Wednesday, August 29, 2018

A Density Functional Tight Binding Layer for Deep Learning of Chemical Hamiltonians

Haichen Li, Christopher Collins, Matteus Tanha, Geoffrey J. Gordon, David J. Yaron (2018)
Highlighted by Jan Jensen


There are increasingly many papers on predicting the molecular energy and other properties using machine learning (ML). Most, if not all, use some similarity measure of the molecular structure to structures in the training set when training. This paper uses DFTB Hamiltonian matrix elements instead and treats the short-range matrix elements as adjustable parameters (weights) to be trained. To make this happen, DFTB is implemented as a layer for deep learning, using the TensorFlow deep learning framework, by recasting the DFTB equations in terms of tensor operations. In this way domain knowledge is incorporated into the ML model. Since the starting values are the "conventional" DFTB parameters one can also view this as refining the DFTB method.

This DFTB-ML approach is evaluated on 15,700 hydrocarbons by comparing the RMSE in energy per heavy atom (Eatom) relative to ωB97X/6-31G(d) reference values. Training on up to 7 heavy atoms and testing on 8 heavy atoms, leads to RMS errors in Eatom of 0.72 kcal/mol, compared to 1.80 using conventional DFTB. Training on up to 4 heavy atoms gives an Eatom RMSE of 1.08 kcal/mol. The results can be further improved by using neural networks to allow the matrix elements to depend on the molecular environment of the atoms.

As the authors point out the performance on the training data remained above chemical accuracy (0.5 kcal/mol) for the total molecular energy, but they offer several interesting ideas on how to improve the performance.


This work is licensed under a Creative Commons Attribution 4.0 International License.

Nano-Saturn: Experimental Evidence of Complex Formation of an Anthracene Cyclic Ring with C60

Yuta, Y.; Eiji, T.; Kan, W.; Shinji, T., Angew. Chem. Int. Ed. 2018, 57, 8199-8202
Contributed by Steven Bacharach
Reposted from Computational Organic Chemistry with permission

It never hurts to promote one’s science through clever names – think cubane, buckminsterfullerene, bullvalene, etc. Host-guest chemistry is not immune to this cliché too, and this post discusses the latest synthesis and computations of a nano-Saturn; nano-Saturns are a spherical guest molecule captured inside a ring host molecule. I discussed an example of this a number of years ago – the nano-Saturn comprised of C60 fullerene surrounded by [10]cycloparaphenylene.

Yamamoto, Tsurumaki, Wakamatsu, and Toyota have prepared a nano-Saturn complex with the goal of making a flatter ring component.1 The inner planet is modeled again by C60 and the ring is the [24]circulene analogue 1. The x-ray crystal structure of this nano-Saturn complex is shown in Figure 1.

1: R = 2,4,6-tri-iso-propylphenyl
2: R = H
Figure 1. X-ray crystal structure of the nano-Saturn complex of 1 with C60.

Variable temperature NMR experiments gave the binding values of ΔH = -18.1 ± 2.3 kJ mol-1 and TΔS = 0.8 ± 2.2 kJ mol-1 at 298 K. To gauge this binding energy, they computed the complex of C60 with the parent compound 2 at B3LYP/6-1G(d)//M05-2X/6-31G(d), unfortunately without publishing the coordinates in the supporting materials. The computed binding enthalpy is ΔH = -50.6 kJ mol-1, but this computation is for the gas phase. The computed structure shows close contacts of 0.29 nm between the fullerene and the C9-proton of the anthracenyl groups, in excellent agreement with the x-ray structure. These weak C-Hπ interactions undoubtedly help stabilize the complex, especially given that the fullerene carries a very tiny Mulliken charge of +0.08 e.

References

1) Yuta, Y.; Eiji, T.; Kan, W.; Shinji, T., "Nano-Saturn: Experimental Evidence of Complex Formation of an Anthracene Cyclic Ring with C60." Angew. Chem. Int. Ed. 2018, 57, 8199-8202, DOI: 10.1002/anie.201804430.

InChIs

1: InChI=1S/C174H180/c1-91(2)121-78-150(97(13)14)164(151(79-121)98(15)16)163-90-128-71-139-70-127-59-109(37-38-120(127)77-162(139)163)110-39-49-140-129(60-110)72-130-61-111(40-50-141(130)165(140)170-152(99(17)18)80-122(92(3)4)81-153(170)100(19)20)112-41-51-142-131(62-112)73-132-63-113(42-52-143(132)166(142)171-154(101(21)22)82-123(93(5)6)83-155(171)102(23)24)114-43-53-144-133(64-114)74-134-65-115(44-54-145(134)167(144)172-156(103(25)26)84-124(94(7)8)85-157(172)104(27)28)116-45-55-146-135(66-116)75-136-67-117(46-56-147(136)168(146)173-158(105(29)30)86-125(95(9)10)87-159(173)106(31)32)118-47-57-148-137(68-118)76-138-69-119(128)48-58-149(138)169(148)174-160(107(33)34)88-126(96(11)12)89-161(174)108(35)36/h37-108H,1-36H3
InChIKey=AMDNULXMAMDTMX-UHFFFAOYSA-N
2: InChI=1S/C84H48/c1-13-61-25-62-15-3-51-33-75(62)43-73(61)31-49(1)50-2-14-63-26-64-16-4-52(34-76(64)44-74(63)32-50)54-6-18-66-28-68-20-8-56(38-80(68)46-78(66)36-54)58-10-22-70-30-72-24-12-60(42-84(72)48-82(70)40-58)59-11-23-71-29-69-21-9-57(39-81(69)47-83(71)41-59)55-7-19-67-27-65-17-5-53(51)35-77(65)45-79(67)37-55/h1-48H
InChIKey=ZYXXLAYETADMDM-UHFFFAOYSA-N


'
This work is licensed under a Creative Commons Attribution-NoDerivs 3.0 Unported License.

Thursday, August 16, 2018

Readily Accessible Ambiphilic Cyclopentadienes for Bioorthogonal Labeling

Levandowski, B. J.; Gamache, R. F.; Murphy, J. M.; Houk, K. N., J. Am. Chem. Soc. 2018, 140, 6426-6431
Contributed by Steven Bacharach
Reposted from Computational Organic Chemistry with permission

I recently posted on a paper proposing 1,2-benzoquinone and related compounds as the diene component for bioorthogonal labeling. Levandowski, Gamache, Murphy, and Houk report on tetrachlorocyclopentadiene ketal 1 as an active ambiphilic diene component.1
1 is sterically congested to diminish self-dimerization and will react with both electron-rich and electron-poor dienes. To test it as an active diene in bioorthogonal labeling applications, they optimized the structures of the transition states at CPCM(water)/M06-2X/6-311+G(d,p)//CPCM(water)/M06-2X/6-31G(d) for the Diels-Alder reaction of 1 with a variety of dienophiles including trans-cyclooctene 2 and endo-bicyclononyne 3. These transition states are shown in Figure 1. The activation free energy is quite low for each: 18.1 kcal mol-1 with 2 and 18.9 kcal mol-1 with 3.

TS(1+2)

TS(1+3)
Figure 1. CPCM(water)/M06-2X/6-31G(d) optimized geometries for the TSs of the reaction of 1 with 2and 3.

Experiments were successfully run using 1 as a label on a neuropeptide.

References

1) Levandowski, B. J.; Gamache, R. F.; Murphy, J. M.; Houk, K. N., "Readily Accessible Ambiphilic Cyclopentadienes for Bioorthogonal Labeling." J. Am. Chem. Soc. 2018140, 6426-6431, DOI: 10.1021/jacs.8b02978.

InChIs

1:InChI=1S/C7H4Cl4O2/c8-3-4(9)6(11)7(5(3)10)12-1-2-13-7/h1-2H2
InChIkey=DXQQKKGWMVTLOJ-UHFFFAOYSA-N



'
This work is licensed under a Creative Commons Attribution-NoDerivs 3.0 Unported License.