search

chembl / mychembl

Resources used to create the myChEMBL virtual machine
56 stars 29 forks source link

Make Notebook 01 more like Chemistry Toolkit Rosetta #9

Open mnowotka opened 9 years ago

mnowotka commented 9 years ago

There is a Chemistry Toolkit Rosetta Wiki, which is a very valuable resource, comparing chemistry toolkit. Since we already have RDkit, OpenBabel and Indigo on board, we can include some of the stuff presented there in our introductory Notebook.

In the introduction notebook, we should put as much references to other notebooks as well to attract the user. Examples may contain for example retrieving data from ChEMBL using raw SQL, ORM, raw REST calls, client.

flatkinson commented 9 years ago

I would be cautious about introducing material using other toolkits without a strong rationale. If we need to use other toolkits behind the scenes (e.g. Indigo for depictions), that's fine, but I'd vote to keep the focus on RDKit for public-facing stuff.

mnowotka commented 9 years ago

Why?

On Fri, Oct 17, 2014 at 9:15 AM, flatkinson notifications@github.com wrote:

I would be cautious about introducing material using other toolkits without a strong rationale. If we need to use other toolkits behind the scenes (e.g. Indigo for depictions), that's fine, but I'd vote to keep the focus on RDKit for public-facing stuff.

— Reply to this email directly or view it on GitHub https://github.com/chembl/mychembl/issues/9#issuecomment-59480713.

flatkinson commented 9 years ago

OK: I, personally (YMMV /etc/.), prefer to focus on one toolset. In the past I've been in the situation of needing to mix and match cheminformatics tools and toolkits (Daylight + MDL + Tripos most notably, with MOE, Schrodinger, Accelrys, OpenBabel and ChemAxon stuff introduced along the way) and it usually leads to pain in the end, due to different chemistry models, assumptions /etc/. Of course, there are always cases where you really have to work with multiple tools at once, but generally I think you're better off focussing on one, even where it's sometimes slightly suboptimal.

Also, fragmentation can be a problem in this field, and it's important to keep up momemtum behind a project. I'd rather see us 'partner with RDKit', if you like, and see if we can help it move forward. In fact, we could approach Greg and see what he'd think of bundling his tutorial material with myChEMBL, so people can give them a go (/e.g./ via notebooks) before they invest time in installing the beast (installation not being it's best feature).

We should also make sure as much of the forthcoming drug-design course material is myChEMBL compatible, so people can take it away and re-run it offline without problems.

On 17/10/2014 09:27, Michał Nowotka wrote:

Why?

On Fri, Oct 17, 2014 at 9:15 AM, flatkinson notifications@github.com wrote:

I would be cautious about introducing material using other toolkits without a strong rationale. If we need to use other toolkits behind the scenes (e.g. Indigo for depictions), that's fine, but I'd vote to keep the focus on RDKit for public-facing stuff.

— Reply to this email directly or view it on GitHub https://github.com/chembl/mychembl/issues/9#issuecomment-59480713.

— Reply to this email directly or view it on GitHub https://github.com/chembl/mychembl/issues/9#issuecomment-59482456.

Dr Francis L Atkinson

Chemogenomics Group European Bioinformatics Institute (EMBL-EBI) European Molecular Biology Laboratory Wellcome Trust Genome Campus Hinxton Cambridge CB10 1SD United Kingdom

(01223) 494473

mnowotka commented 9 years ago

Comparing RDKit to other toolkits doesn't mean we are not focusing on RDKit. I'm not proposing to mix different libraries in our software stack - in fact we use RDKit almost exclusively and this is not going to change. But I can imagine that people using my_chembl can have different backgrounds so a quick guide of how to achieve the same functionality using RDKit can be helpful.