linkers + groupos

[velocirraptor23]

Hi,

I have made some tests, but I would like to make molecules all selcted rgroups against all selected linkers (probably all). I was wondering why it fails when I use this function .
cs.add_rgroups(linkers, [r_groups])
r_groups has embed all r groups. Do I need one by one or can i do it programatically.

Best wishes,

Cesar

[bieniekmateusz]

I am afraid it is not an option right now. I guess a loop would be needed. I wonder if nudging the interface towards working with itertools.product would be useful for this, ie:

cs.add_rgroups( [ (linker1, rgroup1), (linker2, rgroup2), ... ])

which would be equivalent to:

cs.add_rgroups(itertools.product(all_linkers, all_rgroups))

[velocirraptor23]

Thanks for your response.
i did this, seems to work:
for r_group in r_groups:
cs.add_rgroups(linkers, [r_group])

BW,

Cesar

[velocirraptor23]

I did this, it worked perfect:
This does a better job to make all the combinations:

You can select the groupos directly from the lists like this:

r_groups = rgroups.Mol[0:50] 
linker_list = linkers.Mol[6:9]

##and then:
import itertools
for r_group, linker in itertools.product(r_groups, linker_list):
    cs.add_rgroups(linker, [r_group])

Hope this helps anyone interested.

BW,

Cesar

[bieniekmateusz]

Thanks for sharing, just to let you know, the last one will work but will not scale (because of how Dask handles it). If scalibility is ever here crucial please let us know!

[velocirraptor23]

Yes, it would really nice to have scalibility on this, because this is where it really needs to handle lots of molecules.

Thanks for your response!!

Cesar

[velocirraptor23]

By the way, I am doing this way because the graphical selection does not work properly. For some reason, it does not pick the groups. I think that way should not have a problem and make a combination of all groups selected.

[bieniekmateusz]

There is a bit of order dependance when using the graphical selection. I believe there is a function linkers.get_selected() which can help find out if it does select the right thing. Thanks

[velocirraptor23]

Hi,

Looking into the other notebooks I found this that do the kind of work I need:
'''
cs.add_rgroups(linkers.Mol[0:50].tolist(), rgroups.Mol[0:50].tolist())
##or
cs.add_rgroups(linkers.Mol[:50].tolist(), rgroups.Mol[:50].tolist())
'''
However, it does not make cross combinations. It produces only 50 compounds but in reallity if we combine all we should have 2500 compounds. When I do this with itertools it actually does not run properly either the parametrization and the optimization. You are right it breaks the Dask implementation. But the option I mentioned here runs properly.
Any suggestion if there is a way to make cross combination would be really useful.

BW,

Cesar

[bieniekmateusz]

Hey, I added this functionality and will release 2.1.0 with this feature. At the moment the code is available in the pull request if you'd like to try it. It should be merged soon into the main branch. Thanks!

[velocirraptor23]

Thanks a lot for taking the time for this. I am going to test this later today.

Best wishes,

Cesar

[velocirraptor23]

I have tested this in a small set and it works very well. I am doing this in a bigger data set.

Thnaks a lot.