search

Degiacomi-Lab / molearn

protein conformational spaces meet machine learning
https://degiacomi.org/software/molearn/
GNU General Public License v3.0
37 stars 11 forks source link

[JOSS] Manuscript Comments #10

Closed JoaoRodrigues closed 11 months ago

JoaoRodrigues commented 1 year ago

Hello molearn team,

Please find some comments below regarding your manuscript submitted to JOSS.

Statement of Need

On lines 21-22, I'd clarify that MD simulations offer an approximation of atomistic dynamics of biomolecules and make a clear distinction to the data obtained from experiments. These are, after all, simulations. Training NNs on these dynamics is creating a model to reproduce the approximation, not "real" dynamics. You can mention it right at the end when you say that MD is not a "silver bullet".

From a purely formatting perspective, I'd add a line break before you introduce molearn, on line 40.

Package Description

I'd remove the nested parenthesis on line 45-46 for the biobox reference.

On line 60, it should read "structure quality". On the same line, "root mean square deviation" should be explicity written of what. I assume atomic coordinates.

--

Other than these small comments, the manuscript reads very nicely and concisely.

degiacom commented 1 year ago

Thank you for your feedback @JoaoRodrigues. Good points, I have applied the changes as advised.

degiacom commented 11 months ago

@JoaoRodrigues, just checking if you are happy with the changes (for the JOSS publication https://github.com/openjournals/joss-reviews/issues/5523). If so, happy to close this issue?