Guide-linked quiz · Medium Level

Molecular Representations for Machine Learning Quiz

Compare SMILES, fingerprints, molecular graphs and 3D structures while tracking standardization, stereochemistry, collisions and evaluation leakage.

Question 1 of 8

What does a SMILES string encode?

Preview every question8 questions

Answers and explanations appear as you take the quiz.

  1. What does a SMILES string encode?
    • A fixed-length molecular fingerprint
    • Three-dimensional coordinates for one conformer
    • A serialized representation of a molecular graph
    • An experimental property measurement
  2. Which limitation can arise from folding substructure features into a fixed-length fingerprint?
    • They always store full 3D coordinates
    • Different substructures can collide when features are folded into bits
    • They cannot be used for similarity search
    • They directly encode experimental outcomes
  3. How does a molecular graph represent a molecule?
    • A text traversal that serializes atoms and bonds
    • A fixed-length vector of hashed substructure features
    • A set of three-dimensional coordinates for one conformer
    • Atoms as nodes and bonds as edges with features
  4. What extra information can a 3D representation provide?
    • Spatial geometry such as shape and interatomic distances
    • A fixed-length bit vector of hashed molecular fragments
    • A text traversal of atoms and bonds with no coordinates
    • An experimental measurement of affinity
  5. Why can the same molecule have multiple valid SMILES strings?
    • The molecule has a different elemental composition
    • SMILES stores no bond information
    • The graph can be traversed in different atom orders
    • Each SMILES string encodes a different assay result
  6. What does canonicalization aim to provide?
    • A guaranteed physical conformation
    • A standardized string representation under a chosen algorithm
    • A measured molecular property
    • A replacement for structure validation
  7. Why use scaffold-aware splits for some molecular tasks?
    • To increase the fingerprint length
    • To remove stereochemistry from every structure
    • To put close analogs in both training and evaluation
    • To reduce leakage from close chemical analogs across partitions
  8. What should be checked before using a 3D molecular model?
    • Conformer, protonation, stereochemistry and coordinate preparation
    • The order of characters in a canonical SMILES alone
    • The number of fingerprint bits used in a separate model
    • Only the number of atoms, without coordinates or stereochemistry

Keep testing yourself

More quizzes picked for your level.

View all quizzes