Model Runs with Variable Parameter Estimation

[bklein7]

The model runs performed using the 5, "15"-gene networks #265 will be repeated with variable parameter estimation.

Overall, four more runs will be completed for each of the networks. The parameter estimation settings used for these runs will be as follows:

• fixb=0, fixP=0
• fixb=0, fixP=1
• fixb=1, fixP=0
• fixb=1, fixP=1

[kdahlquist]

This is going to be done as part of the testing associated with #323.

[kdahlquist]

This was done as part of #351.

[kdahlquist]

Re-assigning this to @bklein7 as a sub-task of #351. From the discussion at today's meeting where we looked at the L-curve comparisons, we noted that we pay a bigger price for fixing P, than fixing b and that our method for making initial guesses for P may not be so great.

To follow up on this, we asked @bklein7 to plot the initial guesses for production rate next to the optimized production rate when b is fixed. A bar chart is probably the most appropriate for this, but I'm open to other suggestions, like plotting the difference itself against either the initial guess or optimized production rate or some other measure like degradation rate.

I guess I'm asking if there's something special about fast or slow rates.

We hope that seeing the relationship will inform us how to make the initial guesses better (although there may not be a relationship).

[kdahlquist]

I wanted to add that we should compare the production rates we estimate to some from the literature. The Neymotin paper from which we got the degradation rates also has production rates, I think. Look here: https://github.com/kdahlquist/DahlquistLab/tree/master/data/degradation_rates