MCHM3001 · From Molecules to Therapeutics
Hit-to-Lead: Synthesis & Chemical Space
Lectures 12–13 of MCHM3001 turn a hit into a lead series: how to make many analogues quickly (solid-phase and combinatorial chemistry), how to represent and compare molecules (SMILES/InChI, fingerprints and the Tanimoto coefficient), and the developability filters (Lipinski, Veber and their relatives) that a lead must pass. The Tanimoto calculation and violation counting are prime exam material and recur in Test 2 and the final.
What this chapter covers
- 01Hit-to-lead: triage HTS hits into a lead series; classical one-reaction-one-compound synthesis is slow
- 02Solid-phase synthesis: attach to resin (Merrifield polystyrene), react, filter/rinse, cleave; Fmoc protection; Wang/Sasrin linkers
- 03Combinatorial chemistry and split-pool (divide-couple-recombine) generating nˣ compounds; deconvolution and encoding
- 04Chemical space (10¹⁴–10²⁰⁰ estimated) and databases (ChEMBL, PubChem, ZINC); GDB-17
- 05Molecular representation: SMILES (line notation), InChI/InChIKey, registry IDs
- 06Fingerprints (MQN, ECFP) and the Tanimoto coefficient T = c/(a+b+c); 'similar' if T > 0.85 (not a bioactivity guarantee)
- 07Lipinski rule of 5 and Veber (rotatable bonds ≤10, PSA ≤140 Ų); Pfizer 3/75, GSK 4/400 rules
- 08Lead-series criteria and efficiency metrics (LE, LipE/LLE); frequent hitters and toxicophores to avoid
Tanimoto similarity from two fingerprints
- +1Identify the terms: c = bits set in both = 6; a = bits set only in A = 8 − 6 = 2; b = bits set only in B = 8 − 6 = 2.
- +1Substitute into T = c/(a + b + c) = 6/(2 + 2 + 6).
- +1Evaluate: T = 6/10 = 0.60.
- +1Interpret against the taught threshold: molecules are usually called 'similar' when T > 0.85. Here T = 0.60 < 0.85, so the pair is not similar by that rule — and even a T above 0.85 does NOT guarantee similar bioactivity (the 'similarity paradox').
Key terms
- Solid-phase synthesis
- Attaching a substrate to an insoluble resin bead (e.g. Merrifield polystyrene), reacting and simply filtering/rinsing between steps, then cleaving the product; enables rapid parallel and combinatorial synthesis.
- SMILES / InChI
- Text representations of molecular structure: SMILES is a compact ASCII line notation with implicit hydrogens; InChI is a layered, standardised identifier (with a hashed InChIKey) for database lookup.
- Molecular fingerprint
- A bit-vector encoding a molecule's substructures (e.g. ECFP, 1024-bit); the basis for similarity comparisons such as the Tanimoto coefficient.
- Tanimoto coefficient (T)
- A similarity measure between two fingerprints, T = c/(a+b+c), where c is bits common to both, a bits only in the first and b bits only in the second; a value above ~0.85 is conventionally 'similar' but does not guarantee similar activity.
- Veber rule
- A developability guideline: good oral bioavailability is likely when rotatable bonds ≤ 10 and polar surface area ≤ 140 Ų; PSA often predicts permeability better than cLogP.
- Lipophilic efficiency (LipE/LLE)
- A metric rewarding potency that is not just bought with lipophilicity: LipE = pIC50 − logP; a higher value means affinity from genuine interactions rather than greasy bulk.
Hit-to-Lead: Synthesis & Chemical Space FAQ
How do you compute a Tanimoto coefficient from two fingerprints?
Count three things: c, the bits set in both molecules; a, the bits set only in the first; and b, the bits set only in the second. Then T = c/(a+b+c) — the intersection over the union. If A and B each have 8 bits set and share 6, then a = b = 2, c = 6, so T = 6/10 = 0.60. The commonest error is using the full bit counts for a and b instead of the bits unique to each.
If two molecules have T above 0.85, will they have the same activity?
Not necessarily. A Tanimoto value above about 0.85 means the molecules are structurally similar by their fingerprints, but the 'similarity paradox' is that small structural changes can still cause large activity changes (an activity cliff). So high Tanimoto similarity raises the odds of comparable activity but never guarantees it, which is why it is a triage tool, not a proof.
Why does solid-phase synthesis speed up making a lead series?
Because purification collapses to a filtration. The growing molecule is anchored to an insoluble resin bead, so after each reaction you just wash away the excess reagents and by-products and move on, without chromatography at every step. That makes it practical to run many parallel reactions and to build combinatorial libraries, then cleave the finished products from the resin at the end.
Can AI help me with the hit-to-lead and chemical-space material?
Yes. Sia can walk you through a Tanimoto calculation, count Lipinski and Veber violations, explain SMILES versus InChI, or contrast the efficiency metrics (LE versus LipE). It explains the method and checks your working; it does not do graded assessment for you, and University of Sydney academic-integrity rules apply.
Exam move
Drill the two examinable calculations first: the Tanimoto coefficient (extract a, b and c cleanly, then intersection over union) and violation counting against both Lipinski and Veber. Keep the developability rules on one page — Lipinski's four limits, Veber's rotatable-bond and PSA cut-offs, and the one-line 3/75 and 4/400 heuristics — and pair each efficiency metric (LE, LipE) with the failure it guards against. Be able to explain, in a sentence, why solid-phase and split-pool chemistry make lead series fast and why high Tanimoto similarity never guarantees activity. When the metrics or notations blur, ask Sia for fresh fingerprints and structures to score.
Working through Hit-to-Lead: Synthesis & Chemical Space in MCHM3001? Sia is AskSia’s AI Chemistry tutor — ask any MCHM3001 Hit-to-Lead: Synthesis & Chemical Space question and get a clear, step-by-step explanation grounded in how MCHM3001 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
