tag:blogger.com,1999:blog-5744790168115177005.comments2024-03-02T07:26:51.612+01:00Proteins and Wave FunctionsJan Jensenhttp://www.blogger.com/profile/08595894308946022740noreply@blogger.comBlogger171125tag:blogger.com,1999:blog-5744790168115177005.post-3227908912474390412023-01-12T15:44:21.235+01:002023-01-12T15:44:21.235+01:00Fixed. Thanks Graham. Fixed. Thanks Graham. Jan Jensenhttps://www.blogger.com/profile/08595894308946022740noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-33706751038494152812023-01-11T13:26:36.523+01:002023-01-11T13:26:36.523+01:00Hi Jan, just a note to say that the APC for PeerJ ...Hi Jan, just a note to say that the APC for PeerJ Chemistry journals are $1195<br /><br />Cheers,<br /><br />Graham<br /><br />(Graham Smeddle, PeerJ)Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-80667494985832198952023-01-11T13:25:19.793+01:002023-01-11T13:25:19.793+01:00This comment has been removed by the author.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-8779210363155573432022-01-13T12:52:21.274+01:002022-01-13T12:52:21.274+01:00Newly started journal with a fee waiver until the ...Newly started journal with a fee waiver until the end of this year:<br /><br />https://www.journals.elsevier.com/artificial-intelligence-in-the-life-sciencesFilip Miljkovićhttps://www.blogger.com/profile/10190023668899095209noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-80381722886824349262022-01-05T21:11:29.022+01:002022-01-05T21:11:29.022+01:00FixedFixedJan Jensenhttps://www.blogger.com/profile/08595894308946022740noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-2976469430117130282022-01-05T10:13:50.890+01:002022-01-05T10:13:50.890+01:00Hi Jan,
the link for Chemical Science does not wo...Hi Jan,<br /><br />the link for Chemical Science does not work. Could you fix that?Stephanhttps://www.blogger.com/profile/18301132942439461616noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-1761000590830082652022-01-04T12:41:25.534+01:002022-01-04T12:41:25.534+01:00My unofficial criterion is <$2000My unofficial criterion is <$2000Jan Jensenhttps://www.blogger.com/profile/08595894308946022740noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-68623822883013527982022-01-04T12:11:29.442+01:002022-01-04T12:11:29.442+01:00JCheminf? https://jcheminf.biomedcentral.com/JCheminf? https://jcheminf.biomedcentral.com/Emma Schymanskihttps://www.blogger.com/profile/16470756127551057228noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-12995810132189914072021-03-28T17:07:17.597+02:002021-03-28T17:07:17.597+02:00This is not surprising. AI models are simply high...This is not surprising. AI models are simply highly nonlinear (in the parameters) regression models, nothing more and nothing less (except for their sexy name). I see many papers in the literature that fit empirical (thermodynamic and other) models to data and they call it "prediction". This is not prediction, but "correlation". The only possible "predictions" are essentially "interpolations" within the ranges of experimental conditions under which the training set data was collected. Only if you use a model that incorporates basic scientific principles can you make "predictions". An example of a "prediction methodology" is "Accurately Predicting CO$_2$ Reactive Absorption Properties in Aqueous Alkanolamine Solutions by Molecular Simulation Requiring No Solvent Experimental Data", Ind. Eng. Chem. Res., 59, 18254--18268 (2020).WSmithGuelphhttps://www.blogger.com/profile/13919793226142596969noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-80547068637194017212020-12-14T13:54:03.493+01:002020-12-14T13:54:03.493+01:00Dear Prof. Jensen,
I have some problems that invo...Dear Prof. Jensen,<br /><br />I have some problems that involve atom reordering as well: I partly solved these in a program: https://github.com/conradhuebler/curcuma<br /><br />And I would be happy, if there were interested researchers with similar problems, that might have a look at my solution. <br /><br /><br />Thanks and best regards,<br />ConradConradhttps://www.blogger.com/profile/13821134969813214158noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-10595417146189900292020-02-19T13:16:41.001+01:002020-02-19T13:16:41.001+01:00Hi Jan,
I appreciate your candour - to help move ...Hi Jan,<br /><br />I appreciate your candour - to help move things forward here's a unittest file, these are what I think GEDs should be. My view is a unit of GED change should 1 per addition/removal of each atom and 1 per addition or deletion of a bond.<br /><br />Ed<br /><br />import unittest<br />from GED import *<br /><br /><br />class MyTestCase(unittest.TestCase):<br /><br /> def run_tests(self, d):<br /> for n, p in d.items():<br /> with self.subTest(p=p):<br /> GED = calc_GED(p[0], p[1])<br /> self.assertEqual(GED,p[2], msg='GED not correct for ' + n)<br /><br /><br /> def test_aromatic_aromatic(self):<br /> mol_dict = {<br /> 'benzene->pyridine': ('c1ccccc1', 'n1ccccc1',1),<br /> 'pyridine->pyridazine': ('n1ccccc1', 'n1ncccc1',1),<br /> 'benzene->pyridazine': ('c1ccccc1', 'n1ncccc1', 2)<br /> }<br /> self.run_tests(mol_dict)<br /><br /> def test_aromatic_substituents(self):<br /> mol_dict = {<br /> 'benzene->toluene': ('c1ccccc1', 'c1ccccc1C',2),<br /> 'benzene->chlorobenzene': ('c1ccccc1', 'c1ccccc1Cl', 2),<br /> 'benzene->anisole':('c1ccccc1', 'c1ccccc1OC', 3)<br /> }<br /> self.run_tests(mol_dict)<br /><br /> def test_aliphatic_aromatic(self):<br /> mol_dict = {<br /> 'benzene->cyclohexane': ('c1ccccc1', 'C1CCCCC1',6)<br /> }<br /> self.run_tests(mol_dict)<br /><br /> def test_aliphatic_ring_expansion(self):<br /> mol_dict = {<br /> 'cyclopropane->cyclobutane': ('C1CC1', 'C1CCC1',4)<br /> }<br /> self.run_tests(mol_dict)<br /><br /> def test_aliphatic_aliphatic(self):<br /> mol_dict = {<br /> 'ethane->propane': ('CC', 'CCC',2)<br /> }<br /> self.run_tests(mol_dict)<br /><br /><br /><br />if __name__ == '__main__':<br /> unittest.main()Ed Griffenhttps://www.blogger.com/profile/18354525587176891607noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-10567986834747668692020-02-16T12:51:43.117+01:002020-02-16T12:51:43.117+01:00The implementation makes two arbitrary choices:
1...The implementation makes two arbitrary choices:<br /><br />1. The structures are "kekulized" meaning that benzene is C1=CC=CC=C1 rather, meaning that 3 double bonds need to be changed to single bonds. I do this so that the distance between hexatriene (C=CC=CC=C) and benzene is 1. But whether that's a good idea is arguable. <br /><br />2. The Hs are implicit, so when a double bond is changed to a single bond, then the number of Hs on the Cs adjust, and don't contribute to the edit distance.<br /><br />I should also note that there is a bug in the implementation. The way I compare atoms (nodes) is by defining a "bond to itself". This means that additions of atoms is double counted (both as an edge and a node". The the code predicts the GED between C and CC to be 3, when in fact it is 2.Jan Jensenhttps://www.blogger.com/profile/08595894308946022740noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-5977213580530112942020-02-10T14:52:34.077+01:002020-02-10T14:52:34.077+01:00Really nice, however, can you (or perhaps Noel and...Really nice, however, can you (or perhaps Noel and Roger) explain how the distance from benzene to cyclohexane is 3? I might have expected 6, or 12, but 3 is a mystery - my ignorance I suspect.<br /><br />Here's my snippet of code:<br /><br />def calc_GED(s1,n, s2, m):<br /> mol1 = Chem.MolFromSmiles(s1)<br /> mol2 = Chem.MolFromSmiles(s2)<br /> G1 = get_graph(mol1)<br /> G2 = get_graph(mol2)<br /> GDE = nx.graph_edit_distance(G1, G2, edge_match=lambda a, b: a['weight'] == b['weight'])<br /> print(n, m, GDE)<br /><br />mol_dict = {<br /> 'benzene':'c1ccccc1',<br /> 'pyridine':'n1ccccc1',<br /> 'pyridazine':'n1ncccc1',<br /> 'cyclohexane':'C1CCCCC1'<br />}<br /><br />for n,s1 in mol_dict.items():<br /> for m,s2 in mol_dict.items():<br /> if n != m:<br /> calc_GED(s1,n, s2, m)Ed Griffenhttps://www.blogger.com/profile/18354525587176891607noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-87995940454272536792013-01-05T11:59:16.550+01:002013-01-05T11:59:16.550+01:00Mmm. I love keeping my journal's policy on arX...Mmm. I love keeping my journal's policy on arXiv deposition ambiguous. Offers me an additional way to reject the really lousy papers without review...<br /><br />And, at the risk of sounding snarky, the percentage of papers deposited openly as a function of a given field (e.g., math > physics > chemistry) is inversely proportional to that field's likelihood of producing anything having any useful/economic consequence.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-62380974234102621062013-01-02T14:20:30.452+01:002013-01-02T14:20:30.452+01:00Very cool video!Very cool video!Anonymoushttps://www.blogger.com/profile/18123073028209991396noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-79168686382378417242012-12-28T11:12:02.081+01:002012-12-28T11:12:02.081+01:00Did you get the right answer though?Did you get the right answer though?Jan Jensenhttps://www.blogger.com/profile/08595894308946022740noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-8120116365496165362012-12-28T11:11:36.828+01:002012-12-28T11:11:36.828+01:00Thank you!Thank you!Jan Jensenhttps://www.blogger.com/profile/08595894308946022740noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-88641221790264218612012-12-28T07:33:11.029+01:002012-12-28T07:33:11.029+01:00oh one is a carbon and the other is the nitrogen (...oh one is a carbon and the other is the nitrogen (amine/methyl). took me a while to figure out what's the difference between A & B moleculesdimkahttps://www.blogger.com/profile/15283992791038140199noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-34824860159703569292012-12-27T14:58:26.610+01:002012-12-27T14:58:26.610+01:00Very nice example!Very nice example!Anonymoushttps://www.blogger.com/profile/10508646868117037632noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-71666993794830099182012-12-27T11:05:51.723+01:002012-12-27T11:05:51.723+01:00Many people who publish in JCP consider themselves...Many people who publish in JCP consider themselves chemists and most chemists have never heard about arXiv. Most of the ones who have think arXiv is exclusively for physics. It would really help if arXiv made a chemistry section. I wrote to them about this, but no reply.<br /><br />The other, related, issue is whether chemists feel an arXiv submission establishes priority since they live in perpetual fear of being scooped. <br /><br />The final issue is that most chemistry journals appear to be arXiv-hostile, though some are not when you actually write to them and ask.Jan Jensenhttps://www.blogger.com/profile/08595894308946022740noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-14922590270520514122012-12-26T15:25:30.026+01:002012-12-26T15:25:30.026+01:00By the way, http://www.councilscienceeditors.org/f...By the way, http://www.councilscienceeditors.org/files/presentations/2009/Ingoldsby.pdf has some research on what percentage of physics articles are on arXiv. Not surprisingly, J. Chem. Phys. is doing quite bad at around 7%, whereas e.g. Phys. Rev. D has a submission ratio of 97%. It seems there are also some great differences between the different fields in physics.<br /><br />Money quote: "Of course, all Physical Review Letters papers are contained in the arXiv" (though looking at the numbers, only 55% of all Phys. Rev. Lett. papers are on arXiv)<br />Michael Banckhttps://www.blogger.com/profile/10019023009908861248noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-5918037651515797612012-12-25T12:23:46.363+01:002012-12-25T12:23:46.363+01:00It is true that if you use Avnir's method you ...It is true that if you use Avnir's method you will get a non-zero entropy contribution like $S = R \ln(3.9)$, but it won't be exactly the same as the conformational entropy. $S=-R\sum_i^4 p_i\ln(p_i)$ where $p_i=e^{-(G^\circ_i-G^\circ)/RT}$ and $G^\circ=-RT\sum_i^4 e^{-G^\circ_i/RT}$. $G^\circ$ is the free energy of the R-A complex, which has four different binding modes. If A has four-fold symmetry then $G^\circ_1=G^\circ_2=G^\circ_3=G^\circ_4$, $p_i=\frac{1}{4}$ and $S=R\ln(4)$. If you include the symmetry number when computing the entropy of A then this contribution to the free energy is taking care of that way.<br /><br />If A is formally $C_1$ then the four binding free energies will be different and the conformational entropy will be a value between $R\ln(4)$ and $R\ln(1)$. This is also true for Avnir's method but, based on his equations, I don't see why the entropies computed in these two ways would be the same. However, Avnir's method might yield a good approximation to the conformational entropy, I don't know. <br /><br />Notice also that the binding free energies can in principle be similar completely by accident and not to do with symmetry in any way. Jan Jensenhttps://www.blogger.com/profile/08595894308946022740noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-68846013306371665062012-12-24T16:36:07.682+01:002012-12-24T16:36:07.682+01:00Very nice reply. Just some further points:
«Howev...Very nice reply. Just some further points:<br /><br />«However, if molecule A is even *slightly* asymmetric then the effect enters as the conformational entropy of the complex.» Is this a «law» according to Quantum Mechanics? Because, if you apply the continuous symmetry arguments of Avnir, a molecule which is «slightly» assymetric (a flexible molecule) will have a «slight» degree of assymetry. In this case, a flexible A, for example, would have S = R ln(3.9). <br /><br />«Both are formally correct but I would argue the former is more general as it also works for non-symmetric molecules.» I do not fully agree with the reason here, because if for some reason the complex has an higher entropy because the ligand has more possibilities of binding, it means that the ligand itself has some symmetry operation somewhere and it can still be present on the rotational entropy of the ligand.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-54988146399811202862012-12-24T11:45:47.161+01:002012-12-24T11:45:47.161+01:00Good questions! Let me take the last question fir...Good questions! Let me take the last question first since it is a bit easier. If I understood your question correctly your talking about the extra degree of freedom due to rotation around the horizontal axis. In that case it depends on whether you view the model as being 2-dimensional or 3-dimensional. I view it as 2D where this degree of freedom is not allowed. However, if you view it as 3D (i.e. flat molecules in a 3D world) then you are correct.<br /><br />Now to your first question. Yes you can view it like that if the molecule is perfectly symmetric. The this effect is included in the rotational entropy of free A via the symmetry number (<a href="http://molecularmodelingbasics.blogspot.dk/2012/12/conformational-and-rotational-entropy.html" rel="nofollow">see this post</a>). However, if molecule A is even *slightly* asymmetric then the effect enters as the conformational entropy of the complex. So I think in the symmetric case the effect can also be ascribed to a conformational entropy of the complex.<br /><br />One could argue that the entropy of A can be measured to settle this issue. However, I don't think that is so straightforward - even conceptually. The entropy at temperature T is measured relative to absolute zero where the entropy is zero by definition. However, this state cannot be reached, and even at very, very low T symmetric molecules will have so-called residual entropy related to their symmetry which will be hard - if not impossible - to measure accurately. But I am not sure.<br /><br />Bottom line: I think the answer to your question is: ultimately what is measured is an entropy *change* and the higher entropy change for A can be explained *either* as a lower conformational entropy of the complex or a higher rotational entropy of the free ligand. Both are formally correct but I would argue the former is more general as it also works for non-symmetric molecules.Jan Jensenhttps://www.blogger.com/profile/08595894308946022740noreply@blogger.comtag:blogger.com,1999:blog-5744790168115177005.post-72415672642614857652012-12-23T12:17:52.015+01:002012-12-23T12:17:52.015+01:00Just want to comment this paragraph: «If you mix e...Just want to comment this paragraph: «If you mix equal amounts of A, B, and R you will get more R-A than R-B at equilibrium even though the hydrogen bond strength is the same in the two complexes. This is because molecule A can bind in four different ways while B can only bind one way, i.e. the R-A complex has a degeneracy of four (g=4) and the R-B complex has a degeneracy of one (g=1). Put another way, the R-A complex is more likely because it has a higher entropy (S=Rln(4)) than the R-B complex (S=Rln(1)).» <br /><br />Isn´t the entropy of complexation of the case A higher, because the rotational entropy of the unbinded molecule A is smaller, rather than being the entropy of the complex itself higher?<br /><br />Shouldn´t be the degeneracy 8 (A) and 2 (B), instead of 4 (A) and 1 (B)? For molecule A, 4 with the molecule facing up and 4 with the molecule facing down. And for molecule B, one with the molecule facing up and one with the molecule facing down. Anonymousnoreply@blogger.com