Interest in chemistry-focused BoF?

Thanks, @mfh. I would love to see how it plays with Fermi. Do you know if it support single precision?

I am running their Julia example, but I am getting random values… Let’s see if I can figure this out…

Very interested

1 Like

Not easily. As far as I can see double is hard-coded as the internally-used type.

This sounds all super interesting.
From my perspective Julia needs, either a nice wrapper for rdkit, or something similar as a native package (I know that I can call rdkit through python). Maybe a julia package that wraps/calls the rdkit c++ code?

That would be wonderful and would in my opinion boost Julia acceptance in the industrial chemistry domain. Since, rdkit is the defacto standard to manipulate small molecules.

3 Likes

Agreed with that. And selfishly, PyCall to rdkit causes me so many CI problems… :roll_eyes: Certainly directly calling the C code could potentially be a good medium- (or even long-) term workaround.

I did end up submitting a BoF proposal, so hopefully it will be accepted and we can all come together to discuss directions for these things this summer! :slightly_smiling_face:

5 Likes

I do not want to hijack this topic but this could be interesting for the Julia-chemistry community:

Thanks for sharing! I don’t think it’s a hijack at all, because this represents a very plausible solution modality for some of the issues we’ve been discussing that probably hasn’t gotten enough discussion here as yet!

1 Like

Hi there! I’m the author of the RDKit blogpost and a new Julia user :slight_smile: I agree with both of you that having RDKit support for Julia can boost the use of Julia in the small molecule comunity. We, at ChEMBL, use it for almost everything.
I’ve heard about PyCall but it is something that I would prefer to avoid in production.

The MinimalLib will never have the full RDKit functionality but Greg is open to suggestions and I think that it is a quite nice start. People working in ML, for example, could already use it since it is already generating physicochemical descriptors and molecular fingerprints.

In the long I guess that it would make sense to try to write some wraps with CxxWrap.jl but this would certainly be a much complex project than the RDKitMinimalLib thingy I wrote.

7 Likes

Just wanted to let anyone listening here know that the BoF is ON! We have a small team of folks who will be organizing/running it. If you’re interested in helping out, feel free to let me know, or just keep this in the back of your mind and attend in July! :smiley:

11 Likes

Hi guys,

I started ElementaryChemistry I’m only just beginning and I’m certainly no professional. But if there suggestions or ideas I’d be happy to hear them !

1 Like

I am a graduate student working with @SimonEnsemble (GitHub) on modeling and simulation of porous crystalline materials. We are building an ecosystem of packages that handle several tasks mentioned here. Xtals.jl provides utilities for reading/writing various structure file formats (including .xyz and .cif), and manipulating/editing the structures. PorousMaterials.jl provides an interface for Lennard-Jones potential mapping, Henry coefficient calculations, and Grand-Canonical Monte Carlo simulations of adsorption isotherms. Both are available on the general registry, and both are under active development for additional features, along with some soon-to-be released packages for hypothetical structure generation and graph neural net property prediction.

Looking forward to the BoF event!

4 Likes

Nice. Two things: I have developed this package that computes non-bonded energies using cell lists, which might accelerate significantly the computation of energies (this loop in particular). I guess that in the Porous Materials field general periodic boundary conditions are fundamental. I am still implementing support for that in CellListMap, but I hope that will be available very soon.

More generally, I am the developer of Packmol, which some may have heard about, particularly in that field. I am starting the development of a new Packmol version in Julia with more features and flexibility. I would be grateful if your group, if using Packmol, provided some feedback on possible new features that you would like to see in it.

1 Like

Yes, periodic boundaries are always a part of the picture for us–this is automatically handled in our distance calculations. The function you point out could potentially see a speed boost by using your suggestion; thanks! One difficulty is that there is no “master list” for indexing every atom in a simulation, but I will open an issue to flag this for further consideration :slightly_smiling_face:

1 Like

Xtals.jl looks awesome! Are there plans to support other file formats? Even just .cif and .xyz is already awesome; I’ll try to find some time to play with it and maybe incorporate into ChemistryFeaturization.jl, which currently relies on ASE for all its file reading needs…

Looking forward to having you at the BoF!

.cssr, .mol, and .mol2 are already included as well. Other formats can be added, and are from time to time when we need to take in data from new sources. We’re always open to folks to opening issues or PRs for new feature requests :smiley:

Hi again everyone! Just wanted to let everyone know that the conference schedule is out so be sure to put this session on your calendars! Building a Chemistry and Materials Science Ecosystem in Julia :: JuliaCon 2021 (times are UTC) :: pretalx

We’re also hoping that the conversation can continue well past this 90 minutes, so keep an eye out at this website for other ways to stay connected (it’s pretty barebones right now, but there is a Slack link!): https://juliamolsim.github.io/qchem/

Thanks again for your interest; hoping to kick off some awesome conversations!

2 Likes

Hi! Regarding the file I/O issues, I have been developing chemfiles to help with this issue since 2015. It is not a pure Julia package, since all of the I/O is implemented in C++ with a C API (code here), but there are Julia bindings (documentation here). We already support more than 20 formats, and we are adding support for new formats regularly. The code is also quite optimized, see some or our benchmarks here, and benchmarks by other people including Julia code there.

I think that the code being written in C++ and thus accessible from more languages is a strength for chemfiles, since developers using C, Fortran, Python, Julia, Rust, … can use it as well. This mean bugs are discovered faster, and multiple language communities can share the maintenance. The downside of the code not being written in Julia/Python/… is that it make it harder for folks from these communities to contribute.

I’ll try to join the BoF session to discuss all of this, so see you there!

3 Likes

Completely agree - so many edge cases to fix… great to share code across languages

@leandromartinez98 I am starting the development of a new Packmol version in Julia with more features and flexibility. I would be grateful if your group, if using Packmol, provided some feedback on possible new features that you would like to see in it.

I have used Packmol a few times for preparing molecular dynamics simulations. Some of the features that I think would be useful are : 1) being more compatible with different file types and 2) for the output of the optimisation, especially the part where GENCAN loops begin to be a bit more readable. Apart from this, if the software were allowed GPU usage for faster processing, that would be a big plus. I really like packmol and if possible I would like to get involved in the development project of the Julia version of it. Is there any forum where I can engage in further discussions on the matter?

2 Likes

Quick reminder that the BoF session will be tomorrow, so be sure not to miss it.

We’ve also set up a HackMD notebook, where we want to collect some interactive notes during the session for future reference. So in case you won’t manage to attend the live discussion, feel free to take a look at a later point.

Hope to see you all tomorrow :slight_smile:.

2 Likes