Differential APA genes

[thomasthtc]

Dear Dr. Ha,

I would like to know how did you process the output file for differential APA genes? I'm slightly confused by your paper so I wish to know which method did you use? Is there any codes or guides you can provide for such analysis? I really wanted to use QAPA in my publication, but I'm not sure how to proceed with getting differential APA genes. Would you mind suggesting a way to do that?

Thanks,
Thomas

[zhanglab2008]

I have the same request. Can the authors provide script to process the output file...
That will help tons!
Many thanks

[quaidmorris]

Here's the relevant section from the methods, do you have questions that it doesn't answer?

DEXSeq [41] was used to compare 3′ UTR isoform expression changes between ESC and mature neurons. As per the method’s procedure, 3′ UTR isoforms were collapsed and segmented into adjacent bins demarcated by each isoform’s boundaries. In particular, we denote the 5′-most bin in the 3′ UTR as the proximal bin, which is associated with the “common UTR regions” (cUTR) — the region common to proximal and distal isoforms. We denote the remaining bin(s) located 3′ to the proximal bin as distal bin(s), which are associated with “alternative UTR regions” (aUTRs) originating from one or more distal isoforms. We defined a bin to be significantly differentially expressed if it had a |log2 fold change| > 0.5 and FDR < 0.1. For the latter, the same FDR was used as by the DEXSeq authors. In the case of multiple distal 3′ UTRs, we required a significant change for at least one of the distal bins.

[MustafaElshani]

Is there a possibility if these Rscripts for data visualisation can be deposited in github.

In addition this section on your paper, it does make sense what you have done I just can't seem to translate it to R script

we excluded gene pairs whose distal 3′ UTRs had 3′ ends that were within 500 nt of each other on the genome. We also excluded genes with aUTR lengths of less than 100 nt to reduce potentially noisy estimates due to small differences in length between proximal and distal 3′ UTR sequences. We defined the change in proximal poly(A) site usage (∆PPAU) as the difference between the median PPAU of ESC group (DIV −8 and −4) replicates and the median PPAU of the neuron group (DIV 7, 16, 21, and 27) replicates.

Any help would very much be appreciated

Mustafa

[iamnicogomez]

Hi Drs. Ha and Morris,

I too am interested in comparing PAU between groups in a more sophisticated way. Did you simply use the PAU_results.txt as a kind of gene counts matrix as an input for DEXseq? I'm fairly comfortable with RNAseq analysis yet I will say, I too was having trouble understanding the DEXseq part of the methods section. If you could share you R script, that would be greatly appreciated.

Thank you!
Nico