search

BridgesLab / CushingAcromegalyStudy

The source code for the cushing and acromegaly studies, currently ongoing
Other
1 stars 2 forks source link

Compare GSEA results to GOseq Results #12

Closed davebridges closed 10 years ago

davebridges commented 10 years ago

Need to generate code for running GSEA externally. Compare the GO categories from this with the GOseq categories.

davebridges commented 10 years ago

Quynh can you redo the GSEA analysis with the HTseq/DESeq2 data. The input data should now be

data/processed/htseq_Annotated DESeq2 Results - Acromegaly.csv data/processed/htseq_Annotated DESeq2 Results - Cushing.csv

Can you run this against KEGG, GO and TRANSFAC. Thanks

qtran1 commented 10 years ago

Working on it now!

On Jan 19, 2014, at 7:49 AM, Dave Bridges wrote:

Quynh can you redo the GSEA analysis with the HTseq/DESeq2 data. The input data should now be

data/processed/htseq_Annotated DESeq2 Results - Acromegaly.csv data/processed/htseq_Annotated DESeq2 Results - Cushing.csv

Can you run this against KEGG, GO and TRANSFAC. Thanks

— Reply to this email directly or view it on GitHubhttps://github.com/BridgesLab/CushingAcromegalyStudy/issues/12#issuecomment-32708647.

davebridges commented 10 years ago

Can you remove .html from the gitignore file and push the GSEA results from the DEseq2 branch back up

qtran1 commented 10 years ago

Done!

On Jan 26, 2014, at 9:25 AM, Dave Bridges wrote:

Can you remove .html from the gitignore file and push the GSEA results from the DEseq2 branch back up

— Reply to this email directly or view it on GitHubhttps://github.com/BridgesLab/CushingAcromegalyStudy/issues/12#issuecomment-33319252.

davebridges commented 10 years ago

I dont see any of the image (png files) in the DEseq2 branch of the cushing study. Can up upload those to that branch?

davebridges commented 10 years ago

Actually, rather than that, i noticed that the input files are not the same between master (called data/processed/htseq_Annotated DESeq2 Results - Acromegaly.csv of master) and the ones you used (in data/processed/Annotated DESeq2 Results - Acromegaly.csv of DEseq2). I am pretty sure the one in master is correct. I think the easiest thing would be make a new branch out of master, copy over GSEA_inputs_CushingAcromegaly.Rmd and rerun it.

The other issue is why are these file different and are we sure that the one in master is correct

qtran1 commented 10 years ago

I'm not sure why. But we need to clean the branches. Dave, I don't think it matters with GSEA because I used the htseq counts and then let GSEA ranks the list using t-statstics. GSEA required expression values. I used DE genes before, but we agreed to use the whole list. So, the analysis only changes when the counts change.

Now, the question is

  1. Should we use some kind of cutoff to create different gene lists with different types of analysis approaches?
  2. Or we could pre-rank our gene lists based on the p-values from the analysis and put it in GSEA.

On Feb 2, 2014, at 8:56 AM, Dave Bridges wrote:

Actually, rather than that, i noticed that the input files are not the same between master (called data/processed/htseq_Annotated DESeq2 Results - Acromegaly.csv of master) and the ones you used (in data/processed/Annotated DESeq2 Results - Acromegaly.csv of DEseq2). I am pretty sure the one in master is correct. I think the easiest thing would be make a new branch out of master, copy over GSEA_inputs_CushingAcromegaly.Rmd and rerun it.

The other issue is why are these file different and are we sure that the one in master is correct

— Reply to this email directly or view it on GitHubhttps://github.com/BridgesLab/CushingAcromegalyStudy/issues/12#issuecomment-33902202.

qtran1 commented 10 years ago

Dave, should we do gene_set or phenotype permutation? I think we should stick with phenotype. What do you think?

davebridges commented 10 years ago

I think we should stick with phenotype too, if thats not what you did already.

Also can you re-run the GO analysis separating out MF/BP/CC as separate folders. The main go analysis includes them all but i think it would be better to consider them differently.

qtran1 commented 10 years ago

I agree.

Sent from my iPhone

On Mar 9, 2014, at 8:54 AM, "Dave Bridges" notifications@github.com<mailto:notifications@github.com> wrote:

I think we should stick with phenotype too, if thats not what you did already.

Also can you re-run the GO analysis separating out MF/BP/CC as separate folders. The main go analysis includes them all but i think it would be better to consider them differently.

— Reply to this email directly or view it on GitHubhttps://github.com/BridgesLab/CushingAcromegalyStudy/issues/12#issuecomment-37127434.

qtran1 commented 10 years ago

I uploaded the BP, MF, CC GSEA results in bower/bridgeslab/Tran/mar09.

DiffRank: used the difference between 2 groups as the ranking metric Otherwise, t-statistics were used for ranking.