<aside> 💡 Goal – Identify key substructures, fragments, and pharmacophores driving the bioactivity of molecules from the repurposing study

</aside>

Create rdScaffoldNetwork object

• Maps the relationship between molecular fragments, scaffolds, and initial molecules

• Edges of the graph include:

  • Initialization – remove the terminal side chains to form Bemis-Murcko scaffold
  • GenericBond – replace double bonds with generic single bonds
  • Generic – Replace atoms with dummies
  • Fragment – fragments along all single non-ring bonds between a ring atom and any other atom

  

• Need to effectively navigate this large space of scaffolds and fragments

  • Look at the counts object to see the nodes that occur the largest number of times

    

    

    

  • Look at the statistical significance of connections

  • Eliminate nodes that are too simple (e.g. single rings)

  • Enrich for nodes that are initialized from a molecule with high EC50

  • Perform substructure search for these significant substructure across repurposing sets

  • TODO: Rank by molecules similar to Ferrostatin-1, Liproxstatin-1 and then sort by EC50 or buckets (<1, 1-2, <0.1)

• Iterated through the list of substructures identified by scaffold network to try and find if Liprox and ferrostat share the same substructures. Did not identify any significant matches – the only matches were benzene and the substructure below

  

• Similarities of Ferrostatin-1 and Liproxstatin-1 to all of the other molecules in the screen

• Pearson correlations and p-values between EC50s and ferrostat/liproxstat similarities
  • Ferrostatin-1
    • Correlation: -0.23
    • P-value: 0.015
  • Liproxstatin-1
    • Correlation: -0.09
    • P-value: 0.33

Find MCS smarts

• Use MCS() against pairs of molecules to find the maximum common substructures
• Visualize smarts highlighted in the original molecule

Chemical Checker