Data & Code: "Predicting missing links in global host-parasite networks"

Data &amp; Code from Farrell et al. "Predicting missing links in global host-parasite networks"<br><b>Scripts</b><br>Within the <i>scripts </i>folder are scripts to process the raw data and model results:<br>1. Download, clean, and merge host-parasite interaction databases with mammal supertree (<i>process_raw_data.R</i>)<br>2. Re-create raw data plots from the manuscript (<i>raw_data_plots.R)</i><br>3. Plot posterior interaction matrices, scaled trees, and pull out top predicted links (<i>model_summaries.R</i>)<br>4. Re-create diagnostic plots from the manuscript (<i>diagnostic_plots.R)</i><br>5. Functions used for data manipulation and visualization that are sourced by other scripts (<i>network_analysis.R</i>)<br>6. Investiage bias propagation via node degree product (<i>bias_investigation.R</i>)<br>7. Generate risk maps (<i>risk_maps.R)</i><br><br><br><b>Data</b><br><b><br></b>- <i>raw_data:</i> folder includes the data necessary to amalgamate the host-parasite interaction databases (via script '<i>process_raw_data.R</i>').<br>- <i>clean_data: </i>folder includes the full host-parasite interaction list <i>'hp_list' </i>in both .csv and .rds formats, as well as the binary interaction matrices for the full dataset and ones subset by parasite type (virus, bacteria, fungi, etc...), and model diagnostics ('<i>model_diagnostics.csv') </i>used in '<i>diagnostic_plots.R' .</i><br>- <i>model_results: </i>folder contains .rds files per model, which has the output interaction matrix from each simulation ('<i>P'</i>), the table of model diagnostics ('<i>TB</i>'), and the phylogeny scaling parameter (<i>'Eta'</i>), if applicable. Note that to save space the full cross-fold fit posteriors are omitted (these total ~ 4.5GB). Please contact MF if these are required. <br>-<i> literature_results:</i> folder contains a .csv version of the results of the literature search outlined in the Supplementary Information.<br>- <i>plots_tables: </i>folder contains .csv files for the top 100 'missing' links for each model, and a .csv for the top 1000 links from the full model run on the full dataset.<br><br><br><br>