User 234000dc20
28-02-2014 03:52:38
Hello,
I would like to be able to count how many times in a library dimers of fragments appear. I can create my library of fragment in the following way:
* for the rings, I have used the RingChain.xml file provided in the fragmenter example page, and then kept the fragment with a non-0 value after cxcalc ringbondcount.
* for non-rings, I have something very basic: the 4 main halides, alkene, alkyne, cyanide, carbonyl, amide, carboxylic acid, nitro, C, O, N... (plus 4 or 5 containing sulfur). In my scheme, if I join carboxylic acid with whatever at the O position it becomes an ester. If I join O with only 1 fragment it is an alcohol, with 2 fragments it is an ether.
As I said, I would like to construct the dimers of fragments, and then use jcsearch for example to count how many times each dimer appears. I can create the dimers of non-rings fragments by hand, but as soon as I start considering rings it cannot be done anymore (my library is made of 8000+ drugs, and there are 1500 different kind of ring systems).
Thus, I have thought of using "fragment" to fragment each molecule of my library in as many fragment as possible, and then sort the results of the full decomposition to keep structures with only one single bond between two fragments. I have thought of keeping structures with a rotatablebondcount (from cxcalc) of 1. However as explained here https://www.chemaxon.com/forum/ftopic4060.html&view=previous, bonds to terminal atoms are not considered as rotatable. So I will not be able to keep structures such as RING-CH3.
I was wondering if it would be possible to do something with "fragment" to achieve this. Does anybody have an idea?
Thanks a lot.
Nicolas