Inspiration

Realizing the expanse of common sense reasoning and it's abilities to solve certain problems that almost always require human-assistance lead us to thinking about it's application in the domain of retro-synthesis of chemical compounds. The main idea is in the manufacturing of a new drug and how we can best automate a process that uses a lot of manual effort yet significantly based on common-sense reasoning.

What it does

This project validates the output generated by the retro synthesis done by the computer aided software( for e.g reinforcement learning as a game)

How we built it

We used logic programming using prolog and scasp by defining the predicates and the rules governing the basis for the checks done by a chemist.

Challenges we ran into

1. Representation of molecular structure with regards to the number of atoms, covalent bonds, charge of an atom etc.
2. Learning Prolog and it's constructs.

Accomplishments that we're proud of

1. Usage of dictionaries for the representation of the chemical molecules that made it easier for the calculations.
2. The extensive research regarding the retrosynthesis of chemical compounds and the application of common reason as a solution to the validation problem.

What we learned

A new and exciting programming language that is prolog, working in a team, cheminfomatics, retrosynthesis, and chemistry ontology.

What's next for RETROFIER

1. In future this can be integrated with softwares like openbabel to incorporate that properties of reactants, along with expert human assistance.
2. Chemical reactions involve more complex conditions for which simple template rules may not be sufficient, owing to the necessity of new templates.
3. Scope for more inter-disciplinary research.

Built With

• prolog