How does the alternative splicing recognition work?
GrailEXP currently looks for a very specific case of alternative splicing. It looks for fully determined EST evidence that indicates the insertion or deletion of an exon. In other words, if one gene model contains exons A, B, and C, and another contains only exons A and C, then the program identifies this case as an alternative splice. Why doesn’t the program identify other possible cases of alternative splicing? The biggest problem is distinguishing between an UNSPLICED product that has slipped into the EST database, an error in the genomic alignment, and a genuine case of alternative splicing. I am skeptical of many reports of alternative splicing, because they do not take this into account. Just because one rogue EST disagrees with the pack does not necessarily mean it is an alternative splice. Even if you have 3 or 4 that say one thing, how can you be certain it isn’t just a mistake in the alignment or in the ESTs? For that reason, GrailEXP currently recognizes only the most obviou
