Displaying RNA-Seq reads scaled in proportion to the depth of sequencing?

[mudpud]

I have run into a problem with displaying RNA-Seq reads from my own experiments along with RNA-Seq reads from previously published experiments from other groups.

My sequencing runs are within 30% of each other with respect to the total number of reads. Nevertheless, some of the samples I have analyzed with Bowtie/Tophat/Cufflinks from other groups have significantly more or less reads.

I usually use the Integrated Genomics Viewer with the COUNT feature to display tracks showing the quantity of reads mapping to a particular ORF. Nevertheless, with significantly more reads in some samples, one gets the false impression that the same gene in such samples is expressed at much higher levels. This happens solely due to a significantly higher sequencing depth. When the reads from such samples are adjusted for millions of mapped reads, then it is clear that the gene in the RNA-Seq sample with many more reads is actually expressed at a lower level than in my sample (lower FPKM values).

I was wondering if anybody has developed a way to scale reads, aka to scale track heights precisely in order to graph RNA-Seq tracks proportionately to their depth of sequencing?

Otherwise, the point I am trying to make looks false to a reader if I want to graph genome tracks with ORFs. I would have to forego that and switch to bar graphs showing gene expression FPKMs.

Any thoughts are welcome!

[alec]

To scale by total reads, go to View -> Preferences -> Tracks and select Normalize Coverage Data. If you want some other normalization method (like upper-quartile), you would have to scale each track manually.