Arabidopsis transcriptome size?

[thurisaz]

Hi,

I'm trying to run saet to correct miscalled reads on our RNA-seq data, but I'm not sure what to use for the refLength ("expected length of the assembled sequence"). Does anybody know what size would be appropriate for the Arabidopsis transcriptome?

I downloaded the TAIR10 exons, 5' UTR & 3' UTR files and ran them through "grep -v '^>' | wc -m" to come up with a figure around 77 x 10^6 bases, which seems a bit higher (I was expecting around 50-60 x 10⁶ based on ~25,000 genes x 2000bp per gene).

Any thoughts or experience with this? Is there a reference I've missed which gives a reliable number?

[usad]

Hi,

I reckon you used TAIR10_exon_20101028?
it seems that identical exons from multiple splice variants are repeated in this file.

e.g.
>AT1G01020.1|exon-2 | 274-321 | chr1:8417-8464 REVERSE LENGTH=48
GGAAATTGCAAGGAAGTAGCAGATGAGTACATCGAGTGTGAACGCATG

>AT1G01020.2|exon-2 | 274-321 | chr1:8417-8464 REVERSE LENGTH=48
GGAAATTGCAAGGAAGTAGCAGATGAGTACATCGAGTGTGAACGCATG

which will increase your number. You could filter this based on the "representative" gene list also provided by TAIR.
I am unfortunately not familiar with saet.

Cheers,
björn

[thurisaz]

Thanks for pointing that out! Filtering with the representative gene list from TAIR brings the total length down to about 60 x 10⁶ bp, so I guess I'll use that. Hopefully saet isn't too sensitive to this parameter and a +/- 5x10⁶ bp estimate will be good enough...