Skip to content

Latest commit

 

History

History
75 lines (47 loc) · 3.35 KB

README.md

File metadata and controls

75 lines (47 loc) · 3.35 KB
Raw

Model adequacy and the macroevolution of angiosperm functional traits: Analysis

We developed this repository and structured our analysis with the goal of making our methods transparent and our results completely reproducible. Be warned though that doing this takes some time.

Dependencies:

First, install remake from github using devtools (remake is not yet on CRAN as it is under active development).

devtools::install_github("richfitz/remake")

We depend on quite a few packages for the analysis. To install any missing packages, run:

remake::install_missing_packages()

This step also requires devtools to install arbutus and sowsear from github (see remake_sources.yml)

The full list of required packages is in the packages: section of remake.yml, plus the xlsx package to build wright_2004.

Running the analysis

From within the analysis/ directory run remake::make(). This will take several hours, depending on your computer, and print out some information to what it is doing.

Or, in several steps:

  1. Download all the data and build into data sets
remake::make("data")

Note that one of the data sources (Kew) is continuously updated without a versioning control system and therefore the downloaded data may be different from what is presented in the paper. We will organise a cached verison of this data.

  1. Split the data into a large number of subsets and run the analysis on each (both ML and Bayesian)
remake::make("fits")

This step takes a while (about 5 CPU hours to run the ML analyses, and 13 CPU hours for the Bayesian analysis: so about 9 hours total over two processors). If you have a powerful computer, you will find that setting the global mc.cores option before running this helps (e.g. options(mc.cores=8) to use 8 cores). It will also produce close to 1GB of output, which would be a problem if you are very tight for disk space.

  1. Generate the report and figures
remake::make()

This will build an html report (model-adequacy.html) that includes all numbers referenced in the paper, plus all the figures (pdf versions in output/figs, plus summary tables (output/results-ml.csv and output/results-bayes.csv). Note that owing to the stochastic nature of the procedure, re-running the analyses will result in slightly different numbers from those reported in the paper.

To explore further, run:

e <- remake:::make_dependencies(remake::remake(), "model-adequacy.md")
remake::remake_attach(e)

and then any of the code in model-adequacy.R will be runnable.

The data directory.

Most of the data used in this analysis are downloaded from elsewhere, but a few are local:

  • errors.csv: Data value error fixing
  • names-tr.txt: Species name translations to fix spelling errors, etc
  • spermatophyta_synonyms_PLANTLIST.csv list of synonyms derived from the Plant List, by Jon Eastman.

Data downloaded from other sources:

  • kew.csv, kew/: Seed weight data from Kew; http://data.kew.org/sid
  • leda.csv (and leda.txt): SLA data from LEDA traitbase; Kleyer et al. 2008, Journal of Ecology 96:1266-1274.
  • wright_2004.csv (and wright_2004.xls): SLA and leaf nitrogen data from Wright et al. 2004, Nature 428:821-827.