Molecular Orbitals (MOs) are linear combinations of Atomic Orbitals (AOs), which in turn are linear combinations of other functions called ‘basis functions’. A basis, or more accurately a basis set, is a collection of functions which obey a set of rules (such as being orthogonal to each other and possibly being normalized) with which all AOs are constructed, and although these are centered on each atomic nucleus, the canonical way in which they are combined yield delocalized MOs; in other words, an MO can occupy a large space spanning several atoms at once. We don’t mind this expansion across a molecule, but what about between two molecules? Calculating the interaction energy between two or more molecular fragments leads to an artificial extra–stabilization term that stems from the fact that electrons in molecule 1 can occupy AO’s (or the basis functions which form them) centered on atoms from molecule 2.
Fundamentally, the interaction energy of any A—B dimer, Eint, is calculated as the energy difference between the dimer and the separately calculated energies for each component (Equation 1).
Eint = EAB – EA – EB (1)
However the calculation of Eint by this method is highly sensitive to the choice of basis set due to the Basis Set Superposition Error (BSSE) described in the first paragraph. The BSSE is particularly troublesome when small basis sets are used, due to the poor description of dispersion interactions but treating this error by just choosing a larger basis set is seldom useful for systems of considerable sizes. The Counterpoise method is a nifty correction to equation 1, in which EA and EB are calculated with the basis set of A and B respectively, i.e., only in EAB a larger basis set (that of A and B simultaneously) is used. The Counterpoise method calculates each component with the AB basis set (Equation 2)
EintCP = EABAB – EAAB– EBAB (2)
where the superscript AB means the whole basis set is used. This is accomplished by using ‘ghost‘ atoms with no nuclei and no electrons but empty basis set functions centered on them.
In Gaussian, BSSE is calculated with the Counterpoise method developed by Boys and Simon. It requires the keyword Counterpoise=N where N is the number of fragments to be considered (for an A—B system, N=2). Each atom in the coordinates list must be specified to which fragment pertains; additionally, the charge and multiplicity for each fragment and the whole supermolecular ensemble must be specified. Follow the example of this hydrogen fluoride dimer.
%chk=HF2.chk #P opt wB97XD/6-31G(d,p) Counterpoise=2 HF dimer 0,1 0,1 0,1 H(Fragment=1) 0.00 0.00 0.00 F(Fragment=1) 0.00 0.00 0.70 H(Fragment=2) 0.00 0.00 1.00 F(Fragment=2) 0.00 0.00 1.70
For closed shell fragments the first line is straightforward but one must pay attention that the first pair of numbers in the charge multiplicity line correspond to the whole ensemble, whereas the folowing pairs correspond to each fragment in consecutive order. Fragments do not need to be specified contiguously, i.e., you don’t need to define all atoms for fragment 1 and after those the atoms for fragment 2, etc. They could be mixed and the program still assigns them correctly. Just as an example I typed wB97XD but any other method, DFT or ab initio, may be used; only semiempirical methods do not admit a BSSE calculation because they don’t make use of a basis set in the first place!
The output provides the corrected energy (in atomic units) for the whole system, as well as the BSSE correction (which added to the previous term yields the un-corrected energy of the system). Gaussian16 also provides these values in kcal/mol as ‘Complexation energies’ first raw (uncorrected) and then the corrected energy.
BSSE is always present and cannot be entirely eliminated because of the use of finite basis sets but it can be correctly dealt with if the Counterpoise method is included.
We’ve expanded the scope of our research interests from quantum mechanical calculations to docking and MedChem for over a year now; it has been a very interesting ride and a very rich avenue of research to explore. Durbis Castillo has led -out of his own initiative- this project and today he presents us with a guest post on the nuances of his project. Bear in mind that the detail of the calculations and a small -very targeted- tutorial on MAESTRO will be provided later in further posts and that making all this decisions required a long process of trial and error, we can only thank Dr. Antonio Romo for his help in minimizing the time this process took.
HIV is a tricky virus, and even though many of the steps included in its lifecycle are druggable, the chemical machinery making it work has been quite elusive since research groups started studying it. Highly Active Antiretroviral Therapy (HAART) works thanks to the combination of several drugs targeting different proteins such as the HIV protease or reverse transcriptase.
In 1998 the elucidation of the gp120 envelope glycoprotein crystal structure introduced a new step in the drug discovery race: HIV entry. Since drugs targeting gp120 have not been widely explored or developed, we decided to use common methodologies like docking (rigid and fit-induced) and ADME predictions to address the following question: How can we easily discover a molecule that inhibits gp120 binding to the lymphocyte CD4 receptor without having to synthesize it first? The answer was to perform a virtual screening with a bottleneck methodology based on docking calculations.
Docking methodologies are often looked as insufficient, careless or even unscientific, since the algorithms they are founded upon are not as accurate or descriptive as the ones that support DFT or ab initio calculations, for example. But there is a huge advantage to simpler operations: less computational resources are required. Then, following Russia’s example when making tanks during the WWII, why not make thousands or millions of docking calculations to quickly explore an entire chemical space and find which molecules are more likely to bind the protein?
And this is exactly what we did. We built a piperazine-based dataset of 16.3 million compounds, all of them including fragments that are reported in the medicinal chemistry literature, thus having two main characteristics, synthetic accessibility and pharmacological activity. These 16.3 million compounds were thoroughly filtered through several docking steps, each one of them being more accurate and comprehensive than the previous one, abruptly eliminating poorly fitted molecules, leaving us with a total of 275 candidates that were redocked in a different crystal structure and a different program (consensus docking).
After analyzing the ADME properties of the candidates, with descriptors such as human oral absorption and possible metabolic reactions, as well as the Induced-Fit Docking score of these molecules, ten ligands were selected as the best ones inside the analyzed chemical space. You can see ligand 255 (figure 1) as an example of the molecules that obtained the best scores throughout the docking steps.
Many of the colleague researchers related to this kind of topics asked “Why didn’t you download a set of molecules from Zinc or Maybridge?” And the answer to this question includes three aspects: first we wanted to test a combinatorial approach to drug design, second, we wanted to test whether including a piperazine as the core of the set of molecules would immediately grant them activity and high potency, and finally, a built database will always confer a higher degree of novelty to the possible hits when compared to commercially available compounds whose synthesis has already been developed. However, this last point needs to be addressed by an organic chemist since none of the molecules from our database have ever been synthesized (any takers?).
Right now, we are trying to explore further through molecular dynamics simulations using Desmond and Amber. Other future goals for this project include screening large databases of commercial and novel compounds with gp120 and other proteins involved in the HIV lifecycle. Also, we remain open to collaborate with anyone interested in taking the challenge to synthesize our molecules, as well as performing the biochemical assays to get an idea of their activity.
More details on MD simulations and the path of our first virtual hits to follow. Anyone interested in reading my thesis work can contact me through my linkedin profile at https://www.linkedin.com/in/durbisjaviercp/. An article is under preparation and will soon be submitted, stay tuned!
2017 was a complicated year for various reasons here in Mexico (and some personal health issues) but nonetheless I’m very proud of the performance of everyone at the lab whose hard work and great skills keep pushing our research forward.
Four new members joined the team and have presented their work at the national meeting for CompChem for the first time. Also, for the first time, one of my students, Gustavo Mondragón, gave a talk at this meeting with great success about his research on the Fenna Matthews Olson complex of photosynthetic bacteria.
The opportunity to attend WATOC at Munich presented me the great chance to meet wonderful people from around the world and was even kindly and undeservingly invited to write the prologue for an introductory DFT book by Prof. Pedro Cerón from Spain. I hope to Jeep up with the collaborations abroad such as the one with the Mirkin group at Nortgwestern and the one with my dear friend Kunsagi-Mate Sándor at Pecsi Tudomanyegyetem (Hungary), among many others; I’m thankful for their trust in our capabilities.
Two members got their BSc degrees, Marco an Durbis, the latter also single handedly paved the way for us to develop a new research line on the in silico drug developing front; his relentless work has also been praised by the QSAR team at the Institute of Chemistry with which he has collaborated by performing toxicity calculations for the agrochemical industry as well as by designing educational courses aimed to the dissemination of our work and QSAR in general among regulatory offices and potential clients. We’re sad to see him go next fall but at the same time we’re glad to know his scientific skills will further develop.
I cannot thank the team enough: Alejandra Barrera, Gustavo Mondragón, Durbis Castillo, Fernando Uribe, Juan Guzman, Alberto Olmedo, Eduardo Cruz, Ricardo Loaiza and Marco Garcia; may 2018 be a great year for all of you.
And to all the readers thank you for your kind words, I’m glad this little space which is about to become nine years old is regarded as useful; to all of you I wish a great 2018!
One of the most popular posts in this blog has to do with calculating Fukui indexes, however, when dealing with a large number of molecules, our described methodology can become cumbersome since it requires to manually extract the population analysis from two or three different output files and then performing the arithmetic on them separately with a spreadsheet or something.
Our new team member Ricardo Loaiza has written a python script that takes the three aforementioned files and yields a .csv file with the calculated Fukui indexes, and it even points out which of the atoms exhibit the largest values so if you have a large molecule you don’t have to manually check for them. We have also a batch version which takes all the files in any given directory and performs the Fukui calculations for each, provided it can find file triads with the naming requirements described below.
Output files must be named filename.log (the N electrons reference state), filename_plus.log (the state with N+1 electrons) and filename_minus.log (the N-1 electrons state). Another restriction is that so far these scripts only work with NBO population analysis as provided by the NBO3.1 program available in the various versions of Gaussian. I imagine the listing is similar in NBO5.x and NBO6.x and so it should work if you do the population analysis with them.
The syntax for the single molecule version is:
python fukui.py filename.log filename_minus.log filename_plus.log
For the batch version is:
(Por Lote means In Batch in Spanish.)
These scripts are available via GitHub. We hope you find them useful, and you do please let us know whether here at the comments section or at our GitHub site.
If you work in the field of photovoltaics or polyacene photochemistry, then you are probably aware of the Singlet Fission (SF) phenomenon. SF can be broadly described as the process where an excited singlet state decays to a couple of degenerate coupled triplet states (via a multiexcitonic state) with roughly half the energy of the original singlet state, which in principle could be centered in two neighboring molecules; this generates two holes with a single photon, i.e. twice the current albeit at half the voltage (Fig 1).
It could also be viewed as the inverse process to triplet-triplet annihilation. An important requirement for SF is that the two triplets to which the singlet decays must be coupled in a 1(TT) state, otherwise the process is spin-forbidden. Unfortunately (from a computational perspective) this also means that the 3(TT) and 5(TT) states are present and should be taken into account, and when it comes to chlorophyll derivatives the task quickly scales.
SF has been observed in polyacenes but so far the only photosynthetic pigments that have proven to exhibit SF are some carotene derivatives; so what about chlorophyll derivatives? For a -very- long time now, we have explored the possibility of finding a naturally-occurring, chlorophyll-based, photosynthetic system in which SF could be possible.
But first things first; The methodology: It was soon enough clear, from María Eugenia Sandoval’s MSc thesis, that TD-DFT wasn’t going to be enough to capture the whole description of the coupled states which give rise to SF. It was then that we started our collaboration with SF expert, Prof. David Casanova from the Basque Country University at Donostia, who suggested the use of Restricted Active Space – Spin Flip in order to account properly for the spin change during decay of the singlet excited state. A set of optimized bacteriochlorophyll-a molecules (BChl-a) were oriented ad-hoc so their Qy transition dipole moments were either parallel or perpendicular; the rate to which SF could be in principle present yielded that both molecules should be in a parallel Qy dipole moments configuration. When translated to a naturally-occurring system we sought in two systems: The Fenna-Matthews-Olson complex (FMO) containing 7 BChl-a molecules and a chlorosome from a mutant photosynthetic bacteria made up of 600 Bchl-d molecules (Fig 2). The FMO complex is a trimeric pigment-protein complex which lies between the antennae complex and the reaction center in green sulfur dependent photosynthetic bacteria such as P. aestuarii or C. tepidium, serving thus as a molecular wire in which is known that the excitonic transfer occurs with quantum coherence, i.e. virtually no energy loss which led us to believe SF could be an operating mechanism. So far it seems it is not present. However, for a crystallographic BChl-d dimer present in the chlorosome it could actually occur even when in competition with fluorescence.
I will keep on blogging more -numerical and computational- details about these results and hopefully about its publication but for now I will wrap this post by giving credit where credit is due: This whole project has been tackled by our former lab member María Eugenia “Maru” Sandoval and Gustavo Mondragón. Finally, after much struggle, we are presenting our results at WATOC 2017 next week on Monday 28th at poster session 01 (PO1-296), so please stop by to say hi and comment on our work so we can improve it and bring it home!
Out of some +1000 twitter accounts I follow about a quarter are related computational chemistry. The following public list isn’t comprehensive and prone to errors and contains researchers, programmers, students, journals, products and companies who gravitate around the use of in silico methods for the understanding and design of chemical and biochemical compounds.
As is the case of proteins, the functioning of DNA is highly dependent on its 3D structure and not just only on its sequence but the difference is that protein tertiary structure has an enormous variety whereas DNA is (almost) always a double helix with little variations. The canonical base pairs AT, CG stabilize the famous double helix but the same cannot be guaranteed when non-canonical -unnatural- base pairs (UBPs) are introduced.
When I first took a look at Romesberg’s UBPS, d5SICS and dNaM (throughout the study referred to as X and Y see Fig.1) it was evident that they could not form hydrogen bonds, in the end they’re substituted naphtalenes with no discernible ways of creating a synton like their natural counterparts. That’s when I called Dr. Rodrigo Galindo at Utah University who is one of the developers of the AMBER code and who is very knowledgeable on matters of DNA structure and dynamics; he immediately got on board and soon enough we were launching molecular dynamics simulations and quantum mechanical calculations. That was more than two years ago.
Our latest paper in Phys.Chem.Chem.Phys. deals with the dynamical and structural stability of a DNA strand in which Romesberg’s UBPs are introduced sequentially one pair at a time into Dickerson’s dodecamer (a palindromic sequence) from the Protein Data Bank. Therein d5SICS-dNaM pair were inserted right in the middle forming a trisdecamer; as expected, +10 microseconds molecular dynamics simulations exhibited the same stability as the control dodecamer (Fig.2 left). We didn’t need to go far enough into the substitutions to get the double helix to go awry within a couple of microseconds: Three non-consecutive inclusions of UBPs were enough to get a less regular structure (Fig. 2 right); with five, a globular structure was obtained for which is not possible to get a proper average of the most populated structures.
X and Y don’t form hydrogen bonds so the pairing is pretty much forced by the scaffold of the rest of the DNA’s double helix. There are some controversies as to how X and Y fit together, whether they overlap or just wedge between each other and according to our results, the pairing suggests that a C1-C1′ distance of 11 Å is most stable consistent with the wedging conformation. Still much work is needed to understand the pairing between X and Y and even more so to get a pair of useful UBPs. More papers on this topic in the near future.
As if I didn’t have enough things to do I’m launching a new blog inspired by the #365papers hashtag on Twitter and the naturalproductman.wordpress.com blog. In it I’ll hopefully list, write a femto-review of all the papers I read. This new effort is even more daunting than the actual reading of the huge digital pile of papers I have in my Mendeley To-Be-Read folder, the fattest of them all. The papers therein wont be a comprehensive review of Comp.Chem. must-read papers but rather papers relevant to our lab’s research or curiosity.
Maybe I’ll include some papers brought to my attention by the group and they could do the review. The whole endeavor might flop in a few weeks but I want to give it a shot; we’ll see how it mutates and if it survives or not. So far I haven’t managed to review all papers read but maybe this post will prompt to do so if only to save some face. The domain of the new blog is compchemdigest.wordpress.com but I think it should have included the word MY at the beginning so as to convey the idea that it is only my own biased reading list. Anyway, if you’re interested share it and subscribe, those post will not be publicized.
Sometimes you just need to optimize some fragment or moiety of your molecule for a number of reasons -whether because of its size, your current interest, or to skew the progress of a previous optimization- or maybe you want just some kind of atoms to have their positions optimized. I usually optimize hydrogen atoms when working with crystallographic files but that for some reason I want to preserve the rest of the molecule as refined, in order to keep it under a crystalline field of sorts.
Asking Gaussian to optimize some of the atoms in your molecule requires you to make a list albeit the logic behind it is not quite straightforward to me. This list is invoked by the ReadOptimize keyword in the route section and it includes all atoms by default, you can then further tell G09 which atoms are to be included or excluded from the optimization.
So, for example you want to optimize all atoms EXCEPT hydrogens, then your input should bear the ReadOptimize keyword in the route section and then, at the end of the molecule specification, the following line:
If you wish to selectively add some atoms to the list while excluding others, here’s an example:
atoms=C H S notatoms=5-8
This list adds, and therefore optimizes, all carbon, hydrogen and sulfur atoms, except atoms 5, 6, 7 and 8, should they be any of the previous elements in the C H S list.
The way I selectively optimize hydrogen atoms is by erasing all atoms from the list -using the noatoms instruction- and then selecting which are to be included in the list -with atoms=H-, but I haven’t tried it with only selecting hydrogen atoms from the start, as in atoms=H
I probably get very confused because I learned to do this with the now obsolete ReadFreeze keyword; now it sometimes may seem to me like I’m using double negatives or something – please do not optimize all atoms except if they are hydrogen atoms. You can include numbers, ranks or symbols in this list as a final line of your input file.
Common errors (by common I mean I’ve got them):
Lets look at the end of an input I just was working with:
> AtmSel: Line=”P 0″
> Maximum list size exceeded in AddBin.
> Error termination via Lnk1e in…
AtmSel is the routine which reads the atoms list and I was using a pseudopotential on phosphorous atoms, I placed the atoms list at the end of the file but it should be placed right after the coordinates and the connectivity matrix, should there be one, and thus before any external basis set or pseudopotential or any other specification to be read by Gaussian.
As a sort of test you can use the instruction:
%kjob l103 %chk=myfile.chk ...
at the Link0 section (where your checkpoint is defined). This will kill the job after the link 103 is finished, thus you will only get a list of what parameters were frozen and which were active. Then, if things look ok, you can run the job without the %kjob l103 instruction and get it done.
As usual I hope this helps. Thanks for reading except to those who didn’t read it except for the parts they did read.