduced model should be: reduced~surgeon or reduced~1, the variable of interest should be removed which is time. 

LRT deseq2 multiple factor design • 556 views
ADD COMMENT
0
Entering edit mode

Hello again Michael, 

I wanted to include batch effect into my design, in which the 32 samples are distributed among 13 pools of flow cells during  RNA sequencing. So i assigned each sample to the pool it was placed in from p1 to p13 but I got half the genes having padj <0.05 compared to the run without batch effect and in the resultsNames this is what I get I'm not really sure that this must be happening, p1 is not a control to do such computation!! 

resultsNames(dds)
 [1] "Intercept"               "surgeon_Lionel_vs_Bruno" "replicate"              
 [4] "batch_p10_vs_p1"         "batch_p11_vs_p1"         "batch_p12_vs_p1"        
 [7] "batch_p13_vs_p1"         "batch_p2_vs_p1"          "batch_p3_vs_p1"         
[10] "batch_p4_vs_p1"          "batch_p5_vs_p1"          "batch_p6_vs_p1"         
[13] "batch_p7_vs_p1"          "batch_p8_vs_p1"          "batch_p9_vs_p1"         
[16] "time_t1_vs_t0"           "time_t2_vs_t0"           "time_t3_vs_t0"  

Is there anything I'm missing? 

Thank you, 

Sally

ADD REPLY
0
Entering edit mode

Can you make a PCA plot where you color by the flow cell pools? I've found that the sequencing batch doesn't influence the measurements that much (meanwhile, library preparation batch has a great influence). If you see not much grouping by flow cell pool you could skip it, or if you are concerned about unwanted variation you could instead use a method like svaseq or RUVSeq to estimate say 1-3 factors of unwanted variation, and then include these in the design. See here:

http://www.bioconductor.org/help/workflows/rnaseqGene/#removing-hidden-batch-effects

ADD REPLY
0
Entering edit mode
@mikelove
Last seen 1 day ago
United States

If you want to find genes that change over time, but controlling for a surgeon effect it would be a full design of ~surgeon + time with a reduced design of ~surgeon.

ADD COMMENT
0
Entering edit mode

Thank you for your reply :) 

ADD REPLY

Login before adding your answer.

Traffic: 172 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6