idiv2: to do list

[rueuntal]

• smooth out sSBR by averaging it over at a minimum 20 runs. Currently the sSBR looks very jagged which largely reflects the particulars of the random walk that it took in sampling. By doing this several times a smoother and a less biased estimate of the expected S can be obtained although it will be slower.
• improve documentation for function rarefaction so it is clear what the different methods accomplish
• add subset.mob_in function that aids in pulling out a reduced number of rows of the mob_in object - this can be really useful when wanting to only work with certain treatment combinations
• acknowledge Marc Taylor for plotStacked function (https://github.com/marchtaylor/sinkr/blob/master/R/plotStacked.R) also thank Anne Chao and collegues for various functions --> this also requires us to copy and paste there functions for our purposes rather than maintain the dependancies on their package.
• stacked bar chart with pie charts
• implement spatial anaylsis when geographic coordinates are not available this changes interpretation from implicit and explicit spatial aggregation to just focusing on implicit spatial aggregation
• get rid of plot_rarefy and move that code into plot.mob_out which currently calls that function.
• implement the same log argument in plot.mob_out as in mob_out
• remove the function that does the 3d plot (and breaks when the code is executed on a server)
• change ScaleBy to ref_group
• Anova type output based on a ref group
• Agree on how to scale between # individuals and # plots
• Regression type
  • Regression analysis
  • null tests for regression?
    • test 1 - complete individual randomization
    • test 2 -
    • test 3
    • sensitivity analysis
• function at the beginning to check that rows (plots) between two data frames match
• define object class mob_out (print, plot) and mob_in (print, t-test, summary -> t-test table)
  • summary.mob_in should provide p-values rather than quantiles
• other functions
  • print
  • plot
    • rarefaction plots these will pair with the delta S plots
      • rarefaction plot legend sometimes mislabels the curves (high priority)
  • modify plot.mob_out to either plot delta curves or raw curves or both
    • It appears that plot.mob_out sometimes log transforms the x axis depending on mob_out$log_scale and sometimes does not. This needs to be made consistent across the figures. (high priority)
    • improve code in plot.mob_out
    • add number of samples to axis see old axis labels
    • nick requests that for the individual based rarefaction we show the entire curve across all samples and not just for the smallest number of individuals common across the groups.
  • trim delta plots to ignore n < 5 individuals due to the artifact that it must interested at 0.
  • box plots of univariate analyses
    • N vs trt
    • S rarefied vs trt
    • PIE rarefied vs trt
    • Beta PIE rarefied vs trt
    • S vs trt
  • summary
• cleanup and combine categorical case (anova) with continuous case (regression)
  • break out null algos back into separate functions Place null algos in separate functions per #29 #30
• test for composition? (*maybe later)
• complex exp design with more than one factor? (*maybe later)
• package up
• figure out what's going on with null test for N
• modify function rarefaction so it will accept a vector for sample based rarefaction
• add extrapolated S to the boxplot
• fix null test for aggregation
• add output of number of permutations to mobr object output from get_delta_stats
• occupancy data (low priority can only be applied to spatial rarefaction)
• track down warning message from rarefaction for coffee and cattle datasets: "1: In rarefaction(comm, "indiv", inds) :" Also the ant 2014 dataset is producing the following error: "In deltaS_N(comm_perm[plot_levels == x, ], plot_dens_level, ... :
  Extrapolating the rarefaction curve because the number of rescaled individuals is smaller than the inds argument "effort" larger than total number of samples
• document package
  • mobr.package documentation
• note in the documentation of the function rarefaction that sample-based rarefaction is not actually used anywhere in the mob methods.
• provide 3d plot option for headless machine (low priority) currently implemented with rgl
• repeat sensitivity analysis for the discrete case after fixing the warnings & also implement with larger number of iterations

[dmcglinn]

Folks keep adding things to the list of ToDo as you encounter them in the code

[dmcglinn]

@rueuntal and @FelixMay where are we on this list of to do's from last meeting?

[FelixMay]

@dmcglinn I feel responsible for the boxplots and for a vignette for mobr. However, with the latter I would like to wait until the normal function documentation is there. So let me know when you documented the functions. I will only be able to devote time, after November 9, which means after our iDiv conference

[rueuntal]

Hey @dmcglinn - sorry about the delay! My family are in town, so I'm a bit lagging behind. If I remember correctly, there are a few things that we have agreed not to include, at least not in this first release:

• analysis for composition

• complex experiment designs with multiple factors

• (continuous case - did we agree to postpone this as well?)

  The remaining big things on my mind at the moment are:

• make sure that the supporting functions (e.g., plot, summary, etc.) work and the outputs match what we have in mind;

• help documents for each function;

• bump up the number of iterations for the sensitivity analysis, to be done on iDiv's server.

Anything else?

[FelixMay]

Hey @dmcglinn, I will take care of the documentation for the mob_stats and the boxplots function. However, so far I only worked on the master branch. Should I add the new stuff to the 4cur branch? Does that mean I also have to fork this branch or can I chose when I push and pull to which branch this happens? Thanks!

[dmcglinn]

Hey Felix I'm going to merge 4cur with master today after Xiao has a chance to approve my PR. So you'll be able to just work with master shorty. Dan

…

On Nov 29, 2016, at 3:51 AM, Felix May ***@***.***> wrote: Hey @dmcglinn, I will take care of the documentation for the mob_stats and the boxplots function. However, so far I only worked on the master branch. Should I add the new stuff to the 4cur branch? Does that mean I also have to fork this branch or can I chose when I push and pull to which branch this happens? Thanks! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[FelixMay]

Hi Dan, Yeah cool. Then I am in again – and I have no more excuses of not contributing ;o) Felix Von: Dan McGlinn [mailto:[email protected]] Gesendet: Dienstag, 29. November 2016 12:20 An: MoBiodiv/mobr <[email protected]> Cc: May, Felix <[email protected]>; Mention <[email protected]> Betreff: Re: [MoBiodiv/mobr] idiv2: to do list (#21) Hey Felix I'm going to merge 4cur with master today after Xiao has a chance to approve my PR. So you'll be able to just work with master shorty. Dan

On Nov 29, 2016, at 3:51 AM, Felix May ***@***.******@***.***>> wrote: Hey @dmcglinn, I will take care of the documentation for the mob_stats and the boxplots function. However, so far I only worked on the master branch. Should I add the new stuff to the 4cur branch? Does that mean I also have to fork this branch or can I chose when I push and pull to which branch this happens? Thanks! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#21 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AO8i0XO4ErVX85_7Uifr1EZQ2CHmVrDmks5rDApHgaJpZM4JFAVD>.

[dmcglinn]

I just added additional items from the Felix to do list to a more general todo list for @rueuntal and myself