Benzylic amine compounds update

[edwintse]

The trend that has been seen with previous benzylic amine compounds (inherited or synthesised) is as follows: primary and tertiary amines are active, secondary amine was inactive

Following on from my previous post on my next set of compounds (#481), a number of these compounds will now be synthesised. Scale up of the key benzylic ketone compound has been done and a number of reductive aminations can be done for diversification.

n.b. the LHS phenyl ring will be used instead of the 3,4-difluorophenyl ring due to ease of access.

Before further diversification, these compounds will be evaluated first, but if there are any suggestions for other compounds that can be obtained from this ketone, suggest away.

chemdraw.zip

[mattodd]

I'm speaking at this conference at Tres Cantos next week, about OSM and all other things open. I'll include this as an example of a "live" question.

My votes are, in order: 1) hydroxylamine, 2) morpholine, 3) NMeEt.

[MFernflower]

@edwintse @mattodd Sorry for having to repost but here are a few ideas I came up with:

[spadavec]

Using MMP/Regression approach, I developed 47 different ideas for new compounds which can be found here, and the workbook found here (which is essentially a carbon copy of my previous implementation).

For the sake of brevity, here are the top 10 most (predicted) potent compounds--some rather silly, but others perhaps not so much

For what its worth, these are all predicted to be roughly equipotent with the most potent S4 compounds; in or around ~500nM

[MFernflower]

@spadavec My votes for synthesis out of that batch is as follows:

END-D074AC
END-BB326D
END-C22BEE

[mattodd]

Hi @spadavec - just to clarify. All your suggestions have the oxygen in the side chain transposed. @edwintse 's core has the oxygen directly linked to the triazolopyrazine core? If that's a copy/paste error, could you re-predict on Ed's core? Would be v useful.

[spadavec]

@mattodd wow, you're completely right. I initially copied the wrong smiles string and ran this analysis with the wrong seed; I can't believe I didn't catch that!

Updated predictions found here, with top 15 shown here (very little change from before):

Again, a few silly ones, but the trend seems to be for preferring tetrazoles and triazoles substitutions.

[mattodd]

Nice @spadavec . In an ideal world, we'd not introduce another aromatic ring. Are there any simple analogs that involve e.g. similar ring types, but saturated, e.g. pyrrolidine etc?
@edwintse what do you reckon to the above. Accessible?

[spadavec]

@mattodd sadly, all of the substitutions it suggested are aromatic ring derivatives. I'm currently constructing a new tool that does simple organic chemistry rules based enumeration and might be able to have that ready in the next day or two.

[MFernflower]

@spadavec would you be able to screen the proposed compounds I posted? I am away from my linux box at the moment! as always great work!!!

[holeung]

I ran @spadavec's molecules through my 3D-QSAR model. Top rated were F3221C, A26A0A, C52B4B, F961EC, B3D09B. I suggest dropping F3221C because it would be deprotonated at physiological pH.

benzylic_amines_top5.sdf.zip

[spadavec]

@MFernflower I was able to run your ideas, and they are roughly equipotent with each other and the lead compound:

Substitution	SMILES	cLogP	EC50
Pyrrole	FC(F)OC1=CC=C(C=C1)C1=NN=C2C=NC=C(OCC(N3C=CC=C3)C3=CC=CC=C3)N12	4.8626	0.52
pyrrolidine	FC(F)OC1=CC=C(C=C1)C1=NN=C2C=NC=C(OCC(N3CCCC3)C3=CC=CC=C3)N12	4.6086	0.61
azetidine	FC(F)OC1=CC=C(C=C1)C1=NN=C2C=NC=C(OCC(N3CCC3)C3=CC=CC=C3)N12	4.2185	0.63
thianedione	FC(F)OC1=CC=C(C=C1)C1=NN=C2C=NC=C(OCC(N3CCS(=O)(=O)CC3)C3=CC=CC=C3)N12	3.243	0.70

[mattodd]

Good stuff. I'm still greatly preferring the non-aromatic analogs like the pyrrolidine/azetidine. Any further insight/predictions in that direction would be very useful. @edwintse will also love you if it's single-step from ketone. I'm thinking NEt2 vs pyrrolidine would be obviously nice.

[holeung]

@mattodd: I looked into non-aromatic analogs of @spadavec's molecules, but they scored much lower in my models.

[MFernflower]

Some more ideas I came up with: (sorry about having to post the raw CSV - MarvinSketch was not cooperating today!)

regression-predictions.csv.txt

[edwintse]

Three benzylic amines from the original post were synthesised and sent for biological testing. Relevant post here.