Is it possible to distinguish alternative splicing events from INDELs?

[sulicon]

Hi there,

We have resequenced a set of mRNAs and found novel isoforms. Some of them have partial exons when compared reference transcripts and most the of partial exons correspond to non-canonical splicing sites. They may be caused by the indels on the chromosomes or splicing events. Is there any analysis I could perform to distinguish the one possibility from the other?

Of course, this problem could be solved if we had the DNA sequences...

Thanks a lot.

[jmerkin]

Do some genomic sequencing. If it's an indel, the genomic reads will also have it.

If it is only a handful of events, you could also do PCR, either with a primer crossing the putative indel or, if the indel is large, with primers surrounding it so you can differentiate the two on a gel.

[steven]

Hi,

I would first look at the differences between your "isoforms", and in particular i would plot the size distribution of the differences: exon skipping should produce a bump that follows the exonic size distribution, whereas indels of size 1 are very unlikely to be produced by alternative splicing. That should give you the general trend.

In any case, I would be cautious regarding the conclusion of the splice sites being canonical or not, because this is usually defined by the intronic sequence (especially the two first and last positions), which may be difficult -or even impossible- to infer from your cDNAs. For instance, a skipped exon would produce two isoforms with no possible way to conclude if the splice sites are consensus or not, and that is the most frequently observed case of alternative splicing event.

Hope it helps,
s.

[sulicon]

Thanks for your suggestion. We have also discussed the experiments that could be performed. But before that, it would be nice if there are some computational analysis could be applied.

Do some genomic sequencing. If it's an indel, the genomic reads will also have it.

If it is only a handful of events, you could also do PCR, either with a primer crossing the putative indel or, if the indel is large, with primers surrounding it so you can differentiate the two on a gel.

[sulicon]

Hi steven,

I would first look at the differences between your "isoforms", and in particular i would plot the size distribution of the differences: exon skipping should produce a bump that follows the exonic size distribution, whereas indels of size 1 are very unlikely to be produced by alternative splicing. That should give you the general trend.

Yes, it's reasonable to declare a small gap between the alignment of cDNA and genomic sequences is caused by Indels. However, it would be hard if a large chunk of nucleotides miss in the alignment. For example, a gap of 10kb has been observed between the 1st and 2nd exons in the alignment, which corresponding to intron 1 and 2, exon 2, 3' terminal part of exon 1 and 5' terminal part of exon 3 in a reference transcript. What can we do for such an alignment? I have no idea whether to treat it as large Indel (CNV deletion) or consider it as an alternative spliced isoform...

In any case, I would be cautious regarding the conclusion of the splice sites being canonical or not, because this is usually defined by the intronic sequence (especially the two first and last positions), which may be difficult -or even impossible- to infer from your cDNAs. For instance, a skipped exon would produce two isoforms with no possible way to conclude if the splice sites are consensus or not, and that is the most frequently observed case of alternative splicing event.

I agree it's sometimes hard to infer the gene structure from cDNA sequences. But I'm not understand your point in the example. Can't we get the gene structure of an isoform with a skipped exon by aligning its sequence to reference genome?

Thanks,