add alpha diversity box plot at single rarefaction depth

[gregcaporaso]

A boxplot would probably be good here showing, on a per-body-site basis, other (in a box) with self as a single point next to the box. This might involve some custom code, but @jrrideout is the box-plot-master, so best to check with him on that (e.g., it might be possible to hack using the existing code by just providing each self value as a "single value distribution" which might then only plot the median of that "distribution").

[jairideout]

Here's what a matplotlib boxplot looks like with a single-value distribution (with a value of 2). This was generated with generate_box_plots() as follows:

>>> from qiime.pycogent_backports.distribution_plots import generate_box_plots
>>> fig = generate_box_plots([[1, 2, 3, 4], [2]])
>>> fig.savefig('foo.png')

[gregcaporaso]

OK - that's what I was guessing. It's a little bit of a hack, but I'd be OK going with this as a first pass.

[wasade]

What about axis labels?

On Dec 4, 2012, at 15:54, Jai Ram Rideout notifications@github.com wrote:

Here's https://dl.dropbox.com/u/43778879/foo.png what a matplotlib boxplot looks like with a single-value distribution (with a value of 2). This was generated with generate_box_plots() as follows:

from qiime.pycogent_backports.distribution_plots import generate_box_plots fig = generate_box_plots([[1, 2, 3, 4], [2]]) fig.savefig('foo.png')

— Reply to this email directly or view it on GitHubhttps://github.com/qiime/personal-microbiome-delivery/issues/7#issuecomment-11019590.

[gregcaporaso]

Yes, with labels of course (see help(generate_box_plots) for how to set those). I meant that a plot that looks like this will work for the first pass and we can improve from there as necessary.

[gregcaporaso]

We'll generate these plots from the collated alpha data (same input as to make_rarefaction_plots.py). To get that data for a single rarefaction depth, you can do the following:

from qiime.parse import parse_rarefaction
from qiime.compare_alpha_diversity import extract_rarefaction_scores_at_depth
extract_rarefaction_scores_at_depth(10,parse_rarefaction(open('alpha_div_collated/observed_species.txt','U')))

For the first-delivery milestone, we'll generate these data at 10k sequences/sample, but this value should be a parameter to the command line script.

[jairideout]

Looking at this now- wouldn't there be distributions for both self and other, since there are multiple iterations and samples at a particular body site? Or am I misunderstanding what you're wanting to be plotted?

From what I can tell from the description, we're needing a figure for each body site, with two boxplots per figure (one for self, one for other), with each distribution made up of the alpha diversity metric values. Is that right?

[gregcaporaso]

You're right - we will have a distribution for each (I realized that after this conversation and forgot to follow up). I'd actually like to have all of these in a single box plot so it's easy to tell which sites are most/least diverse relative to each other.

[jairideout]

Great, thanks for explaining. Will put all body sites in a single plot.