From owner-chemistry[ AT ]ccl.net Mon Nov 11 16:12:01 2024 From: "m.dominic.ryan#gmail.com" To: CCL Subject: CCL: Computational Chemistry Ontology Message-Id: <-55250-241111091243-24527-jlxpyt9Y/CgpJL3xeq66Pw,+,server.ccl.net> X-Original-From: Content-Language: en-us Content-Type: multipart/alternative; boundary="----=_NextPart_000_0113_01DB3419.D85334C0" Date: Mon, 11 Nov 2024 09:12:33 -0500 MIME-Version: 1.0 Sent to CCL by: [m.dominic.ryan++gmail.com] This is a multipart message in MIME format. ------=_NextPart_000_0113_01DB3419.D85334C0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable I think the pointers previously cited give a good overview of the = problem of ontology in computational chemistry. That term is too broad = for usefully defining an ontology. You really need to identify the = problem space you are working in such as cheminformatics based on data = extracted from ELNs, or assay definitions and types that you might troll = through in pulling data from PubChem etc. Or, the various tools used in = QM applications with terms ranging from basis sets to density = functionals etc. =20 But on the philosophy side I=E2=80=99ll quibble a bit on what was said.=20 =20 The quantization of physical phenomena is the =E2=80=98law of = nature=E2=80=99. It is captured effectively in Schroedinger=E2=80=99s = equation due to that mathematical formalism. Quantum mechanics is an = approximation to solutions of Schoedinger=E2=80=99s equation. As = computational power has increased over the decades we are increasingly = able to approach useful problem spaces. That is often through the = combination of methods of varying treatment of electron correlation and = dispersion. The severe scaling of problems means that we still can only = approximate fine details of larger problems. =20 The question usually comes down to what level of detail matters to your = problem. Do you want to get some idea of how a protein can move around = your ligand? Various approaches to molecular dynamics will do very = nicely. But, you could add a more sophisticated handling of water = through additional DFT evaluation of each step and add correction factor = forces to the MD. Are you trying to model light harvesting proteins and = account for energetics of photon to electron shift? That=E2=80=99s going = to need more.=20 =20 DFT, coupled cluster methods, or pure molecular mechanics are simply = empirical formalisms that are useful in the right setting.=20 =20 We can extend this to QSAR in a dramatic example. If you compile enough = examples of a problem space and develop a mathematical model that = represents that space with high confidence, have you identified a = =E2=80=98law of nature=E2=80=99? The recent Nobel Prize on protein = folding is just that. All these =E2=80=98AI=E2=80=99 methods are well = fitted models, often simply with very deep (many layered) neural nets to = model the phenomena. Even in this space though, there are still = exceptions. =20 So, how much =E2=80=98truth=E2=80=99 do you want with your ontology?=20 =20 Dominic Ryan =20 > From: owner-chemistry+m.dominic.ryan=3D=3Dgmail.com+*+ccl.net = On Behalf Of = David Young zqmuser]![gmail.com Sent: Sunday, November 10, 2024 8:18 AM To: Ryan, M Dominic Subject: CCL: Computational Chemistry Ontology =20 Darren, =20 Quantum chemistry is an application of quantum mechanics, which is = generally accepted as a law of nature, so follow the ontology for laws = of nature. =20 Molecular mechanics & dynamics are based on other laws of nature (i.e. = Hook's Law of Elasticity), which are not strictly the correct = formulation but have been validated as giving results that match to = experiment to a reasonable accuracy within a certain region (near = equilibrium geometry). The constants in these equations are fitted to = reproduce experimental data. =20 Some of the cheminformatics methods, like QSAR and QSPR, are completely = empirical. Although they are not derived from any law of nature, they = can give results accurate enough to be useful for certain research = needs. Dave Young (he, him, Dr., scientist, fencer, author) =20 The wise person tries to find the truth. The foolish person believes whoever tells them what they want to hear. =20 =20 =20 On Sun, Nov 10, 2024 at 6:40=E2=80=AFAM Darren Rhodes = darren.rhodes!^!gmail.com > wrote: Sent to CCL by: "Darren Rhodes" [darren.rhodes]|[gmail.com = ] Hi All Can you point me in the direction of an appropriate ontology for use in=20 computational chemistry? all the best Darren. darren.rhodes.:.REMOVEgmail.com -=3D This is automatically added to each message by the mailing script = =3D-
= or use:E-mail to administrators: CHEMISTRY-REQUEST[a]ccl.net = or use

I think = the pointers previously cited give a good overview of the problem of = ontology in computational chemistry. That term is too broad for usefully = defining an ontology. You really need to identify the problem space you = are working in such as cheminformatics based on data extracted from = ELNs, or assay definitions and types that you might troll through in = pulling data from PubChem etc. Or, the various tools used in QM = applications with terms ranging from basis sets to density functionals = etc.

 

But on = the philosophy side I=E2=80=99ll quibble a bit on what was said. =

 

The = quantization of physical phenomena is the =E2=80=98law of = nature=E2=80=99. It is captured effectively in Schroedinger=E2=80=99s = equation due to that mathematical formalism. Quantum mechanics is an = approximation to solutions of Schoedinger=E2=80=99s equation. As = computational power has increased over the decades we are increasingly = able to approach useful problem spaces. That is often through the = combination of methods of varying treatment of electron correlation and = dispersion. The severe scaling of problems means that we still can only = approximate fine details of larger problems.

 

The = question usually comes down to what level of detail matters to your = problem. Do you want to get some idea of how a protein can move around = your ligand? Various approaches to molecular dynamics will do very = nicely. But, you could add a more sophisticated handling of water = through additional DFT evaluation of each step and add correction factor = forces to the MD. Are you trying to model light harvesting proteins and = account for energetics of photon to electron shift? That=E2=80=99s going = to need more.

 

DFT, = coupled cluster methods, or pure molecular mechanics are simply = empirical formalisms that are useful in the right setting. =

 

We can = extend this to QSAR in a dramatic example. If you compile enough = examples of a problem space and develop a mathematical model that = represents that space with high confidence, have you identified a = =E2=80=98law of nature=E2=80=99? The recent Nobel Prize on protein = folding is just that. All these =E2=80=98AI=E2=80=99 methods are well = fitted models, often simply with very deep (many layered) neural nets to = model the phenomena. Even in this space though, there are still = exceptions.

 

So, how = much =E2=80=98truth=E2=80=99 do you want with your ontology? =

 

Dominic = Ryan

 

From:<= /b> = owner-chemistry+m.dominic.ryan=3D=3Dgmail.com+*+ccl.net = <owner-chemistry+m.dominic.ryan=3D=3Dgmail.com+*+ccl.net> On = Behalf Of David Young zqmuser]![gmail.com
Sent: Sunday, = November 10, 2024 8:18 AM
To: Ryan, M Dominic = <m.dominic.ryan+*+gmail.com>
Subject: CCL: Computational = Chemistry Ontology

 

Darren,

 

Quantum chemistry is an = application of quantum mechanics, which is generally accepted as a law = of nature, so follow the ontology for laws of = nature.

 

Molecular mechanics & = dynamics are based on other laws of nature (i.e. Hook's Law of = Elasticity), which are not strictly the correct formulation but have = been validated as giving results that match to experiment to a = reasonable accuracy within a certain region (near equilibrium = geometry).  The constants in these equations are fitted to = reproduce experimental data.

 

Some of the cheminformatics = methods, like QSAR and QSPR, are completely empirical. Although they are = not derived from any law of nature, they can give results accurate = enough to be useful for certain research = needs.


Dave Young (he, him, Dr., scientist, fencer, = author)

 

The wise person tries to find the = truth.

The foolish = person believes whoever tells them what they want to = hear.

 

<= p class=3DMsoNormal = style=3D'margin-left:.5in'> 

 

On Sun, Nov 10, 2024 at = 6:40=E2=80=AFAM = Darren Rhodes darren.rhodes!^!gmail.com = <owner-chemistry[a]ccl.net&g= t; wrote:


Sent to CCL by: "Darren  Rhodes" = [darren.rhodes]|[gmail.com]
Hi All

Can you point me in = the direction of an appropriate ontology for use in
computational = chemistry?

all the = best

Darren.
darren.rhodes.:.REMOVEgmail.com



-=3D= This is automatically added to each message by the mailing script = =3D-<br


E-mail to subscribers: CHEMISTRY[a]ccl.net or use:
    =   http://www.ccl.net/cgi-bin/ccl/send_ccl_message
=
E-mail to administrators: CHEMISTRY-REQUEST[a]ccl.net or use
  =     http://www.ccl.net/cgi-bin/ccl/send_ccl_message
= <br
      http://www.ccl.net/chemistry/sub_unsub.shtml
Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/chemistry/announcements/conferenc= es/

Search Messages: http://www.ccl.net/chemistry/searchccl/index.shtml<= br><br
      http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/instructions/=

------=_NextPart_000_0113_01DB3419.D85334C0--