To review or not to review?

I'll admit that I don’t always see eye-to-eye with journal editors although it has been a while since I last blogged on this theme.  Recently we published an article on prediction of alkane/water partition coefficients and, as was noted at the end of the blog post, the article had originally been submitted to another journal which declined to review the submission.  Much is written about peer review and, if your manuscript actually gets reviewed, you’ll get a good idea about why the journal said nein.  Put another way, there is some accountability even if you don’t like the outcome.

It can be very different when the journal declines to review and I see this as a sort of shadow land in the manuscript review world.  You often (usually?) don’t get any useful feedback as to why your submission was not even worthy of review and the explanation will often be that the manuscript lacked broad appeal.  Sometimes the manuscript will be farmed out to a member of the editorial advisory board who will effectively make the decision for the editor but you never get to see what actually got said.   This is why I talk about a shadow land. 

Our alkane/water partition coefficient manuscript was originally submitted to the Journal of Chemical Information and Modeling and I really did think it was a JCIM sort of manuscript.  Those open science enthusiasts reading this might be interested to know that we were providing source code (albeit source that used the proprietary OEChem toolkit from OpenEye) and the data that was used to build and test the model.  I went through the submission process which included creating a graphical abstract and providing email addresses for potential reviewers (whom I contacted to see if it’d be OK).  The response from JCIM was swift:

'Thank you for considering the Journal of Chemical Information and Modeling for your manuscript submission. The Journal is committed to promoting high standards in the fields that it covers, so we will be quite selective. Most ACS journals including JCIM are also experiencing substantial increases in submissions that are burdening the editorial and reviewing systems. Consequently, most editors are now prescreening and returning ca. 30% of submissions without further review.

For JCIM, some typical problems that will lead to return of a manuscript by the Editor include papers that:

(A) fall outside the aims and scope of JCIM, sometimes characterized by few if any references to papers in JCIM or other mainstream journals in computational chemistry,

(B) are routine studies, do not provide significant new results, or promote new methods without demonstration of advantages over prior alternatives,

(C) center on developments and interest in research areas that are tangential to the scope of JCIM, or

(D) are characterized by poor presentation of the work that is inconsistent with the style and quality of typical papers in JCIM. In particular, QSAR/QSPR submissions are welcome, but authors must be careful to avoid problems under heading (B).

In my judgment, your submission is inappropriate for JCIM.  The journal has limited interest in QSAR/QSPR studies; ones that are considered need to adhere to the guidelines published in JCIM 46, 937 (2006).

I regret any inconvenience, but hope that this prompt return will allow you to consider alternative journals without further delay.' (07-Mar-2013)

I was genuinely surprised by the decision but take the view that journals can publish anything they want to. Nobody is entitled to have their manuscripts published.  Unless of course they have paid for the service as is the case for Open Access.  I was a bit alarmed that the journal appeared to be using number of references to its own articles ('…fall outside the aims and scope of JCIM, sometimes characterized by few if any references to papers in JCIM or other mainstream journals in computational chemistry') as a criterion for deciding whether to send a manuscript for review. However, life must go on and so I replied as follows:

'Thank you for your prompt response to our manuscript submission. While we are disappointed to learn that JCIM is not interested in prediction of alkane/water partition coefficients, we fully accept your decision and will seek a more appropriate home for the manuscript.' (07-Mar-2013)

The manuscript was duly submitted to JCAMD which accepted it but just as the article went to press, I noticed that JCIM had recently published an article on alkane/water partition coefficients.  I was aware that our article had cited two or three articles from JCIM’s predecessor JCICS but only one from JCIM itself.  Given that that it appeared that JCIM were actually interested in alkane/water partitioning, I sought specific feedback as to why our submission was not considered worth sending for review:

'Although our alkane/water partition article is currently in press at JCAMD, I must now confess to being perplexed by your decision to reject the manuscript without review given the recent publication of "Structural Determinants of Drug Partitioning in n-Hexadecane/Water System" (  http://dx.doi.org/10.1021/ci400112k ) in JCIM.  The possibility that number of references to JCIM articles might be used as a criterion for deciding whether to send a manuscript for review did worry me (and continues to do so) when you first communicated the decision in March.  I would be interested to know if you are able to comment more specifically on this matter.' (23-May-2013)

JCIM have not yet responded to my query and I am still perplexed.

Literature cited

Kenny, Montanari, Propopczyk (2013) ClogP_alk: A method for predicting alkane/water partition coefficient. JCAMD 27:389-402 DOI

Natesan, Wang, Lukacova, Peng, Subramaniam, Lynch, Balaz (2013) Structural determinants of drug partitioning in n-hexadecane/water system. JCIM 53:1424-1435 DOI

Malan's Ten Rules for Air Fighting

I'm going to introduce the #SavingPharma hashtag in this post but I'm also going to introduce WW2 fighter pilot (and apartheid opponent)  Adolph Gysbert Malan (better known as Sailor).  Some of what these combat pilots had to say is still relevant and resonant today.  One of my favorites is the cricketer Keith Miller (and former Mosquito pilot) responding to an interviewer's question about the pressures of Test Cricket by noting that, "Pressure is a Messerschmitt up your arse, playing cricket is not".  Malan's Ten Rules for Air Fighting can teach those who would claim to be leading Drug Discovery and I reproduce them here:

1. Wait until you see the whites of his eyes. Fire short bursts of one to two seconds only when your sights are definitely "ON".

2. Whilst shooting think of nothing else, brace the whole of your body: have both hands on the stick: concentrate on your ring sight.

3. Always keep a sharp lookout. "Keep your finger out".

4. Height gives you the initiative.

5. Always turn and face the attack.

6. Make your decisions promptly. It is better to act quickly even though your tactics are not the best

.

7. Never fly straight and level for more than 30 seconds in the combat area.

8. When diving to attack always leave a proportion of your formation above to act as a top guard.

9. INITIATIVE, AGGRESSION, AIR DISCIPLINE, and TEAMWORK are words that MEAN something in Air Fighting.

10. Go in quickly - Punch hard - Get out

Presentations on slideshare

I recently uploaded some presentations to slideshare including my recent Gordon conference talk and the slide set used on visits to a couple of Pharmaceutical companies before the conference.  My slideshare profile is pwkenny and there are currently 13 presentations uploaded.  Be warned that there is plenty of overlap between some slide sets and it is fair to say that there are more presentations than themes.  I hope that you'll find the slide sets useful.

Some reflections on computer-aided drug design (after attending CADD Gordon conference)

I’ve just returned to Trinidad where I’ve been spending the summer.  I was in the USA at the Computer-Aided Drug Design (CADD) Gordon Conference (GRC) organized by Anthony Nicholls and Martin Stahl.  The first thing that I should say is that this will not be a conference report because what goes on at all Gordon Conferences is confidential and off-record. This is intended to make discussions freer and less inhibited and you won’t see proceedings of GRCs published.  Nevertheless, the conference program is available online so I think that it’ll be OK to share some general views of the CADD field (which have been influenced by what I picked up at the conference) even though commenting on specific lectures, posters or discussions would be verboten.

The focus of the conference was Statistics in CADD and the stuff that I took most notice of was the use of Baysian methods although I still need to get my head round things a bit more. Although modelling affinity/potency and output of ADMET assays (e.g. hERG blockade) tends to dominate the thinking of CADD scientists, the links between these properties and clinical outcomes in live humans are not as strong as many assume.  Could the wisdom of Rev Bayes be applied to PK/PD modeling and the design of clinical trials?  I couldn’t help worrying about closet Bayesians in search of a good posterior and what would be the best way to quantify the oomph of a ROC curve...

Reproducibility is something that needs to be addressed in CADD studies and if we are to improve this we must be prepared to share both data and methods (e.g. software).  This open access article should give you an idea of some of the issues and directions in which we need to head. Journal editors have a part to play here and must resist the temptation to publish retrospective analyses of large proprietary data sets because of the numbers of citations that they generate.  At the same time, journal editors should not be blamed for supplemental information ending up in PDF format.  For example, I had no problems (I just asked) getting JMC (2008), JCIM (2009) and JCAMD (2013 and 2013) to publish supplemental information in text (or zipped text) format.

When you build models from data, it is helpful to think of signal and noise.  The noise can be thought of as coming from both the model and from the data and in some cases it may be possible to resolve it into these two components.  The function of Statistics is to provide an objective measure of the relative magnitudes of signal and noise but you can’t use Statistics to make noise go away (not that this stops people from trying).  Molecular design can be defined as control of behavior of compounds and materials by manipulation of molecular properties and can be thought of as being prediction-driven or hypothesis-driven.   Prediction-driven molecular design is about building predictive models but it is worth remembering that a much (most?) pharmaceutical design involves a significant hypothesis-driven component. One way of thinking about hypothesis-driven molecular design is as framework for assembling structure activity/property relationships (SAR/SPR) as efficiently as possible but this is not something that statistical methodology currently appears equipped to do particularly well.

The conference has its own hashtag (#grccadd) and appeared to out-tweet the Sheffield Cheminformatics conference which ran concurrently.  Some speakers have shared their talks publically and a package of statistical tools created especially for the conference is available online.   

Literature cited and links to talks

WP Walters (2013) Modeling, informatics, and the quest for reproducibility. JCIM 53:1529-1530 DOI

CC Chow, Bayesian and MCMC methods for parameter estimation and model comparison. Link

N Baker, The importance of metadata in preserving and reusing scientific information Link

PW Kenny, Tales of correlation inflation.  Link

CADD Stat Link

Prediction of alkane/water partition coefficients

Those of you who follow this blog will know that I have a long standing interest in alkane/water partition coefficient and I’d like to tell you a bit about the ClogP_alk model for predicting these from molecular structure that we published during my time in Brasil. Some years ago we explored prediction of ΔlogP (logP_oct - logP_alk) from calculated molecular electrostatic potentials and this can be thought of as treating the alkane/water partition coefficient as a perturbation of the octanol/water partition coefficient.  One disadvantage of this approach is that it requires access to logP_oct and I was keen to explore other avenues.  The correlation of logP_alk with computed molecular surface area (MSA) is excellent for saturated hydrocarbons and I wondered if this class of compound might represent a suitable reference state for another type of perturbation model.  Have a look at Fig 1 which shows plots of logP_alk against MSA for saturated hydrocarbons (green), aliphatic alcohols (red) and aliphatic diols (blue).  You can see how adding a single hydroxyl group to a saturated hydrocarbon shifts logP_alk down by about 4.5 units and adding two hydroxyl groups shifts logP_alk further still.

 

The perturbations are defined substructurally using SMARTS notation. Specifically, each perturbation term consists of a SMARTS definition for the relevant functional group and a decrement term (e.g. 4.5 units for alcohol hydroxyl).  The model also allows functional groups to interact with each other.  For example, an intramolecular hydrogen bond ‘absorbs’ some of a molecule’s polarity and manifests itself as an unexpectedly high logP_alk value.  Take a look at this article if you’re interested in this sort of thing.  The interaction terms can be thought of as perturbations of perturbations. The ClogP_alk model is shown in Fig 2.

The performance of the model against external test data is shown in Figure 3.  There do appear to be some issues with some of the data and measured values of logP_alk were found to differ by two or more units for some compounds (Atropine, Propanolol, Papavarine).  Also there are concerns about the self-consistency of the measurements for Cortexolone, Cortisone and Hydrocortisone. Specifically, the logP_alk of Cortexolone (-1.00) is actually lower than that for its keto analogue Cortisone (-0.55).

 

 

The software was built using OpenEye programming toolkits (OEChem and Spicoli) and you’ll find the source code and makefiles in the supplementary information with all the data used to parameterize and test the models. It’s not completely open source because you’ll need a license from OpenEye to actually run the software.  However, the documentation for the toolkits is freely available online and you may be even able to get an evaluation license to see how things work.  You’ll also find the source code for SSProFilter in the supplemental material and this is an improved (it also profiles) version of the Filter program that I put together with the Daylight toolkit back in 1996. Very useful for designing screening libraries and you might want to take a look at this post on SMARTS from a couple of years ago.

There's some general discussion in the article that is not specific to the ClogP_alk model and I'll mention it briefly since I think this is relevant to molecular design. Those of you who believe that the octanol/water partition coefficient is somehow fundamental might like to trace how we ended up with this particular partitioning system.  We also address the question of whether logP or logD is the more appropriate measure of lipophilicity measure and some ligand efficiency stuff from an earlier post makes its journal debut.  

That’s about all I wanted to say for now and I’ll finish by noting that the manuscript was originally submitted to another journal but that's going to be the subject of a post all of its very own...

Literature cited

Toulmin, Wood, Kenny (2008) Toward prediction of alkane/water partition coefficients. J Med Chem 51:3720-3730 DOI

Kenny, Montanari, Prokopczyk (2013)ClogP_alk: A method for predicting alkane/water partition coefficient. JCAMD 27:389-402 DOI

Thermodynamics and molecular interactions

So it’s #RealTimeChem week on twitter and I thought I’d get into the spirit with a blog post.  The article that I’ve selected for review focuses on non-additivity of functional group contributions to affinity.  The protein in question is Thrombin and ligand binding was characterised using protein crystallography and isothermal titration calorimetry.  

Before reviewing the article, it’s probably a good idea to articulate my position on the thermodynamics of ligand-protein binding.  Firstly, G, H and S are three state functions, each of which can be written in terms of the other two, but only one of which is directly relevant to the binding of ligands to proteins.  K_d is no less thermodynamic than ΔH° or ΔS° and the contribution of a particular intermolecular contact to ΔG° (or ΔH° for that matter) is not in general an experimental observable.  Thermodynamics with state functions is like accountancy in that if you over-pay one interaction, the other interactions will lose out.

Now back to the featured article.  One observation presented as evidence for non-additivity is that the slopes of plots of ΔG° against hydrophobic contact area differ according to whether X (see figure below) is H or NH2.  Inhibitors with the amino group bind with greater affinity than the corresponding inhibitors lacking the amino group which interacts (in its cationic form) with the protein.   However, the difference is not constant and actually increases with hydrophobic contact area.   I would certainly agree that something interesting is going on and if we’re ever going to understand ligand-protein then combining affinity measurements for structurally-related ligands with structural information will be very useful.   

About enthalpy measurements, I am a lot less sure.  Isothermal titration calorimetry (ITC) is a direct, label-free method for measuring affinity but the fact that you get ΔH° from the experiment at no extra cost doesn’t by itself make ΔH° useful.     If measured values of ΔH° lead to improved predictions of ΔG° or provide clear insight into the nature of the interactions between ligand and protein then I certainly agree that we should make use of ΔH°.  However, there is still the problem that there is no unique way to distribute ΔG°  (or, for that matter, ΔH°) over intermolecular contacts and biomolecular recognition also takes place in aqueous media.    The cohesiveness of liquid water that drives hydrophobic association in aqueous media is a consequence of strong, cooperative hydrogen bonds between water molecules and I like to think of the hydrophobic force as a non-local, indirect, electrostatic interaction.  This non-local nature of hydrophobic interactions complicates interpretation of affinity measurements in structural terms.

Let's see what the authors have to say:

"Analysis of the individual crystal structures and factorizing the free energy into enthalpy and entropy demonstrates that the binding affinity of the ligands results from a mixture of enthalpic contributions from hydrogen bonding and hydrophobic contacts, and entropic considerations involving an increasing loss of residual mobility of the bound ligands."

I'm going to put my cards on the table and say that I believe this statement represents an exercise in arm waving.    See what you think and ask yourself the question as to whether this statement would help you design a higher affinity Thrombin inhibitor?  It's also worth thinking carefully about the relationship between mobility and entropy.   One way of looking at entropy is as the degree to which systems are constrained and a more highly constrained system will be less mobile.    The authors state:

"The present study shows, by use of crystal structure analysis and isothermal titration calorimetry for a congeneric series of thrombin inhibitors, that extensive cooperative effects between hydrophobic contacts and hydrogen bond formation are intimately coupled via dynamic properties of the formed complexes."

I think that one needs to be very careful when talking about 'dynamic properties' in the context of equilibrium thermodynamics.  Ultimately entropy is determined by the characteristics of potential energy surfaces and Statistical Mechanics tells us that entropy (and other thermodynamic properties) can be calculated from the partition function.  It may be instructive to think how one would use the partition function to put these 'dynamic properties' on a quantitative basis.

I'm now going to change direction because there is one factor that the authors appear not to have considered  and I think that it could be quite important.  The ligand amino group (see figure above; it's protonated under assay conditions) interacts with the protein but it can also affect affinity in another way.  Have a look a this figure (showing the stricture of complex with 3e) from the article which shows how one of the inhibitors lacking the amino group binds.  Now I'd like you to look at one of the dihedral angles in the ligand structure and you can see how it is defined by looking at the substructures inset in the histograms below.  

 

The histograms show the distributions of (the absolute value of) this dihedral angle observed for the instances of substructure in the CSD. The histogram on the left suggests that the dihedral angle will tend to be 180° when the carbon next to the carbonyl carbon has two attached hydrogens. The histogram on the right shows how a substituent (X ¹ H) on that carbon shifts the distribution of dihedral angles.  Now let's go back to the structure of complex 3e, which can be downloaded from the PDB as refcode 2ZIQ, and we can see that the relevant dihedral angle is 117°.  Comparing the two histograms tells us is that a substituent (such as an amino group) on the carbon next to the carbonyl group will tend to stabilise the bound conformations of these inhibitors in addition to making direct interactions with the protein.  Do the ITC results tell us this?       

Literature cited

Baum, Muley, Smolinski, Heine, Hangauer and Klebe (2010) Non-additivity of functional group contributions by protein-ligand binding: A comprehensive study by crystallography and isothermal titration calorimetry. J Mol Biol 397:1042-1054  doi

Unknown knowns and binding kinetics

I hope that you all enjoyed the April Fool post but now it’s time to get back to business.  Before returning to Thermodynamics, I’d like to point you towards a couple of articles that you might find to be of interest.  In the first of these, Ben Davis and Practical Fragments’ Dan Erlanson take a look at some of what can go wrong when you screen fragments and it is noteworthy that they actively sought material from the fragment community before writing the review.  If you’re thinking about running a fragment screen, you do need to read this article which has already been featured in other blog posts ( 1 | 2 | 3).    

The article by Göran Dahl and Tomas Åkerud focuses on binding kinetics, specifically in the context of pharmacokinetics. In multistep processes, timescale determines relevance.  My take is that slow binding is equivalent to slow distribution and, at some point when the topic is raised, I will usually get round to asking whether slow distribution is something that you would want to design into a drug. The authors conclude that ‘the drug-target residence time appears to be limited in scope for prolonging duration of effect’ and the article provides a timely reminder that drugs act within a pharmacokinetic framework.

Literature cited

Davis & Erlanson (2013) Learning from our mistakes: The ‘unknown knowns’ in fragment screening. Bioorg Med Chem Lett In press DOI

Dahl & Åkerud (2013) Pharmacokinetics and the drug-target residence time concept. Drug Discov Today In press DOI

Isotopic enrichment by recrystallisation from water

Isotopic enrichment doesn’t normally fall under the molecular design remit.  However, a radically new approach has been recently discovered and predictive modelling was very important in showing that the method would indeed be feasible.  Isotopic enrichment is typically carried out in gas phase and if the elemental form is involatile it is necessary to prepare a compound which can be more readily persuaded into the gas phase. So if you want to enrich uranium you first make the relatively volatile hexafluoride (fluorine is good because it’s only got one natural isotope) and then stick it in a centrifuge and away you go.

The isotopic enrichment industry has been watching developments in aqueous solubility prediction with great interest.  As you’ll know aqueous solubility is a very important physicochemical property and a key determinant of the efficiency with which a drug is absorbed from the gastrointestinal tract.  By focussing on aqueous solubility you can get away from all the technological demands of working in gas phase.  You can do this in your garage.

Most pharmaceutical researchers are familiar with the idea that aqueous solublility decreases with molecular weight and there is large literature of many seminal articles on the subject.  One consequence of this anti-correlation is that the average solubility for compounds above a particular molecular weight cut off will always be lower than the average solubility for the compounds with molecular weights below the cut off.  The selection of the optimal cut off is a challenging problem and it was believed that the only option available was to design compounds such that the masses of two isotopic forms straddled the value of the molecular weight cut off.  Only recently has the seminal 1940 study of Bronshtein and Mercader, the result of a short collaboration in Mexico City, come to light.  In a nutshell, their study unequivocally demonstrated that the cut off could be set to the centroid of the isotopic masses for any compound.  The math in the article is truly formidable but the essence of the method is that diagonalisation of the reduced mass tensor is NKVD-complete.  Thus the cut off can be matched to compounds which is much easier than having to design compounds to the bracket the cut off.

Literature cited

Bronshtein and Mercader, Мисс Скарлетт с ледорубом в исследовании.  Dokl Akad Nauk 1940, 7, 432-456 doi

Efficient Voodoo Thermodynamics

So you’ve got a little data analysis problem.  You have some compounds with a range of IC₅₀ values and you’d like to explore the extent that molecular size contributes to potency/affinity.  Here’s one suggestion for starters.  Plot pIC₅₀ against your favourite measure of  molecular size which could be number of number of non-hydrogen atoms, molecular weight and look at the residuals which tell you how much each compound beats the trend (or is beaten by it).  Let’s start by defining  pIC₅₀ which you can calculate from:

 pIC₅₀ = -log(IC₅₀/M)                                                                              1

You might be asking why I’m not suggesting that you use the standard Gibbs free energy of binding which is defined by:

ΔG° = RTln(K_d/C°)                                                                                 2

The main reason for not doing this is that we can’t.  You’ll notice that ΔG° is calculated from K_d and not  IC₅₀ and the two are not the same thing even though they both have units of concentration.  Those of you who have worked on kinase projects may have even used IC₅₀ values measured at different ATP concentrations to get a better idea how much kick your inhibitors will have at physiological ATP concentration.  Put another way, you can measure the concentration of sugar in your coffee and plug this into equation 2 but that does not make what you calculate a standard Gibbs free energy of binding.   If you’ve measured IC₅₀ then you really should use  pIC₅₀ in this analysis.  Converting pIC₅₀ to ΔG° is technically incorrect and arguably pretentious since the converted pIC₅₀ can give the impression that it is somehow more thermodynamic than that from which it was calculated.  Converting pIC₅₀  to ΔG° also introduces additional units (of energy/mole) and there is always a degree of irony when these units, which may have been introduced to just make biological data look more physical, get lost when the results are presented.

So let’s get back to the data analysis problem.  Suppose that we’ve plotted  pIC₅₀ against number of heavy (i.e. non-hydrogen) atoms and the next step is to fit the data.  Best way to start is to fit a straight line although you could also fit a curve if the data justifies this.  Let’s assume that we’re fitting the straight line:

  pIC₅₀ = A + B×N_HA                                                                            3

I realise that using an intercept term (A) will cause a few eyebrows to become raised.  Surely the line of fit should go through the origin?   There is a problem with this line of thinking and it’s helpful now to talk instead in terms of affinity and ΔG° to develop the point a bit more.  You might be thinking that in the limit of zero molecular size a compound should have zero free energy of binding.  However, a zero free energy of binding corresponds to K_d being equal to the standard concentration and you’ll remember that the choice of standard state is arbitrary.  If you must derive insights from thermodynamic measurements then the very least that you can do is to ensure that any insights you derive are invariant with respect to the value of the standard concentration.  

When you use ligand efficiency (-ΔG°/N_HA) you’re effectively assuming that the value of ΔG° should be directly proportional to the number of heavy atoms in the ligand molecule.  One consequence of defining ligand efficiency in this manner is that relative values of ligand efficiency for compounds with different numbers of heavy atoms will change if you change (as thermodynamics tells us that we are allowed to do) the standard concentration used to define ΔG°.  I've droned on enough though and it's time to check out.  I will however, leave you with the question of whether it makes sense to try to correct ligand efficiency for the effects of molecular size. 

The wrong kind of free energy

It has been some time since I last blogged.   There has been the distraction of a little drug-likeness project and there is still much to do before I leave Brasil at the end of May.  I have, however, set up a twitter account (@pwk2013) which I use in futile attempts to engage minor celebrities (like God and Richard Dawkins) in conversation.   I’ll be taking a look at some aspects of thermodynamics in this post and I’ll start by writing the relationship between dissociation constant and the standard free energy of binding:

ΔG° = RTln(K_d/C°)

I realise that this might look a bit different to what you’re used to so I’ll try to explain why its been written like this.  First of all K_d has units of concentration and logarithms are only defined for numbers (i.e. unitless quantities) so you might want to consider the possibility that what you normally write is wrong (apologies for hideous pun).   The other thing that you need to remember is that the standard free energy of binding depends on the choice of standard state and writing the equation like this makes this connection explicit.   ΔG° is negative for a 10nM compound when the standard concentration is 1M.  However, if we were to define the standard concentration to be 1nM then   ΔG° would be positive.  If you consider the idea of a 1nM standard concentration to be offensive, think of a 1M solution of your favourite protein...

One of the misconceptions that drug discovery researchers tend to have is that you can’t call it thermodynamics unless you measure both enthalpy and entropy.  One reason for this is the use of isothermal titration calorimetry (ITC) to measure affinity.  ITC is a direct, label-free method for measuring affinity and you get the enthalpy as part of the package.  One of the important things to remember about thermodynamics is that you need to use the most appropriate thermodynamic quantity to describe the phenomenon that you’re interested in and for binding at constant pressure this is the Gibbs free energy.  In contrast, the engineer scaling up a synthetic reaction for manufacturing a drug will usually be more interested in the heat and volume (i.e. gaseous bi-products) generated by the reaction because failure to control these can result in the Big Kaboom.

One idea that has emerged as ITC has become more accessible in drug discovery is that enthalpy-driven binding is somehow better than entropy-driven binding.  I remain sceptical and ask the rhetorical question of how an isothermal system such as a human taking a drug senses the degree to which engagement of the drug’s target(s) is driven by enthalpy changes.   If measuring enthalpy of binding in addition to affinity helps us to predict affinity for compounds yet to be synthesised or gives us clear (i.e. keeping arms stationary) insights into the nature of binding then it makes sense to do it.  However, I’m not seeing this being done convincingly and sometimes wonder whether enthalpy optimisation is just a case of the wrong kind of snow...

We tend to interpret affinity in terms of contacts between protein and ligand although one must always remember that the contribution of a particular contact to affinity is not in general an experimental observable.   Proteins and their ligands associate in an aqueous environment and the non-local nature of the hydrophobic effect further complicates attempts to use structure to rationalise affinity.  I often hear hydrophobic interactions being described as non-directional and in my view this is complete bollocks since interactions are forces and forces are vectors which by definition have direction.  On this subject, I can't resist telling the tale of the Human Resources department getting the Scalar Quantity Award for having magnitude but no direction...

I think this is a good place to leave things and I'll try to be back in a week or some more specific stuff in a follow up to this post.