openmolecules.org Forum: Cheminformatics

Home » DataWarrior » Cheminformatics » fragments

Show: Today's Messages :: Show Polls :: Message Navigator

fragments [message #521]

Thu, 11 April 2019 19:34

pc419714@ohio.edu
Messages: 20
Registered: February 2019

Junior Member

Thomas,

How does data warrior define a fragment? Why do you use fragments? My guess would be because of metabolism. Does a fragment correspond with a functional group?

Thanks,

Patrick

Report message to a moderator

Re: fragments [message #528 is a reply to message #521]

Sun, 14 April 2019 22:10

thomas
Messages: 247
Registered: June 2014

Senior Member

Hi Patrick,

in DataWarrior the word fragment usually refers to a substructure fragment, which in contrast to a normal molecule is incomplete such open valences are not considered to be filled with hydrogen atoms. Fragments may also contains atom and bond query features.

In some cases, however, 'fragment' also refers to a molecule, e.g. the functionality 'Add largest fragment', which identifies all non-connected parts of a molecule (e.g. in salts) and just copies that one with the highest number of non-hydrogens atoms into a new column. This column then contains 'molecules', where all free atoms valences are implicitly considered to be filled with hydrogen atoms.

Finally it is a matter of the context. I assume this does not fully answer your question. What was the context in your case?

Thomas

Report message to a moderator

Re: fragments [message #531 is a reply to message #528]

Mon, 15 April 2019 22:10

pc419714@ohio.edu
Messages: 20
Registered: February 2019

Junior Member

I was just wondering how fragments were defined for purposes of calculating the drug score. Thanks!

Report message to a moderator

Re: fragments [message #534 is a reply to message #531]

Tue, 16 April 2019 16:19

thomas
Messages: 247
Registered: June 2014

Senior Member

The drug score does not use fragments. It is a mathematical equation combining other properties.
The druglikeness, however, uses >5000 substructures. Everyone has an associated druglikeness contribution value, some positiv, some negativ, which were derived from the frequency these fragments occur in commercial drugs or organic chemicals. I attach a text file, which when opened from DataWarrior shows the chemical structures and the contributions of these fragments.

Attachment: druglikenessFragments.txt
(Size: 158.46KB, Downloaded 249 times)

Report message to a moderator

Re: fragments [message #535 is a reply to message #534]

Tue, 16 April 2019 22:56

pc419714@ohio.edu
Messages: 20
Registered: February 2019

Junior Member

This is great! Thanks! One idea my boss had was to modify the code so that it calculated the drug score each generation and picked the best ones. I might try to modify the source code to try this but the way it is set up, if you select compounds with a high drug score to begin with, it generates compounds with high drug scores. We ended up doing 2 evolutions.

This is very helpful knowing more details about the fragments.

Report message to a moderator

Previous Topic:	issue with permeability
Next Topic:	Flexophore

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Tue Dec 10 03:42:12 CET 2019

Total time taken to generate the page: 0.00350 seconds