Hi,
I have RNA-Seq data for a case vs. control experiment involving D. melanogaster (n=6: 3 cases vs. 3 controls).
The sequencing depth is high – approx. 20 million reads per sample – but coverage is comparable across replicates (the fold-change of mean sequencing depth between conditions is 1.1).
When I map to a reference genome and compare expression between conditions, it notice a systematic bias that is different for lowly expressed genes compared to highly expressed genes (see plot below).
This is not an artefact that is attributable to a particular normalization method, as I see the same trend when creating MA plots with DESeq, DESeq2, or with cufflinks-derived FPKM values (below).
Does anyone have any idea what could be causing this?
Thanks
I have RNA-Seq data for a case vs. control experiment involving D. melanogaster (n=6: 3 cases vs. 3 controls).
The sequencing depth is high – approx. 20 million reads per sample – but coverage is comparable across replicates (the fold-change of mean sequencing depth between conditions is 1.1).
When I map to a reference genome and compare expression between conditions, it notice a systematic bias that is different for lowly expressed genes compared to highly expressed genes (see plot below).
This is not an artefact that is attributable to a particular normalization method, as I see the same trend when creating MA plots with DESeq, DESeq2, or with cufflinks-derived FPKM values (below).
Does anyone have any idea what could be causing this?
Thanks