weird fpkm value

[bioliyezhang]

Hi, all:

when I checked my paired-end RNAseq data, I saw some weird result.

http://www.freeimagehosting.net/o9akt

basically two genome features have same coordinate(as highlighted in red), however, their fpkm value is different. I wonder anyone have some ideas why this occur?

By the way,
I am using Tophat v2.0.4, Cufflinks v2.0.2 and cummeRbund v2.0.0.

Thanks.

Best,
Liye

[blanco]

Hi,
I suppose there could be a couple of reasons for this. It could be two different transcripts from the same gene. Or it could be antisense transcription.

If you look at the column gene_short_name and look up those names you will indeed see that the second one is an antisense transcript. What I find a bit strange is when I look up DDX11L10 it is listed on chromosome 16, not chromosome 1. The gene on chromosome 1 at this locus is called DDX11L1.

[bioliyezhang]

Hi, Blanco:

Thanks for checking that out.
Yes, the annotation file I used have some potential problem(DDX11L1 gene have one annotation at chr1 and chr16 as well). But I think I know what is the reason.

I check the annotation for DDX11L1 of chr1, which spans 11869 to 13670.
And RP11-34P13.3 spans 29554 to 30039.
And Has-mir-1302-2 starts from 30366 to 30503.

Therefore, I think the locus represents a transcript contig defined by cufflinks, which may include multiple transcripts. And each can have an unique ID. Therefore, both XLOC0001 and XLOC_00002 is within contig chr1:11868-31109.

Best,
Liye