rMATS - Question

[mahek jain]

Hello Eric and other group members, 

I hope this email finds you all well. I am trying to use rMATS to quantify alternative splicing events. I read through your paper - "rMATS: Robust and flexible detection of differential alternative splicing from replicate RNA-Seq data", it was very informative and insightful, thank you for the same. However, I have a question - 

1. Iam trying to use rMATS on unpaired data and each group has different number of samples; for example, group 1 - has 5 and group 2 has 13, in that case should the number of samples in both the groups should be of an equal number to be able to use rMATS? If no, how do I determine the number of samples that I have to include? The paper mentions that within group variability and read numbers are important factors to consider, but if there are any additional factors that I must consider in the case of unpaired unequal data, please let me know. 
2. What would be the best way to determine within group variability for RNA sequencing data?

I appreciate your help. 

Thank you so much in advance!

Best,

Mahek

[kutsc...@gmail.com]

Hi Mahek,

You can have any number of samples in the two groups. Since you have 5 in group 1 and 13 in group 2 you can use all 5 in --b1 and all 13 in --b2: https://github.com/Xinglab/rmats-turbo/tree/v4.3.0
> The number of replicates can differ between sample 1 and sample 2 (example: 3 BAMs in --b1 and 1 BAM in --b2)

For within group variability, the paper that you mentioned describes how the variation of inclusion level among replicates within a group is modeled as a logit-normal distribution. Based on that you could get a variance for each group for each event. You could also make a PCA plot to see how the individual samples group in the plot: https://github.com/Xinglab/rmats-turbo/issues/398#issuecomment-2391636164

Eric