On making pretty structures

I had an exchange with @fluorogrol today talking about poorly thought out chemical structures.  It began when the pseudonymous blogger tweeted out the (less than optimal) structure of this alkaloid:

Wrong, indeed. Representing complex stereochemistry can be quite tricky.  But bonds extending halfway across the molecule is generally not the way to go.  Chemical structures must satisfy two requirements: 1) they must be unambiguous, and 2) they must accurately represent the 3-dimensional shape of the the actual molecule.  I responded with my take:

…which was straightened out and rotated a bit, giving us this final, unambiguous structure:

No exaggerated bonds, no stereocenters consisting of a cluster of wedges.  Plus, now you can quickly get an idea for the actual shape of the molecule.  Now, admittedly, I have no idea where the original structure appeared, but I’m going to assume it’s from a publication.  Which means a group of scientists, and presumably at least one organic chemist, collectively decided that structure was the best way to represent akuammine (the wikipedia page structure is equally bad).

And to think, this tragedy could have been easily avoided.  Here’s a quick checklist on how to draw an accurate, sexy structure of your molecule.

1) Does the molecule have any cyclic or bicyclic motifs common in organic chemistry?

To list a few...

…to list a few

If yes, start there.  These structures are so ubiquitous in organic chemistry that they give readers a quick and easy 3-D reference point.  Not sure if there’s one of these in your molecule? Proceed to step 2.

2) Load you molecule up in the three-dimensional molecule viewer of your choice.  The professional edition of ChemDraw comes with Chem3D, which will do the trick.  As will whatever you may use for energy calculations.  Don’t have access to any 3D chemistry software?  MolView has you covered with its in-browser tool, which can even perform simple energy minimization calculations.

Rotate the structure around a bit.  Look for any of the aforementioned motifs.


Still can’t find one?  Then…

3) Find and angle from which all atoms are more or less visible.  Of course, your ChemDraw structure doesn’t need to perfectly match all bond angles and distances, but try to replicate them as faithfully as possible.  A good angle is one from which stereochemistry is unambiguous and doesn’t require dashes and wedges everywhere to makes sense of the structure.  Also, don’t forget the ever underutilized “structure perspective” tool in ChemDraw; make judicious use of it.

Maybe I’ll start featuring a “bad structure of the week” here.


Strain energy for days: an in silico study of xinghaiamine A

SeeArrOh (via Twitter) reminded me of something I’ve been wanting to check out since this paper surfaced.

The paper in question features a supposed natural product, named “xinghaiamine A,” with some pretty wonky bonding.  Readers at Just Like Cooking and In the Pipeline brought up some issues regarding the evidence for this compound’s existence.  And rightly so; there appears to be something off about the supplemental data¹.  But, ignoring the (very real) issues readers have brought up with the supporting info for this paper, just look at this structure:

Oh dear

One half of the proposed compound.

At first glance, there’s some serious strain going on in there.  I figured I’d take a look at what xinghaiamine A looks like in 3D-space.  Getting it to behave in Spartan was a challenge on its own.  Chiefly, that bicyclo[2.2.0]hexane system was quite problematic.  Initial geometry optimizations at the semi-empirical level of theory produced some odd results.  I ended up settling on MMFF geometry optimization, which gave me the reasonably acceptable structure shown below²:

MMFF geometry optimization

MMFF geometry optimization of the xinghaiamine A “monomer”

Check out that bowl-shaped aromatic system.  That thing is supposed to be planar.

Check out that torsion angle: 40 degrees!

A torsion angle of 40 degrees.  And how!

The next logical step is to figure out exactly how much strain energy is in this thing.  This was done by taking the MMFF optimized geometry of xinghaiamine A and using it as a starting point for Spartan’s “T1 thermochemical recipe.”³  The T1 recipe is a post-Hartree-Fock method which consists of:

  • A quick and dirty HF/6-31G* geometry optimization
  • MP2 single point energy calculation with expanded basis set

This set of calculations yielded a heat of formation for xinghaiamine A of 1098.65 kJ/mol

Now, if we break that C-C bond joining the acenaphthalene and the bicyclo[2.2.0]hexane systems, repeat the calculations, and compare the results we can get a pretty decent idea of how much strain energy this proposed structure contains:

That planar acenaphthalene system looks so much happpier

That planar acenaphthalene system looks so much happpier

The answer is: a lot.

Breaking that one bond liberates quite a bit of energy.  But that’s not what makes this structure so implausible.  No, as others have pointed out, some of the motifs in this molecule have never been seen in a natural product.  And if you’re going to propose something never-before-seen, you best have the evidence to back it up.

Which raises the question: did the authors think the chemistry community would look at that structure and collectively go “yup, looks good to me, moving on then”?

  1. Something that rhymes with “fata dabrication”
  2. Note: the published structure is a dimer.  I’ve modeled it as a monomer, with a methyl R-group for computational simplicity
  3. Not a shill for Spartan, I promise

Reporting Adverse Results in Publications

Here’s a question: to what extent should authors describe the limitations of their results in publications?

I’m not talking about failed experiments and negative results or null-hypotheses.

For example, say a compound is synthesized, but is highly unstable in air, causing it to decompose rapidly.  Should that be noted?  Or maybe someone makes a new polymer that undergoes depolymerization after a couple days on the bench.  Should the researchers describe that in their publication?

Of course, it’s contextual.  Does the negative observation directly impact potential applications of the material?  If yes, then one would think it imperative accurately describe the limitations of research, especially if they are known at the time of publication.  It seems irresponsible to intentionally leave out these details.  And it is exceedingly frustrating as a scientist to discover these sort of details independently, when replicating others’ results.

Obviously, it hurts your chances of getting into a high impact journal if your results are muddled by adverse circumstances at the back-end.  And not every single negative result is noteworthy.  Library synthesis papers don’t describe every single failed substrate, but they often note something along the lines of “[class of compound] did not react to produce the desired product.”

I don’t have a perfect answer to this question.  Thoughts are welcome by comment or email.