DE analysis with only two groups

[yanav...@gmail.com]

Dear authors, first of all I would like to thank you for the development of this tool. 

Currently I am using your R script and I have a couple of questions:

My dataset is very simple and I just have two conditions to compare ("Control", "treatment") at one specific time point.  For each condition I have three biological replicates. 

1) In your R script, you are using, glmQL, glm and limma functions to estimate the DE genes. According to edgeR authors, this is the best approach if you have a complex experimental design but the classical analysis should be used in a Pairwise comparisons between two groups. What do you think about that? Should I use your approach or the classical one?

2) If I can use the more complex approach. How would you recommend me to define the "set contrast groups"?

Thanks in advance

Ana

[wenbin guo]

Dear Ana,

(1) What does your "Classical analysis"? Do you mean e.g. student t-test and anova analysis? 

Our 3D RNA-seq App also fit to simple pair-wise comparison (only two conditions). In our 3D pipeline, we have various steps to provide more accurate results, e.g. we have data pre-processing step to estimate mean-variance trend of read counts, remove low expressed genes and transcripts, estimate batch effects of biological replicates and data normalisation. And after analysis, we use the statistics of adjusted p-value, log2 fold change and delta percent spliced to generate significant results. Please read our pre-print in bioRxiv for method details: https://www.biorxiv.org/content/10.1101/656686v1. I don't think the classical t-test or anova method can directly incorporate the above complex steps. 

(2) In complex approach, you still use Control vs Treatment as contrast group in your dataset. You can take a look of our 3D App user manual: https://github.com/wyguo/ThreeDRNAseq/blob/master/vignettes/user_manuals/3D_RNA-seq_App_manual.md

Best regards,

Wenbin

[yanav...@gmail.com]

 (1) Thank you for your answer. I just want if this approach fit to a simple pair-wise comparison.

(2) I know that your web app use a simple example as welll. But I want to run my data locally using your R script. So my question was how should I define this groups on the R script. I just want to confirm that I am not doing something wrong. 

Thanks in advance for your quick response. 

Ana

[wenbin guo]

Hi Ana,

You only need to set contrast group as e.g. contrast = "Control-Treatment" (Control and Treatment are the labels in your metadata table of conditions) in the corresponding place.

Best,

Wenbin