Do reported disturbances in weekly survey show up in unweighted unifrac over time?

[DDomogala3]

When looking at the disturbance list, do disturbances reported in this list correlate with the plots of unweighted unifrac over time that Gilberto put together? What we need to do is change the weeks in the disturbance list to map to weekssincestart column in the mapping file.

[jairideout]

@DDomogala3, I cannot access the second link (unifrac over time)- it says the document doesn't exist anymore. I've requested access to the first link via my Google account (I currently cannot access it).

[DDomogala3]

Ok, you should be able to look at both, sorry I didn't realize you needed access to view these things in google docs. -Dan

On Tue, Nov 27, 2012 at 10:52 AM, Jai Ram Rideout notifications@github.comwrote:

@DDomogala3 https://github.com/DDomogala3, I cannot access the second link (unifrac over time)- it says the document doesn't exist anymore. I've requested access to the first link via my Google account (I currently cannot access it).

— Reply to this email directly or view it on GitHubhttps://github.com/gregcaporaso/student-microbiome-project/issues/16#issuecomment-10768624.

[jairideout]

Thanks!

[DDomogala3]

Hey these simple text files with only the PIDs and weekssincestart might help.

[jairideout]

@DDomogala3 and I met today to discuss this, and I referred him to QIIME's make_distance_comparison_plots.py as this can generate plots very similar to the ones @floresg created. Tutorial can be found here:

http://qiime.org/svn_documentation/tutorials/creating_distance_comparison_plots.html

@DDomogala3 will also be getting me the necessary information to map his disturbance week numbers to @floresg's "weeks since start" column. I'll write some code to create 3 new columns in the official SMP mapping file that indicate whether a particular sample was taken during a disturbance event or not (binary "yes" or "no" for sickness, menstruation, and antibiotics).

[gregcaporaso]

Thanks!

[DDomogala3]

Hi so I've included the code that @gregcaporaso provided earlier for creating an antibiotic disturbance mapping file column.

[gregcaporaso]

Where is the code? That link doesn't point to the right place.

[DDomogala3]

Sorry, it is in the issue 16 folder. Is that where you would want me to put code? https://github.com/gregcaporaso/student-microbiome-project/tree/master/analysis-results/issue16

[gregcaporaso]

@DDomogala3 will also be getting me the necessary information to map his disturbance week numbers to @floresg's "weeks since start" column. I'll write some code to create 3 new columns in the official SMP mapping file that indicate whether a particular sample was taken during a disturbance event or not (binary "yes" or "no" for sickness, menstruation, and antibiotics).

@jrrideout, do you have what you need for this now?

[jairideout]

I think so- I'll be taking a look at this later today and will let you guys know if I need anything else.

[DDomogala3]

@jrrideout Thanks for getting the script ready so quickly! I just added the make_distance_comparisons.py results here.This compares weeks 0, 0.5, and 1 to every other week that students turned in samples using the unweighted unifrac metric. @gregcaporaso would you want me to add this to a wiki page?

[gregcaporaso]

This needs to be done on a per-individual basis or the between individual differences are going to swamp the within individual, across timepoint differences. We'll also want to run this only for individuals included in the timeseries. I'm not sure which issue this plot is in response to, but have you applied @jrrideout's new code to generate the per-individual box plots colored by whether the individual had each of the disturbances that we're currently listing?

[DDomogala3]

@gregcaporaso I have started generating the per individual box plots. I used this command for the first one. Let me know if this is incorrect: (make_distance_boxplots.py -d /home/ubuntu/data/college_studentmicrobiome/forehead_only/forehead_even10000_unweighted_unifrac_dm_tc_only.txt -o /home/ubuntu/data/college_studentmicrobiome/forehead_only/unweighted_boxplots_forehead_PID/forehead_even10000_unweighted_unifrac_box_plots_antibiotic -m /home/ubuntu/data/college_student__microbiome/StudentMicrobiomeProject_map.tsv -f "Antibiotic_disturbance,PersonalID" --sort --height 8 --suppress_all_between --save_raw_data --color_individual_within_by_field "Antibiotic_disturbance")

This takes more than an hour to run each command.. -Dan

[gregcaporaso]

Yes, that looks right.

[gregcaporaso]

The answer to this appears to be 'yes' for some body site/disturbance pairs (e.g., gut/abx (as expected), gut/menstration) based on analysis of "adjacent unifrac distances". I will summarize this data in a new wiki page shortly.

[gregcaporaso]

See example plots here.

[gregcaporaso]

Also see plots here.This is an example of one per-individual PCoA trace - we would have these for all individuals.

My goal with these is to present paired plots of PCoA and adjacent unifrac (weighted and unweighted) to get an idea of exactly when the disturbances occur (adjacent-unifrac, we'll see what this adds to PCoA), and how resilient the communities (PCoA). In a per-individual adjacent unifrac plot, we indicate disturbance types in different colored vertical bars. In a per individual PCoA w trace file, we indicate the disturbances with letters presenting the points where a disturbance has occurred.