How can we estimate total percentage of alternative splicing in a specie, for example when we day percentage of AS in Arabidopsis is around 61 % and in human it is around 95 %, how we estimate this percentage?
Is the number of multi-transcript genes reflects AS percentage ?
Depending on the paper you're referencing those are either the percentage of genes with more than one annotated transcript/isoform, or the empirically identified number of genes with multiple isoforms. It's worth remembering that these percents are always floors, just because a specific gene is not alternatively spliced in a specific study doesn't mean it never undergoes AS.
I think it may help as you try to characterize specific events.
For example, some programs (like MISO or MATS) will calculate values between 0 and 1 for frequencies of splicing between two isoforms. However, other programs (like JunctionSeq or DEXSeq) will look at junction or exon differences independently (looking more at abundance than frequencies). If you have IsoSeq PacBio data, you may be able to get whole transcript annotations, but I still don't think that is quantitative (and you'll probably want to focus on the most abundant transcripts, and be aware that you may be enriching for longer transcripts, which could cause over-estimates of those transcripts compared to what you can independently validate).
I don't want to say one strategy will be universally better than the other, but I would recommend testing multiple freely available programs for your project (and eventually try to characterize some specific results).
I think that may be hard to define, in that it may differ considerably with the different methods that you test.
I see that this may be frustrating in terms of your original question, but I was trying to suggest alternatives to thinking about a "total AS" level (such as trying to find splicing events that may be most relevant for the phenotype / comparison you are trying to study). I should also probably mention those suggestions were for "differential splicing" between conditions.
My understanding is that there tend to be 1 or 2 dominant isoforms for a gene in a given condition (although I'm having some difficulty finding the Genome Biology paper that I am thinking about, with a quick Google search). I think that also some some effect asking a general question about splicing.
However, if it helps for the general discussion, I would probably place less emphasis on splicing events that occur within a small fraction of reads (and I think it sometimes helps to check the alignment, because sometimes the difference in splicing frequency reported by a program may be larger that what you would directly calculate from the read proportions).
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
I doubt one can come up with a stable answer for this, since we may never fully know all possible isoforms that are generated over the lifetime of an organism.
You may find this paper interesting (human only).
Those numbers come from using the number of genes as the denominator. If a gene has at least one exon (or intron) that is alternatively spliced (or not spliced), that gene is often considered alternatively spliced. If we're using exon (or intron) as the denominator, the percentage of alternative splicing will obviously be much lower.