Add new notebooks

[lilyminium]

Fixes #99
Fixes #45
Related to #105

These are largely notebooks I made earlier during GSOD and updated a little. However, the diffusion map notebook is new.

Links (forthcoming from RTD):

New

• Diffusion map
• Hydrogen bond analysis
• Density analysis
• Hole remix
• ParmEd

Slightly updated

• PCA (description)

Add to #112

[orbeckst]

Sorry, only managed to look at 3 notebooks so far:

hole2 notebook

• I would move some of the general HOLE description for the PDB section to the front, then say that you can either run it like the hole binary (PDB) or on a trajectory/Universe, just to make clearer why you'd do one or the other
• MDAnalysis.analysis.hole.HOLE -> MDAnalysis.analysis.hole2.HOLE
• **tmp*/pdb_name.pdb**: : not properly formatted

Diffusion map

• "where the transitions between conformational states is not well-sampled" -> "where the transitions between conformational states are not well-sampled" (is -> are)

Just for the future: maybe we should use the AdK Transitions DIMS Dataset for the DiffusionMap analysis – instead of a single DIMS trajectory? Might make the output clearer.

Or even better, use a trajectory that shows repeated transitions between states – could be a small organic molecule or a fancy folding trajectory.

PCA

• Reference to a general paper on PCA (like in DiffusionMap), e.g. (pretty much randomly pulled from my master bibtex file):
  • [1] A. Amadei, A. B. Linssen, and H. J. Berendsen, “Essential dynamics of proteins,” Proteins, vol. 17, pp. 412–25, Dec 1993.
  • [2] I. T. Joliffe, Principal component analysis. Springer Series in Statistics, New York: Springer, 2 ed., 2002.
  • [3] F. Sittel, A. Jain, and G. Stock, “Principal component analysis of molecular dynamics: on the use of cartesian vs. internal coordinates,” J Chem Phys, vol. 141, p. 014111, Jul 2014.
  • [4] F. Sittel and G. Stock, “Perspective: Identification of collective variables and metastable states of protein dynamics,” The Journal of Chemical Physics, vol. 149, no. 15, p. 150901, 2018.
• What is the warning "For best results, your trajectory should be aligned on your atom group selection before you run the analysis. Setting align=True will not give correct results in the PCA." about? If align=True in PCA does not do the right thing then we should remove it, shouldn't we?
• The projection equation \mathbf{r}_i(t) = \mathbf{w}_i(t) \times \mathbf{u}_i + \mathbf{\overline{r}} uses "x" for the outer product. This seems prone to misinterpretation. Maybe write it out with explicit sums or use a different symbol (not the "cross product").

[lilyminium]

Thanks for the review, @orbeckst! I edited the things you mentioned, although I may leave migrating the DiffusionMap dataset for the future.

What is the warning "For best results, your trajectory should be aligned on your atom group selection before you run the analysis. Setting align=True will not give correct results in the PCA." about? If align=True in PCA does not do the right thing then we should remove it, shouldn't we?

This is from 0.20.1 -- it has been fixed for 1.0.0. I remove this warning in #120.

[orbeckst]

Continuation review... still not done (but wasted more time on 3D contour plotting than I'd care to admit...)

ParmEd

Very nice, didn't find anything.

The OpenMM part is a bonus and shows how MDA is useful in analysis.

Density

Very useful example with the PBC.

• nice to have: embedded image of the view1 and view2 of the system
• GROMACS trjconv argument in the table comes as a surprise. Add a sentence with a link to the gmx trjconv program before the table to note that these are operations commonly performed with gmx trjconv. Btw, what are the equivalent operations in VMD/pbc util or cpptraj? – @IAlibay ?
• link to docs for on-the-fly transformations
• Instead of calculating your own midpoints, use dens.density.midpoints. In general, I'd use the Density object, which is the result of the analysis instead of private attributes of DensityAnalysis
• I have more suggestions for 3D plotting in https://gist.github.com/orbeckst/a6a61c8f20a434c1c7520cd51c281522 (including nglview density and matplotlib surface contours, mostly based on https://stackoverflow.com/a/35472146

[orbeckst]

Continuation review ... and final notebook:

H-bond analysis

• An introductory sentence on what is generally considered a hydrogen bond would be nice.
• add link to API docs (I'd prefer if all mentioning of MDAnalysis objects such as MDAnalysis.analysis.hydrogenbonds.HydrogenBondAnalysis linked to the docs but presumably there's not markup in the notebooks equivalent to the sphinx roles that enable intersphinx linking?)
• The conditions for guessing were clearly spelled out – better than in the API docs! Is there a way for a user to create a list of all donors/acceptors that were considered for their system, e.g., so that they can check that whatever they think ought to be included is actually included (e.g., Methionine S_delta as acceptor (Gregoret et al, Proteins 9 1991, doi:10.1002/prot.340090204.))?
• update_selections should be explained and point out that performance can be increased with False if the groups are static.

SUMMARY

Great content, we just need to make sure that users find it!

[lilyminium]

@orbeckst thanks for the reviews and sorry for taking months to get back to you. In the interest of not forgetting about this for more months, I might merge tomorrow morning unless new changes are requested.

[orbeckst]

No problem. Merging sounds good to me.

…

Am 6/20/21 um 13:51 schrieb Lily Wang ***@***.***>:  @orbeckst thanks for the reviews and sorry for taking months to get back to you. In the interest of not forgetting about this for more months, I might merge tomorrow morning unless new changes are requested. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.