Compound Enumeration and Ideation for Series 4

[spadavec]

It looks like we are still open to the idea of exploring the SAR of some compounds in series 4 (Issue 483, Issue 481), with a goal of keeping the LogP down and potency in the same ballpark. As such, I've been working on a matched molecular pairs approach (described here) that attempts to explore the chemical space around the lead compound:

MMV669844: O=C(NC1=CC=NC(C(F)(F)F)=C1)C2=CN=CC3=NN=C(C4=CC=C(OC(F)F)C=C4)N32

I was able to generate a list of compounds that have some predicted properties that might spark some ideas among those interested:

For the LogP and predicted pEC50 values, please check my lab notebook, but suffice to say they are all predicted to be LogP < 5 and pEC50 < 0. If there is significant interest, I can enumerate more compounds using other lead compounds as seed ideas.

Some of the ideas aren't so hot (e.g. END8, END34, END11, etc), but I had a rather crude filter to get down to the most decent ideas; this filter can easily be adjusted to something more reasonable by someone more knowledgeable than I.

[mattodd]

Hi @spadavec - this is excellent. MMP analysis is something OSM needs on this series. Can I immediately request a filter tweak? I'd like cLogP < 3 ideally. Are there compounds that pass that?

[MFernflower]

@spadavec @mattodd I would also like a PAINS filter... That peroxy-epoxy compound and that quinone are borderline rocketfuel!

I really do like a few of these hops however! The triethylsilane replacing the phenyl OCHF2 seems like a neat idea for example!

[spadavec]

@mattodd in this dataset, I'd say there are only 4 compounds which aren't fragment-like and have cLogP < 3:

However, I'm sure I could generate more ideas if I were to use more compounds as seeds for the process. Are there any particular compounds I should use? Unfortunately I can't use all of series 4, as that would likely explode to hundreds of millions of compounds :)

@MFernflower no problem! I think I have some SMARTS implementations of PAINS laying around and will filter out those as well.

EDIT: Adding table, as the text in images came in so poorly

[MFernflower]

@spadavec MMV669304 would be interesting as there is no oxygen linker

See issue #400 for some other seed ideas!

[spadavec]

I re-ran the calculations using the isoindole (MMV669848), amide/pyridine (MMV670944), and two ethers (MMV670437, MMV670947) as seed compounds, which generated 327 unique compounds which passed PAINS and had following properties:

cLogP < 3
275 < molwt < 500
# hbond accp < 10
# hbond donor < 5

There's a bit to sift through, so I've attached a spreadsheet here: OSM_Series4_MMP.zip

My filters still let through a few fragment like compounds through, but I left those in just for curiosities sake.

EDIT: For the more visually inclined, here are the 29 compounds with cLogP < 2 with predicted potency values ~<1 microM

[MFernflower]

@spadavec @edwintse I picked out three that I think might be viable from the zip posted 9 hours ago:

[spadavec]

Continuing from issue #488 : @MFernflower

Given the accuracy of the model, these are equipotent and only END7 and END7-MOD3 are more potent

[mattodd]

Some v interesting predictions above in your set of 29, Vito, wow. @MFernflower The Mod of END196 is similar to MMV688898 (OCHF2 vs CN) which is inactive. END137 is similar to MMV672723 (OCHF2 vs CN, and Ar-Fs) which was 100 nM. END102 has not been made, I don’t think. @edwintse ?

I know the model has not been validated @spadavec but I'm interested: what's the prediction for the Norcross compounds? You have an O-bridge in your set (END127).

[spadavec]

@mattodd I've attached the results for the Norcross compounds here
and the Google sheet here:

Given the accuracy of the model (pEC50 +/- ~0.5), in order to stay roughly equipotent with current best-in-class, predicted pEC50 need to (at minimum) come in <~0. Given this, it looks like the best bets are N12, N11, N15, N16, and N13

[mattodd]

Great - thanks. Will incorporate this when we re-invigorate approaches to the Norcross compounds which have stalled recently.