Running Rowan's pKa Prediction Workflow

Transcript

Hi, I'm Corin, CEO and co-founder of Rowan, and in this video we're going to talk about pKa prediction: what pKa means, how to run pKa calculations through Rowan, and how to interpret the results.

First off, what does pKa mean? So pKa allows us to quantify the acidity or basicity of different molecules and different functional groups within molecules. So basically the pKa is a single parameter which refers to how easy or hard it is to remove or add protons from a given site. This is a pretty foundational concept in organic chemistry. So a lot of intro organic chemistry classes will involve students having to look through and memorize lists and charts of pKa data for given molecules, to build up an intuition about how different substitution and different functional groups will impact the pKa of different sites. This is done because pKa is super important: it determines reactivity, it determines whether a molecule will be charged or neutral under various conditions, and pKa-related thinking is used to inform a whole lot of other sort of structure–activity relationships in organic chemistry.

If you've been around organic chemistry for a while, you might recognize this as the Evans pKa table. It's pretty famous, at least for people within a certain subculture, and it's just composed of lists and lists and lists of pKa values for different compounds in water and DMSO. This is pretty nice if the compound you care about is actually in this table, but another fact about pKa is that it's very sensitive to substitution. In fact, pKa trends are used to compare other things that are sensitive to substitution through Hammett relationships. Because pKa values are so sensitive to substitution, it's nice to be able to compute them from scratch for the molecule we care about, instead of having to dig through tables like this or the literature to find reliable reference data for the most similar compound we can find.

So let's look at what it takes to actually calculate a pKa value in Rowan. So we'll do so for pyridine, which is a really nice example compound. It's one of the most fundamental heterocycles, and the value we'll try to match for this compound here with R=H is 5.21. So let's keep that in the back of our minds, and then we'll run a pKa calculation on pyridine in Rowan. So we'll change tabs and click on the "new pKa prediction" button from the home screen and that will take us here and we can draw a 3D structure: we can say add from library, go to rings, phenyl, then we'll go to our periodic table, we can add a nitrogen to this benzene ring to make it into a protonated pyridine and then remove that proton to just give us pyridine. So we'll call this "pyridine pKa."

And now we have a bunch of choices on this settings half of the screen. And the reason we have so many choices is that there's actually a lot of different sites where we can add or remove protons, even on a molecule as simple as pyridine. So we can imagine adding a proton to every heavy atom here, so each of these five carbons plus this nitrogen. And then we can imagine removing any of these five protons. So altogether, that's 11 different sites for pKa, and that's usually more than we care about. We have two different ways to filter down to the set of questions that you actually have in Rowan. One of this is setting this min and max pKa window which essentially says that we're not going to look for pKas which are crazy low or crazy high, that might exist only under extreme conditions; we're just going to restrict ourselves to asking about pKas within the range 2 and 12 which is pretty normal for physiological or close-to-physiological conditions. If we were looking at, for instance, enolate deprotonation or enolate generation, this sort of strong base chemistry or superacid chemistry, we might want to make this window a little bit larger so that we can actually see the pKas that we care about for that. And then the other thing that we have here is we want to select the elements that we're going to protonate or deprotonate. And in this case,the default settings will be that we only protonate on nitrogen, and we only deprotonate from nitrogen, oxygen, and sulfur. So notice how carbon is on neither of these lists, and if you want to run a pKa calculation that involves deprotonating carbon, again, for studying enolate chemistry, you might want to add that to the "deprotonate elements" list.

Finally, we also have our mode setting. And in this case, at a high level, this controls the trade-off between error and efficiency. And what this usually boils down to is actually how we generate all of the conformers that we need to, for this calculation. So for each charge state that we look at, we're going to have to do a conformer search, and that can be pretty expensive for more complex molecules. And so we can tune it between reckless and careful to try to figure out how much we want to spend on each calculation for higher and more precise results. In most cases careful is appropriate, and so we'll just click "submit."

Behind the scenes what this is doing is, as I alluded to before, we're going to iterate over all the sites and compute which protonation or deprotonation sites are reasonable then we'll run a conformer search on each compound, optimize each of them, get a final sort of proton affinity, and then convert that to a pKa. And what this finally gives us now for this simple molecule is just a single site where we can protonate, and that's right here. So the value we were trying to match is 5.21, and the value we got is 5.11, so that's pretty good. And we can view the conjugate acid here, to confirm that protonation is happening as we expected, and this is exactly what we expect protonated pyridine to look like.

Okay, so that was pretty straightforward: we can see how long that calculation took in total, and that was under 10 seconds. So let's now try again on a slightly more ambitious molecule. We'll just add some extra functionality to this to tax ourselves a little bit more, so we can go to the periodic table here: we can add an oxygen, maybe we can add another oxygen, and we can add a carbon here. So now there's at least three sites that we might care about, and potentially more. We'll say "complex pyridine pKa" because I don't want to try to systematically name this compound on video, we'll leave our same settings because they seem good, and we'll say "submit pKa" again.

It's worth noting that Rowan's pKa workflow gives microscopic pKas and not macroscopic pKas. So microscopic pKas are broken down by atom, whereas macroscopic pKa reflects the propensity of the molecule as a whole to add or remove protons. We focused on microscopic pKas for now because they're more fundamental, in a sense. The microstates are the underlying unit from which we build up the macrostates, and also let us break it down by functional group, like this. In the future, we'll look into incorporating macroscopic pKas into Rowan, but we haven't yet done that.

And so, you know, now what we see is, first of all, we optimized the geometry. So we have this cool internal hydrogen bonding network—that's always interesting to see. And now we actually have three relevant sites. So we have our original nitrogen again, which is now a bit more basic, which makes sense because we've added these electron donating groups onto the ring, as well as this methyl group. And we also have one oxygen and two oxygens. What we can see, interestingly enough, is that this one is a bit more acidic than this one. So if we do deprotonate at a given site, you know, we expect it to be from there, the one that can form the pyridone-looking tautomer, as opposed to this one, which is going to be slightly less acidic. And that's a difference of about one and a half pKa units. So it's substantial, but it's not insane.

Another thing to note here is that we've colored the two different sorts of sites differently. So really, pKa is used in sort of two different ways in Rowan and in most other discussions of pKa outside the physical chemistry literature. We can refer to the pKa of adding a proton somewhere and the pKa of removing a proton from somewhere and strictly, the pKa that we're looking at here is the pKa of the conjugate acid. So really, it's this molecule that properly has a pKa of 6.79, and we should be using the term pKb to be precise. But because it's pretty obvious when we talk about the pKa of a pyridine as being 5 or 6 that we're really referring to the conjugate acid, it's most common just to talk about it in these sort of joint terms. To help disambiguate these in Rowan, we color the pKbs as blue when we're adding a proton, and the pKas where we remove a proton, we color those red. Thanks for watching!