differential expression (DE)

[anj_1]

Hi,

I am trying to compare results between isoform and gene level DE for a set of genes to figure out if any of the genes were affected by splicing. Any suggestions on how to do this?

[blancha]

I'm not sure I understand your question very clearly.
If you are trying to determine differential splicing between conditions, there are multiple programs that will attempt to do this, as a quick Google search will show.

Here are just a few of the programs the will do a differential splicing analysis.
Each program has its specific strengths and weaknesses.
MATS, Cuffdiff, DEXSeq, MISO, ...

[anj_1]

so what i meant is:

i have genes_exp.diff
isoform_exp.diff

Both generated from Cuffdiff. I want to compare results between isoform and gene level DE to identify any genes that may be affected by splicing.

How would I go about doing this?

[blancha]

If you want to analyze splicing with Cuffdiff, you would want to look at the file "splicing.diff".
Generally, the results of the splicing analysis by Cuffdiff are useless though.
Few, if any significant results, ...
I generally don't even bother looking at it, but I have seen some papers reporting the results of splicing.diff.

You're much better off analyzing splicing using a different software.
I really like MATS.
It's very simple to understand the underlying principle, based on comparing the counts for reads overlapping junctions.

For a specific junction, you can also visually determine differential splicing by generating Sashimi plots with IGV after loading the BAM files.
This is my also one of my favourite methods for determining differential splicing for specific junctions.

Be warned that researchers unfamiliar with RNA-Seq invariably will complain about the results of the differential splicing analysis. Detecting a differential splicing event is dependent on the sequencing depth at the splicing site, and how great the differences in splicing are. If the depth is low at the splicing site, or the differences in splicing are small, you will not be able to detect them. To detect differential splicing, using in my opinion the most reliable method, you need a significant difference between the conditions being compared in the reads spanning the junction.

[anj_1]

thank you Blancha!