Understanding Expression Analysis

[jawalsh2]

Hello,

I am working on analyzing differential expression across several samples.

My data consists of 10 RNAseq libraries from an F2 population.

I aligned my reads with novoalign, using the sorghum gene models as my reference (Sorbi1_GeneModels_Sbi1_4_nt.fasta).

After alignment, I got the raw read counts for uniquely aligned reads, and plugged them into DEGseq in R to call differentially expressed genes.

One of the outputs from DEGseq shows the amount of reads mapped to each gene model and another the total reads in the library. In some cases, I have 5-10 or more million reads more in some of the libraries compared to others. However, DEGseq also creates an output table with a "log2(Fold_change) normalized" category. Is this enough to normalize my libraries or do I need a normalization step before differential expression analysis?


I have posted the output graphs from my DEGseq run at:

About the same # of reads:

http://www.life.illinois.edu/ming/output52v30.html

More in 1 library:

http://www.life.illinois.edu/ming/output23v1.html

Another example:

http://www.life.illinois.edu/ming/outputlvh.html

I would greatly appreciate any comments and suggestions on my workflow. Also, if I need normalize what is the best/easiest way to do so? Is there a .bed file available for sorghum?

Thank you for your help! And please contact me if I left anything out of my description!

- James