Used FindMutantVsWildtypeDEGenes-DESeq2.Rmd as an example of how to create my markdown.

Some things I changed and created:

• output file name
• the 'v' object which grabs genotype A & V and now the time chunk that I want
• coldata now includes temperature instead of time
• the design was changed to genotype + temperature with an interaction term with both of them
• the results now doesn't include a contrast but just a results name because the interaction term already made the comparison
• I added a new column to the result. It is called time and includes the time chunk data that was used in the analysis. This column should now show up in the output files
• I decided to change the time chunk and run the analysis each time point. In other words, I reran the analysis for each timepoint. This was due to needing fast results. I hope to go back and change the code to loop through each time and produce an output file that combines all of them.
• For now I made an output txt file for each time and genome but didn't leave it that way.
• I took all of the output txt files and combined them into excel file sheets. One for SL4 and one for SL5.
• Results file names:
  MvW-temp-SL4.xlsx
  MvW-temp-SL5.xlsx
• I then made sure the excel file were easy to read and the top row was frozen when scrolling down.

Next Step: Write and explain everything in the markdown.

Branch: https://bitbucket.org/mdavis4290/molly3-flavonoid-rnaseq/branch/IGBF-3460

File names:

• 72_F3H_PollenTube/FindTreatmentEffectAcrossGenotypes-DESeq2.Rmd
• 72_F3H_PollenTube/FindTreatmentEffectAcrossGenotypes-DESeq2.pdf
• 72_F3H_PollenTube/results/MvW-temp-SL4.xlsx
• 72_F3H_PollenTube/results/MvW-temp-SL5.xlsx

Notes: There is some more work to do obviously and improvements to make but the results are good to go and are ready to be shown.

Adding Molly's fork as a new remote to my local repo copy. Starting to look at her new work. Thanks [~molly].

For the next draft, I would do all four temperatures in the same Markdown.

Probably the easiest way to code it would be to transform the block of code inside the "for" loop starting on line 62 into a function that accepts a data frame with counts and a time point. Then, the time point parameter would then get used in line 70 to subset the columns.

The function would then return the DESeq object.

Then, you'd add additional code chunks, one per time point, where you run the function and save the returned DESeq object in a list. Because you would now have eight DESeq objects, you'd need to use a different key than in the current code. For example, instead of saving the results objects with assembly names as keys, you could use assembly names plus time point as keys.

I think it would be a good idea to do something like the above because currently, to re-run this code, we have to manually change it by editing the Markdown.

If this could be done, then the Makdown code can write out the file, instead of having to manually combine everything.

Also, another benefit is that the Markdown can then summarize the results by reporting the number of comparisons in each time point that met the target false discovery rate (Q) threshold.

It looks like none of the comparisons met the threshold. This is a negative result. That's OK. The most important thing here, and with all analyses, is to "show your work" so that whoever is reading this can believe that the analysis is correct and then move on to the next question.

Also, thanks for rebasing your branch

Branch: https://bitbucket.org/mdavis4290/molly3-flavonoid-rnaseq/branch/IGBF-3460b

[~aloraine]

Please submit PR.

PR : https://bitbucket.org/hotpollen/flavonoid-rnaseq/pull-requests/26

I have reviewed the new content and have change requests. Moving to Done!