RPKM for alternative splicing genes?

[baohua100]

When calculate a PRKM for a gene locus

R=109C/NL
where C is the number of mappable reads that fell onto the gene’s exons, N is the total number of mappable reads in the experiment, and L is the sum of the exons in base pairs.

for genes with alternative splicing, like a intron retain, how to determin the L ? is this intron count into the L ?

[juan]

L is the sum of bases for the gene, not the exon, right?

RPKM does not account for splicing, nor does it distinguish between transcript isoforms.

I would include the intron in the L.

[steven]

I would suggest to "project" or "collapse" all the exons onto the genomic sequence and work in genomic space. Then your C will be the number of mappable reads that fall in an exonic region (of one of the transcripts of your gene), and L the total exonic space of the gene (number of genomic positions that are found in a mature transcript, not the "sum of the exons"). Like this, overlapping parts of exons from different transcripts are not counted several times. Also, transcribed areas that are alternatively retained into splice products (like your "retained intron") will be part of L.

AFAIK, nothing has been published yet regarding relative quantification of transcript variants from alternative splicing. Does anyone have more info?