<cranix>
and not sure what to think about this approach
<Armael>
it's probably too complicated for its own good and awkward to manipulate
<cranix>
i need this data structure for representic organic chemical compunds
<dmbaturin>
Representing adamantanes or propellanes is going to be fun. :)
<cranix>
yes
<cranix>
i will start with something simpler
<cranix>
or cubane
<cranix>
basicly i'm investigating possibilty of using formal grammars and parsers for finding route of synthesis of organic compounds
<cranix>
and compiler theory
<cranix>
there is risk that i will need context sensitive grammar and there is no information about parsing them
<dmbaturin>
And then there's the issue of chirality.
<cranix>
right
<cranix>
i plan to work on it later
<cranix>
but i can define another type of bond
kakadu has quit [Remote host closed the connection]
<dmbaturin>
cranix: And then there are rotaxanes and molecular knots, but let's not think too hard about it...
<dmbaturin>
The idea is interesting though. I don't know if it's viable or not, but it's in the "huge if true" category.
<cranix>
the problem is that chemists do not know anything about computer science and computer sciencists about chemistry so i can not consult anybody
<dmbaturin>
cranix: How do you plan to deal with routes that require adding protective groups to be selective?
<cranix>
i'm thinking about this problem. i can define set of nonterminals that is 'safe' (no protection required) for given reaction
<cranix>
or introduce context to grammar
<cranix>
i'm afraid that protective groups will change length on right side of production rule less than left side -> turing completeness
FreeBirdLjj has joined #ocaml
<cranix>
i plan to encode reactions as production rules
<cranix>
at first stage i plan some lexer that will convert just atoms into nodes composed of a few atoms that have more sematnics
<cranix>
*semantics
<dmbaturin>
Are things like LiAlH4 going to be outside of the normal encoding system?
<cranix>
yes
<cranix>
i'm going to focus on main molecule
<dmbaturin>
I think Grignard can be a good smoke test for the system.
<cranix>
like we have ester and convert it into alcohol (part from carboxylic acid, assuming that alcohol part is small and not intresting like ethyl group)
<cranix>
i'm thinkig about this reaction at first too
<dmbaturin>
If it can handle that, it's probably useful enough but that point.
<cranix>
then some aldol condensations
kakadu has joined #ocaml
<cranix>
i'm thinking about fexofenadine as first 'real world' target
<cranix>
it is complex enough
<cranix>
but realtively few corner cases
<cranix>
i would love to have 1)algebraic structure made of set of organic compund and binary operation as reaction between them
toppler has quit [Ping timeout: 268 seconds]
<cranix>
2)something like turing-compltenesess ie. set of simple substrates and reactions that will generate enough big set of intresting compounds
<dmbaturin>
Macrolydes are living corner cases.
FreeBirdLjj has quit [Ping timeout: 250 seconds]
kakadu_ has quit [Ping timeout: 276 seconds]
<dmbaturin>
* macrolides
<cranix>
yes
<cranix>
thats why i plan to start with fexofenadine
<cranix>
i plan to treat cycle (cyclic parts of compund) as one entity
<cranix>
becouse chemist think about cycle as one logical part with sematnics attached to it
<cranix>
at first i hoped that parsers are reasearched very good and i will be able to easly find way to parse context sensitive grammar, but this is not true
<dmbaturin>
Well... if you are to find a route to something explicitly polycyclic like anthracene, you have to deal with cycle closing.
<cranix>
yes
<cranix>
fused rings are another topic
toppler has joined #ocaml
<cranix>
if You are intrested i can later write You about progress of this topic
KeyJoo has quit [Quit: KeyJoo]
<dmbaturin>
cranix: Feel free to email me at daniil@baturin.org, or PM.
<cranix>
i will
<cranix>
it is very motivating that someone else is intrested in this topic
<cranix>
my mail is cranix@hackerspace.pl
<dmbaturin>
cranix: Sent you a test message.
<cranix>
got it
<cranix>
i will need to find a good way to input structure of compound to ocaml
kleisli has joined #ocaml
<dmbaturin>
There are existing formats that you may want to parse. Those used by quantum chemistry tools.
kleisli_ has joined #ocaml
<cranix>
i will have a look at them
<cranix>
SMILES is simple
<cranix>
and ASCII
<cranix>
but not sure about converting it to graphs