dib-lab / eel-pond

mRNAseq assembly, evaluation, and expression analysis
2 stars 2 forks source link

Thoughts on formatting 7-differential-expression.rst? #36

Open johnsolk opened 7 years ago

johnsolk commented 7 years ago

Looking at options for different formats:

Previous version of eel pond used RSEM, which basically allowed the user to type a command and voila, the expression analysis is done, output tables generated. Common analysis packages now, e.g. DESeq, DESeq2, edgeR, etc. are a bit more interactive in R with command running, examining output, then tweaking.

Curious to hear thoughts on formatting, with the idea in mind to containerize, since executing R commands in a bash shell won't work, but executing a single R script will. So, perhaps a combination of options 2,1 or 2,3 below?

  1. set up an R jupyter notebook, examples:

https://github.com/ljcohen/Porites_astreoites_larvae_salinity/blob/master/RNAseq/P_astreoides_larvae_salinity_exposure_RNAseq_edgeR_DE.ipynb

https://github.com/ljcohen/osmotic/blob/master/notebooks/F_chrysotus.ipynb

  1. single script that should just run and give output, similar to the behavior of the RSEM command

https://github.com/ljcohen/osmotic/blob/master/DE_analysis_by_species/R_scripts_DE_analysis/killifish_DE_combined.R

https://github.com/ljcohen/ECE221_final_project/blob/master/DESeq2/DESeq2_flies.R

  1. R commands in blocks of code on the .rst?

https://github.com/ngs-docs/msu_ngs2015/blob/master/hands-on.Rmd

https://angus.readthedocs.io/en/2015/_static/SLDC-code.html

standage commented 7 years ago

I think adding a notebook to the setup would be overload. A single script would be easiest, but also shield the students from the R internals (for better and for worse). I would lean toward having R command for them to copy and paste interactively, but I could be convinced otherwise with a compelling argument.

ctb commented 7 years ago

For the first run through, I'm -1 on requiring interactivity. What we've done in classes is give them a cookbook R script that does the right thing, and then shown them where and how to edit it for their own samples. Conveniently we already have such scripts :) -- see

https://github.com/ngs-docs/2016-aug-nonmodel-rnaseq/blob/master/quantification.rst

and

https://github.com/ngs-docs/2016-aug-nonmodel-rnaseq/blob/master/files/nema.salmon.R

standage commented 7 years ago

Fair enough :)

johnsolk commented 7 years ago

Thanks! See PR #37