Understanding percent spliced in

Question

[alex.v.nesta]

I am using SUPPA2 to analyze alternative splicing between tumor and normal samples. I am not sure if I am correctly interpreting the results.

For example:

Samples: SampleA = Normal, Sample B= Tumor

Event: A single exon skipping event

Calculations: PSI is calculated separately for sample A and sample B = (Number of reads covering included exon) / (number of reads covering skipped exon + number of reads covering included exon)

Let's say total reads for the splice junction of both samples is 100. When calculating PSI for normal there are 80 reads supporting inclusion and 20 reads supporting skipping. The PSI would be 80%: (80)/(20+80)

PSI for tumor would be 20 reads supporting inclusion and 80 reads supporting skipping. The PSI would be 20%. (20)/(20+80)

So we have Tumor PSI: 20% and Normal PSI 80%.

delta PSI = A - B OR 20% - 80% = -60%

So the -60% means the exon is skipped more in tumor.

Here is my question. Does this also mean the exon is included more in normal? Is it included 60% more? Why/why not.

I think the answer is yes, but I really need the affirmation of people on the internet.

The alternative answer would be as follows - Since we are comparing splicing of A - B only in this situation, one cannot assume that B - A is also true?

Thanks for your help!

[Eric Lim]

Does this also mean the exon is included more in normal? Yes. The direction in which people describe the PSI difference can depend on the biological context. Sometimes it's easier to focus on skipping. Sometimes it's inclusion. Is it included 60% more? Yes, in PSI. Depending on my audience, sometimes I find it more intuitive to simply describe an 80/20 difference as a 4-fold difference.

[alex.v.nesta]

Thanks for the confirmation Eric!