Skip to content

Rule-based models can be unified #37

Open

Description

@sfluegel05

SOTA: Independent tools

Currently, we support several rule-based systems:

  • ChemLog (subdivided into chemlog-peptides and element-based rules, pulled from different repositories) - formally MSOL, in practice we use a Python algorithm
  • Lopster - available via ChemLog, reimplementation of the original rules using clingo (?)
  • C3P - Python programs

In the future, we will probably add rules from NL2FOL / chebILP. Both use ASP syntax and clingo.

Goal: One format

This format will likely be ASP with clingo as the classification engine. That way, we could maintain one set of rules and don't have to worry about a multitude of tools.

I have looked into the peptide formalisation with the conclusion that it can be expressed in ASP syntax and classification with clingo is efficient (slightly slower than algorithmic, but still acceptable). The C3P programs are more difficult. Here we would need to check on an individual basis whether the programs use rdkit calls or complex constructions that can't be expressed in ASP.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions