GitHub - bleinwand/ACRONYM: ACRONYM: Augmented degree corrected, Community Reticulated Organized Network Yielding Model · GitHub
Skip to content You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session. Dismiss alert Branches Tags Open more actions menu Latest commit History 4 Commits 4 Commits Folders and files Name Name Last commit message Last commit date Congress Congress MainSim MainSim MouseRetina MouseRetina Random1200 Random1200 Subnetworks Subnetworks README.md README.md RunEverything.R RunEverything.R dependencies.R dependencies.R Repository files navigation ACRONYM ACRONYM: Augmented degree corrected, Community Reticulated Organized Network Yielding Model This is the code used to for the paper "ACRONYM: Augmented degree corrected, Community Reticulated Organized Network Yielding Model" There are 2 data examples and 3 types of simulations.
Each of these are in their own folder. Datasets are not included but information about how to retrieve these datasets are included below. Installing R library dependencies are included in the file "dependencies.R" "RunEverything.R" runs all the code, but sequentially, so it will take a long time (like a week). As it is, without the datasets, the simulations will run, but the Congress and Mouse analyses won't. Each of the 5 analyses can be run in parallel and parts of each can be run in parallel as well.
If chained together more intentionally using the instructions below, and given enough clusters, it can all be done in under a day. Data: Congressional Twitter Dataset Runs ACRONYM on the full dataset and creates 100 partially hidden networks and runs ACRONYM on them, too. Data availability: The .zip file can be retrieved from here https://zenodo.org/records/8253486 We use the file congress_network_data.json This is read in and formatted at the beginning /Congress/CongressParallelInit.R Running our code: - Download congress_network_data.json to /Congress - Change workspace to /Congress - Run CongressParallelInit.R This creates common networks and does community detection - Once that is complete, you can run all of CongressParallelU.R CongressParallelV.R CongressParallelEig.R In parallel, as they are estimating the same networks using different community detection approaches - When those 3 files have run, use CongressPostAnalysis2.R to view results Mouse Retina Dataset Runs ACRONYM on the full dataset and creates 30 partially hidden networks and runs ACRONYM on them, too.
Data availability: The mouse retina.db file can be retrieved from here github.com/ericmjonas/circuitdata/tree/master/mouseretina We use the file mouseretina.db This is read in and formatted at the beginning /MouseRetina/MouseInit.R Running our code: - Download mouseretina.db to /MouseRetina - Change workspace to /MouseRetina - Run MouseInit.R This creates common networks and does community detection - Once that is complete, you can run all of MouseU.R MouseV.R MouseEig.R In parallel, as they are estimating the same networks using different community detection approaches - When those 3 files have run, use MousePostAnalysis2.R to view results Simulations: Main Simulation.
Section 6 in the paper, with more detail in the appendix Creates one 1200x1200 network, and 30 partially hidden networks from this network. Runs ACRONYM on all of them. Running our code: - Change workspace to /MainSim - Run MainSimInit.R - This creates the networks we will be modeling, and does community detection - Once that is complete, you can run all of MainSimU.R MainSimV.R MainSimEig.R In parallel, as they are estimating the same networks using different community detection approaches - When those 3 files have run, use MainSimPostAnalysis.R to view results Networks with random parameters.
See the appendix Creates 30 1200x1200 networks with random parameters that kind of mimic what we see in the main simulation. Runs ACRONYM on all of them. Running our code: - Change workspace to /Random1200 - Run Redo_RandomAcronym30NestedFull2.R This creates the networks and runs ACRONYM using Eigenvector community detection - Once that is complete, you can run Redo_SingularValuesTest.R Redo_SingularValuesTestV.R In parallel, as they are estimating the same networks using left and right-singular vectors for community detection - When files have run, use postEstimationAnalysisforRandom1200networks.R to view results Subnetworks.
Simulation to remove the effect of community detection Creates 240 200x200 subnetworks generated from 3 different states, and using 4 different H-functions 20 replicates of each state/H-function combination Runs ACRONYM on all subnetworks Also runs ACRONYM with only the ``Normal" H-function available, rather than choosing the H-function based on the data Also runs ACRONYM with the estimate of sigma set to 0 (Redo_AcronymParallel200sigma0.R) Also creates a new set of 240 200x200 subnetworks where the Psi values are drawn from Beta distributions, and runs ACRONYM on those networks (AcronymParallel200Beta.R) Running our code: - Change workspace to /Subnetworks - Run Redo_AcronymParallel200.R - Can run AcronymParallel200Beta.R at the same time - Once Redo_AcronymParallel200.R is complete, run Redo_AcronymParallel200sigma0.R - When all files have run, use postEstimationAnalysisfor200Subnetwork.R to view results About ACRONYM: Augmented degree corrected, Community Reticulated Organized Network Yielding Model Resources Readme Activity Stars 0 stars Watchers 1 watching Forks 0 forks Report repository Releases Packages Contributors Languages You can’t perform that action at this time.