search

Murali-group / supervised-grns

Supervised inference of gene regulatory networks from single-cell RNA-sequencing data
GNU General Public License v3.0
0 stars 0 forks source link

Set up k-fold validation for any k and obtain results #2

Open tmmurali opened 3 years ago

tmmurali commented 3 years ago

At the moment, we generate results for 10-fold cross-validation. We should

  1. generalise to any k, passed in the command line,
  2. run for several iterations, also passed in the command-line, and
  3. plot the results across all iterations, e.g., in a box-plot.
tmmurali commented 3 years ago

@malabikasen please work on this issue with help from @adyprat. We can discuss here how to structure the code between functions and scripts.

adyprat commented 3 years ago

Added functionality to 1) Specify k in kfold-cv; 2) take in a random seed list or a number (nTimes in config.yaml) to generate nTimes different random seeds internally. The random IDs are stored in the output as part of parseoutput funciton. Fixed in 8dab7e5.

Plotting is still missing, however.

tmmurali commented 3 years ago

@aditya please make the minor changes to the field names in the output files.

@malabikasen please create a new issue for AUROC and AUPRC using the prroc R package.

We have decided to do plotting outside the codebase, e.g., in notebooks.