[BioC] Chip-seq pipeline and cloud ami questions

[Christopher Howerton]

Hello,

I have used Bioconductor for several local analyses of microarray experiments, but I have just started to work on an analysis of a Chip-seq experiment and I am finding myself a bit lost. The basic experimental design is a 2 (sex) X 2 (treatment) and for each biological replicate I have: Input and two ChIP samples (H3K4me3 and O-GlcNAc, a post-translational modification). Our sequencing core has provided me with BED files with the aligned sequences, peak finding with HOMER and has compiled two very large tables for each ChIP mark with the counts of reads that fall within any genomic range that was deemed a 'peak.' Using the final tables I have been able to find differential peaks using edgeR (which by the way has great documentation), and I am confident that I understand the background statistics, etc.

Despite, this I feel like there is much more information to be gleaned from this dataset. Some of these questions are 1) how do the two chip marks interact? 2) is this interaction dependent upon treatment condition? 3) is there a broader perspective than just a simple per genomic range analysis? 4) how does the chromatin state affect gene expression (I have microarray data from these samples)?

I think part of my problem, is that I don't know how to translate what I want into the vernacular of bioinformatics, so am easily confused by the dizzying array of available packages. Finally, after trying a few tutorials (i.e. DiffBind) it became clear to me that I will need to use parallel or cloud computing, but I am not familiar with using any of these resources. Looking at the bioconductor cloud ami help page, it gets me to the point of loading the instance, etc, but it's not clear to me where the data will live. Therefore my questions to the group at large is: 1) Is there a suggested pipeline of packages that could achieve my above stated goals? 2) Is there a basic example of a computationally intensive package using the cloud ami (I'm looking for a short vignette so I can understand how to appropriately interact with the EC2 instance)?

-- 
Christopher Howerton, PhD
Postdoctoral Researcher
(215) 898-1368
University of Pennsylvania
201E Vet
3800 Spruce Street
Philadelphia, PA 19104-6046