From: Peter M. <Pet...@sp...> - 2007-08-31 07:35:22
|
Hi Rajarshi, Thanks for you comments. Actually I already added hydrogens in my code when needed. But it's good to know that one should always use the explicite Hydrogens to be on the save site. It might be wise to put in the documentation somewere. All databases I have around and I think most commercially available database, do not have explicite hydrogens. (Our corporate database is MDL's ISIS/host). Therefor I work with 2D SD-files without explicite hydrogens. Cheers, Peter -----Original Message----- From: Rajarshi Guha [mailto:rg...@in...] Sent: Thursday, August 30, 2007 11:45 PM To: Peter Maas Cc: cdk...@li... Subject: Re: [Cdk-user] H-Bond Donor Descriptor On Aug 30, 2007, at 8:14 AM, Peter Maas wrote: > > Hi all, > > Actually I had exactly the same behavour. > The H-bond descriptor seems only the be working properly when you add > implicite hydrogens (which is also stated in the documentation). > > Brings me to the question. How is this handled in the Descriptor > Engine? > Should one always add implicit or even explicit hydrogens to be on the > save side? > Are there descriptors which have problems with explicite hydrogens? In general descriptors will remove H's if they are meant to work on a H-depleted graph. I should probably check up on this. > Also looked at the source code of Rajarshi Guha's Descriptor GUI. I > can't find were the hydrogens are checked or added. I tried the H-bond > Donor descriptor from this GUI but and it also return almost only > ZERO's. The descriptor GUI will not add/delete H's - for the case of 3D structures the 3D desciptors will expect H's to be present and topologicals will ignore the H's. In general, it's best to supply molecules that have H's. Now the problematic case I think is when you try and caculate H-bond donor/acceptors. In this case you don't need 3D structures, but you'd need H's. Are you supplying 3D structures? Or SMILES? Implicit H's is an ongoing problem I think. In the case of the H-bond descriptor IIRC the code is basically a set of if/then's. It might be time to convert it to a SMARTS based form which will be more reliable (though it doesn't necessarily remove the implicit H problem) ------------------------------------------------------------------- Rajarshi Guha <rg...@in...> GPG Fingerprint: 0CCA 8EE2 2EEB 25E2 AB04 06F7 1BB9 E634 9B87 56EE ------------------------------------------------------------------- Science kind of takes the fun out of the portent business. -Hobbes |