![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F89367%2F2c874ac2e0acaaf437f21a828fc12614%2Fmola.PNG?generation=1561900643669650&alt=media\")
This is an observation that is puzzling me. Disclaimer: I know very little about chemistry and less about quantum physics. I do believe there is a relatively simple explanation for this - maybe one of you with more insight knows the answer.
\n\nSo here it goes: take, as an example, molecule dsgdb9nsd_011591 with atom indexes 12, 3. It has an fc of 105.25.
\n\nIn contrast, molecule dsgdb9nsd_012826 with atom indexes 17, 7 has an fc of less - namely 95.3125. Yet both are of the same type (1JHC), have the same number/type of atoms, and what's more - around the two atoms we care about, they are locally very similar (inside a sphere with radius 2.5 around the two relevant atoms, the configuration of atoms is the most similar I could find in the training set). To illustrate, here are two images of these two molecules:
\n\ndsgdb9nsd_011591:
\n\n
dsgdb9nsd_012826:
\n\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F89367%2Fb95b22022440bce19791690a2821dfed%2Fmolb.PNG?generation=1561900665295359&alt=media\")
In these images, the atom with the first link index is red, the one with the second index is green. The other atoms are yellow (C), white (H) and blue (O). Chemists will likely cringe - sorry.
\n\nAs you can see, around the red/green pair the configuration is very similar. Any idea what causes the large difference in fc anyone?
\n\nI can supply the Blender files I used if anybody wants to examine this in 3D.
", "messages": [ { "id": "565121", "postDate": "06/30/2019 13:20:50", "content": "This is an observation that is puzzling me. Disclaimer: I know very little about chemistry and less about quantum physics. I do believe there is a relatively simple explanation for this - maybe one of you with more insight knows the answer.
\n\nSo here it goes: take, as an example, molecule dsgdb9nsd_011591 with atom indexes 12, 3. It has an fc of 105.25.
\n\nIn contrast, molecule dsgdb9nsd_012826 with atom indexes 17, 7 has an fc of less - namely 95.3125. Yet both are of the same type (1JHC), have the same number/type of atoms, and what's more - around the two atoms we care about, they are locally very similar (inside a sphere with radius 2.5 around the two relevant atoms, the configuration of atoms is the most similar I could find in the training set). To illustrate, here are two images of these two molecules:
\n\ndsgdb9nsd_011591:
\n\n
dsgdb9nsd_012826:
\n\n
In these images, the atom with the first link index is red, the one with the second index is green. The other atoms are yellow (C), white (H) and blue (O). Chemists will likely cringe - sorry.
\n\nAs you can see, around the red/green pair the configuration is very similar. Any idea what causes the large difference in fc anyone?
\n\nI can supply the Blender files I used if anybody wants to examine this in 3D.
", "rawMarkdown": "This is an observation that is puzzling me. Disclaimer: I know very little about chemistry and less about quantum physics. I do believe there is a relatively simple explanation for this - maybe one of you with more insight knows the answer.\n\nSo here it goes: take, as an example, molecule dsgdb9nsd_011591 with atom indexes 12, 3. It has an fc of 105.25. \n\nIn contrast, molecule dsgdb9nsd_012826 with atom indexes 17, 7 has an fc of less - namely 95.3125. Yet both are of the same type (1JHC), have the same number/type of atoms, and what's more - around the two atoms we care about, they are locally very similar (inside a sphere with radius 2.5 around the two relevant atoms, the configuration of atoms is the most similar I could find in the training set). To illustrate, here are two images of these two molecules:\n\ndsgdb9nsd_011591:\n\n\n \ndsgdb9nsd_012826:\n\n\n\nIn these images, the atom with the first link index is red, the one with the second index is green. The other atoms are yellow (C), white (H) and blue (O). Chemists will likely cringe - sorry.\n\nAs you can see, around the red/green pair the configuration is very similar. Any idea what causes the large difference in fc anyone?\n\nI can supply the Blender files I used if anybody wants to examine this in 3D.", "votes": null }, { "id": "565595", "postDate": "07/01/2019 06:42:25", "content": "I can't tell on your pictures, but does the top one have a double-bond O, and the bottom contains an OH?
\n\nThat would be enough to affect the electron density on the C that the H is bonded to and change the J-coupling strength.
", "rawMarkdown": "I can't tell on your pictures, but does the top one have a double-bond O, and the bottom contains an OH?\n\nThat would be enough to affect the electron density on the C that the H is bonded to and change the J-coupling strength.", "votes": null }, { "id": "565615", "postDate": "07/01/2019 07:05:42", "content": "Thanks - this is a good pointer! I'll read up a bit on electron densities and double bindings and then investigate further.
", "rawMarkdown": "Thanks - this is a good pointer! I'll read up a bit on electron densities and double bindings and then investigate further.", "votes": null }, { "id": "565665", "postDate": "07/01/2019 08:49:42", "content": "some cool tools to draw molecular orbitals\nhttps://iwatobipen.wordpress.com/2017/07/30/quantum-chemistry-calculation-with-python/#jp-carousel-1824\nhttp://gabrielelanaro.github.io/blog/2015/01/14/visualizing-molecular-orbitals-ipython-notebook.html\nhttps://orbkit.github.io/\n![\"\"](\"https://iwatobipen.files.wordpress.com/2017/07/screen-shot-2017-07-30-at-23-29-31.png?w=1200&h=\")
This looks very helpful - thank you, I will check that out!
", "rawMarkdown": "This looks very helpful - thank you, I will check that out!", "votes": null } ], "comments": [ { "id": 565595, "author_name": "phucco", "author_url": "", "post_date": "07/01/2019 06:42:25", "content": "I can't tell on your pictures, but does the top one have a double-bond O, and the bottom contains an OH?
\n\nThat would be enough to affect the electron density on the C that the H is bonded to and change the J-coupling strength.
", "votes": null, "replies": [ { "id": 565615, "author_name": "chriskorn", "author_url": "", "post_date": "07/01/2019 07:05:42", "content": "Thanks - this is a good pointer! I'll read up a bit on electron densities and double bindings and then investigate further.
", "votes": null, "replies": [] } ] }, { "id": 565665, "author_name": "hengck23", "author_url": "", "post_date": "07/01/2019 08:49:42", "content": "some cool tools to draw molecular orbitals\nhttps://iwatobipen.wordpress.com/2017/07/30/quantum-chemistry-calculation-with-python/#jp-carousel-1824\nhttp://gabrielelanaro.github.io/blog/2015/01/14/visualizing-molecular-orbitals-ipython-notebook.html\nhttps://orbkit.github.io/\n
This looks very helpful - thank you, I will check that out!
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "565121": "This is an observation that is puzzling me. Disclaimer: I know very little about chemistry and less about quantum physics. I do believe there is a relatively simple explanation for this - maybe one of you with more insight knows the answer.\n\nSo here it goes: take, as an example, molecule dsgdb9nsd_011591 with atom indexes 12, 3. It has an fc of 105.25. \n\nIn contrast, molecule dsgdb9nsd_012826 with atom indexes 17, 7 has an fc of less - namely 95.3125. Yet both are of the same type (1JHC), have the same number/type of atoms, and what's more - around the two atoms we care about, they are locally very similar (inside a sphere with radius 2.5 around the two relevant atoms, the configuration of atoms is the most similar I could find in the training set). To illustrate, here are two images of these two molecules:\n\ndsgdb9nsd_011591:\n\n\n \ndsgdb9nsd_012826:\n\n\n\nIn these images, the atom with the first link index is red, the one with the second index is green. The other atoms are yellow (C), white (H) and blue (O). Chemists will likely cringe - sorry.\n\nAs you can see, around the red/green pair the configuration is very similar. Any idea what causes the large difference in fc anyone?\n\nI can supply the Blender files I used if anybody wants to examine this in 3D.", "565595": "I can't tell on your pictures, but does the top one have a double-bond O, and the bottom contains an OH?\n\nThat would be enough to affect the electron density on the C that the H is bonded to and change the J-coupling strength.", "565615": "Thanks - this is a good pointer! I'll read up a bit on electron densities and double bindings and then investigate further.", "565665": "some cool tools to draw molecular orbitals\nhttps://iwatobipen.wordpress.com/2017/07/30/quantum-chemistry-calculation-with-python/#jp-carousel-1824\nhttp://gabrielelanaro.github.io/blog/2015/01/14/visualizing-molecular-orbitals-ipython-notebook.html\nhttps://orbkit.github.io/\n", "565758": "This looks very helpful - thank you, I will check that out!" }, "source": "meta" }