liveBook · Manning

Constraint-satisfaction problem

As explained in Build AI Drug Discovery Pipelines

Constraint satisfaction is the process of finding outputs that obey specified hard requirements while possibly optimizing additional soft objectives.

Constraint satisfaction separates hard requirements that every candidate must meet from soft objectives to optimize within the feasible set. In drug design, this can prevent strong predicted potency from compensating for a failure to meet a structural requirement. section 13.3.1

Molecular design requirements can include:

  • Hard constraints: at most one chiral center, molecular weight of 300–500 Da, or a required scaffold. section 13.3.1
  • Soft objectives: maximize predicted potency and minimize lipophilicity while satisfying the hard constraints. section 13.3.1
  • SMARTS pattern matching or scaffold-constrained generation can enforce structural constraints; a feasible solution must exist. section 13.3.1

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Defines hard and soft criteria, gives molecular examples, and names SMARTS matching and scaffold-constrained generation; notes feasible solutions must exist.
Examples
Uses a chiral-center limit, molecular-weight range, and required scaffold as hard constraints, with potency and lipophilicity as soft objectives.
What you can do after
Apply structural constraints to molecular generation while optimizing desired properties among candidates that satisfy them.
How it connects
Connects molecular generation with multi-objective optimization, SMARTS pattern matching, conditional generation, and guided optimization.

Also explained in

Compare how 3 books explain constraint-satisfaction problem →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build AI Drug Discovery Pipelines. The section links point to the parts of the book it draws on.