Re: [Cdk-devel] atomic number in Element constructor?

[Egon W. <ego...@gm...>]

On Fri, Dec 10, 2010 at 2:57 PM, Rajarshi Guha <raj...@gm...> wrote:
>> new Atom(Elements.CARBON);
>>
>> which would actually set both, because we are clearly reading from a
>> controlled vocabulary here, and this is actually what you can do
>> nowadays.
>
> Yes, this is nice when you are constructing atoms by hand. How would
> the SMILES parser make use of this? Would we have to rewrite it to
> have a 100-line if-else block?

Fair. We would need an ElementFactory for that:

IElement element = ElementFactory.getElement((String)symbol);

This might actually be good for memory use too, as we will not have
many "C" Strings instantiated...

>> However, I do not think class are using that yet... but this
>> would be semantically correct... passing a String which happens to be
>> "C" is less clear...
>
> It would be simply enough to do a look up on the periodic table by
> symbol and get the atomic number and the appropriate details. If an
> invalid symbol is provided, then atomic number is null

Yes.

>> I recommend using the Atom(IElement) constructor. That should set the
>> atomic number.
>>
>> Would that work for you?
>
> Not really, because right now, molecules obtained from a SMILES string
> have null for atomic number

I have no objections against setting the atomic number in the SmilesParser.

Egon

-- 
Dr E.L. Willighagen
Postdoctoral Research Associate
University of Cambridge
Homepage: http://egonw.github.com/
LinkedIn: http://se.linkedin.com/in/egonw
Blog: http://chem-bla-ics.blogspot.com/
PubList: http://www.citeulike.org/user/egonw/tag/papers