search

SystemsGenetics / KINC

Knowledge Independent Network Construction
MIT License
11 stars 4 forks source link

Example data #136

Closed spficklin closed 4 years ago

spficklin commented 4 years ago

@bentsherman . This PR contains a new directory named example. It has a quite small GEM in it 60x1167. But the genes in it were selected using GSForge (thanks @biggstd) to be those most indicative of the experimental conditions. The example data runs very fast and generates some nice condition-specific networks.

I've also updated the documentation to use this example data file. That means new users can actually follow along using this same data by cut-and-pasting the commands from the documentation. I've added some Bash scripts to the example directory as well that contain all of the commands from the tutorial. So you can run the entire network construction just by calling, for example, kinc-gmm-run.sh or kinc-traditional-run.sh. I've also got some visualization scripts as well for the GMM networks.

Hopefully this will help not just our end users but also us do debugging and testing. Perhaps we should also include a final network and use that with Travis CI to make sure we don't break KINC. As long as we get the same network out from this sample data odds are KINC is still working.

spficklin commented 4 years ago

Also, this PR is based off of the current master branch....

spficklin commented 4 years ago

One more thing. If you use the 126-corpower-slowness branch you can generate a proper network to which you can compare for trying to resolve the issues currently in the master branch.

bentsherman commented 4 years ago

@spficklin This is awesome we've needed this for a while. I'll go ahead and merge and then I'll be able to test my changes to corrpower and cond-test.

If you could add a traditional and gmm network at some point that would be great. I believe KINC is deterministic in it's output because it uses the same random seed every time. But if you give me some outputs I can set up a basic sanity test that we could eventually use for a Travis CI script.