Skip to content

Physicochemical Properties, SAR and QSAR

CSIR NET medicinal chemistry: logP and logD, pKa and ionisation, Lipinski rules, pharmacophores, bioisosterism, Hammett-Hansch QSAR and ligand efficiency.

  • 17 min read
  • Updated
On this page · 15 sections
  1. Prerequisites
  2. The Hook: The Same Pharmacophore, Ten Different Fates
  3. Lipophilicity
  4. Acidity, Basicity and Hydrogen Bonding
  5. Drug-Likeness Rules
  6. From SAR to a Pharmacophore
  7. QSAR: Making the Relationship Quantitative
  8. Efficiency Metrics
  9. Memory Tricks
  10. Don’t Confuse
  11. Common Mistakes
  12. Practice
  13. Quick Revision Table
  14. Key Takeaways
  15. What’s Next

Prerequisites

The Hook: The Same Pharmacophore, Ten Different Fates

Lipophilicity

Partition coefficient

$$P = \frac{[\mathrm{drug}]_{\text{octanol}}}{[\mathrm{drug}]_{\text{water}}} \quad (\text{un-ionised species}), \qquad \log P \text{ is the usual scale}$$

Octan-1-ol/water is the convention: octanol has a hydrogen-bonding hydroxyl and a long chain, a crude but useful model of a membrane. \( \log P \) is additive to a first approximation, and the substituent constant

$$\pi_X = \log P_{\mathrm{R-X}} - \log P_{\mathrm{R-H}}$$

is the lipophilicity analogue of Hammett’s \( \sigma \): positive for alkyl, halogen and CF\(_3\); negative for OH, NH\(_2\), CONH\(_2\), SO\(_2\)NH\(_2\).

Distribution coefficient

For an ionisable compound, what matters at a given pH is \( D \), which counts all species in water:

$$\log D_{\mathrm{pH}} = \log P - \log\!\left(1 + 10^{\,\mathrm{pH} - \mathrm{p}K_a}\right) \ \ (\text{acid}); \qquad \log D_{\mathrm{pH}} = \log P - \log\!\left(1 + 10^{\,\mathrm{p}K_a - \mathrm{pH}}\right) \ \ (\text{base})$$

In both cases the subtracted term is \( -\log(\text{fraction un-ionised}) \): the exponent is written so that it is positive when the compound is mostly ionised.

This is the pH–partition idea: only the neutral form crosses a lipid membrane by passive diffusion, so weak acids are absorbed relatively better where the pH is low and weak bases where it is higher, while ionisation keeps the compound in solution and often improves its ability to reach a charged counter-group in the binding site.

Why both too little and too much are bad

\( \log P \) too low\( \log P \) too high
poor membrane permeationpoor aqueous solubility
rapid renal clearance of the unchanged drughigh plasma-protein binding, low free fraction
little non-specific binding, so often cleanmore metabolic oxidation, more promiscuous off-target binding

The result is an optimum, which is the physical origin of the parabolic term in the Hansch equation.

Acidity, Basicity and Hydrogen Bonding

GroupTypical character at pH 7.4
Carboxylic acidionised (anion)
Phenol, sulfonamide NH, tetrazoleweakly acidic; tetrazole is a common carboxylic-acid replacement of similar acidity but higher lipophilicity
Aliphatic amineprotonated (cation)
Aromatic amine, pyridinemostly neutral
Amideneutral; a hydrogen-bond donor and acceptor

The Henderson–Hasselbalch relation, \( \mathrm{pH} = \mathrm{p}K_a + \log([\mathrm{A^-}]/[\mathrm{HA}]) \), and nothing more, answers all the “fraction ionised” questions. Electron-withdrawing substituents lower the \( pK_a \) of an acid and of a protonated amine alike; the Hammett treatment of substituted benzoic acids is the quantitative version.

Hydrogen bonds pay for themselves only if the donor or acceptor is buried against a partner: every polar group costs desolvation energy, which is why adding an NH that finds nothing to pair with usually reduces affinity.

Drug-Likeness Rules

Lipinski’s “rule of five” (an empirical pattern for orally absorbed compounds, derived from marketed drugs — a guideline, not a law):

PropertyLimit
Molecular mass≤ 500
\( \log P \)≤ 5
Hydrogen-bond donors (OH + NH count)≤ 5
Hydrogen-bond acceptors (N + O count)≤ 10

Two or more violations flag likely poor oral absorption. Natural-product-derived and actively transported compounds are recognised exceptions.

Veber’s criteria add shape and polarity: rotatable bonds ≤ 10 and polar surface area ≤ 140 Å\(^2\) (often ≤ 90 Å\(^2\) for compounds intended to cross the blood–brain barrier). Fragment work uses the correspondingly scaled “rule of three”.

From SAR to a Pharmacophore

Structure–activity relationship (SAR) work is systematic: make analogues that change one feature at a time and read what the activity does.

TacticQuestion it answers
Delete a groupis it needed at all?
Replace with an isostereis its size, shape or electronics what matters?
Homologate (CH\(_2\) at a time)how long should the chain be?
Rigidify (ring closure)what conformation binds?
Change stereochemistrywhich enantiomer or diastereomer is active?
Scan substituents on a ringwhich positions tolerate bulk, which need electronics?

The distilled result is a pharmacophore: the three-dimensional arrangement of features — hydrogen-bond donor, acceptor, positive ionisable centre, aromatic ring, hydrophobic volume — required for activity, stripped of the scaffold that happens to carry them. Two chemically unrelated series sharing a pharmacophore is the basis of scaffold hopping.

Bioisosterism

A bioisostere replaces a group with one of similar size, shape, electron distribution or hydrogen-bonding pattern, keeping activity while improving a property.

ClassicalNon-classical
F for H; OH, NH\(_2\), CH\(_3\) interchanged; Cl for SHtetrazole or acylsulfonamide for COOH
O, S, NH, CH\(_2\) interchanged in a chainamide replaced by a trans-alkene or a 1,2,4-oxadiazole
N for CH in an aromatic ring (benzene → pyridine)cyclopropyl or bicyclo[1.1.1]pentyl for a phenyl ring

Typical motives: block a metabolically labile position (aromatic F), raise or lower \( pK_a \), add or remove a charge, improve solubility, or escape a patent.

QSAR: Making the Relationship Quantitative

The central assumption is an extrathermodynamic one: the free energy of binding, and so \( \log(1/C) \) where \( C \) is the concentration producing a standard response, is a sum of contributions from electronic, hydrophobic and steric features of the substituents.

The Hansch equation

$$\log\frac{1}{C} = -k_1(\log P)^2 + k_2\log P + k_3\sigma + k_4E_s + k_5$$
  • \( \log P \) or \( \pi \): hydrophobic term, usually parabolic (an optimum exists).
  • \( \sigma \): Hammett electronic term (\( \sigma^+ \) or \( \sigma^- \) where through-conjugation matters; Taft \( \sigma^* \) for aliphatic systems).
  • \( E_s \): Taft steric parameter; molar refractivity MR or Verloop’s sterimol parameters are the common alternatives.

Differentiating the parabola, the optimum lipophilicity is \( \log P_0 = k_2/(2k_1) \).

Choosing the analogues to make

MethodIdea
Craig plotplot \( \pi \) against \( \sigma \) for candidate substituents and pick from all four quadrants, so the two parameters are not accidentally correlated
Topliss decision treea non-statistical, stepwise scheme: make a small set, and let the result (more, equal or less active) dictate the next substituent
Free–Wilson (de novo) analysisactivity = a constant plus an additive contribution for each substituent at each position; no physical parameters needed, but no extrapolation outside the set either
3D-QSAR (CoMFA-type)align the molecules, compute steric and electrostatic fields on a grid, and regress activity on the field values by partial least squares

Colinearity is the standard pitfall. If every analogue with a bulky group also has a lipophilic one, the regression cannot tell the two effects apart. Craig plots exist to prevent this. A fit must also be validated: a high \( r^2 \) on the training set means little without cross-validation and a test set, and the number of descriptors must be small compared with the number of compounds.

Efficiency Metrics

Potency gained simply by adding mass is a poor bargain, because mass costs solubility and metabolic stability. Two normalisations are standard:

$$\text{Ligand efficiency } \mathrm{LE} = \frac{-\Delta G^\circ}{N_{\text{heavy atoms}}} \approx \frac{1.37\,\mathrm{p}\mathrm{IC}_{50}}{N_{\text{heavy atoms}}}\ \ (\text{kcal mol}^{-1}\text{ per heavy atom, 298 K})$$$$\text{Lipophilic ligand efficiency } \mathrm{LLE} = \mathrm{p}\mathrm{IC}_{50} - \log P$$

Memory Tricks

Don’t Confuse

ThisvsThat
\( \log P \) (un-ionised partition)\( \log D \) (apparent, pH-dependent)
\( \pi \) (lipophilicity of a substituent)\( \sigma \) (electronic effect of the same substituent)
Pharmacophore (3D feature arrangement)Scaffold (the framework carrying them)
Bioisostere (property kept, group changed)Analogue (any structural relative)
Hansch (physicochemical parameters)Free–Wilson (purely additive group contributions)
Potency (pIC\(_{50}\))Efficiency (LE, LLE — potency per atom or per log P)
Lipinski rule of fiveVeber criteria (rotatable bonds, polar surface area)
\( r^2 \) on the training setCross-validated \( q^2 \) and performance on a test set

Common Mistakes

Practice

All questions below are original practice questions written for this page. They are not from any CSIR NET paper.

Level 1: Foundation

  1. Define \( \pi \) for a substituent.
  2. Write the Lipinski limits.
  3. What does a negative coefficient of \( (\log P)^2 \) in a Hansch equation imply?
Answers
  1. \( \pi_X = \log P_{\mathrm{R-X}} - \log P_{\mathrm{R-H}} \): the change in \( \log P \) caused by replacing H with X.
  2. Mass ≤ 500, \( \log P \) ≤ 5, H-bond donors ≤ 5, acceptors ≤ 10.
  3. The activity passes through a maximum at \( \log P_0 = k_2/(2k_1) \).

Level 2: Part B Style

Practice Q1. A base has \( \log P = 3.0 \) and \( \mathrm{p}K_a = 9.4 \). Its \( \log D \) at pH 7.4 is closest to:

(a) 3.0 (b) 1.0 (c) 0.0 (d) −2.0

Answer: (b) 1.0

For a base, \( \log D = \log P - \log(1 + 10^{\mathrm{p}K_a - \mathrm{pH}}) = 3.0 - \log(1 + 10^{2.0}) = 3.0 - 2.00 = 1.00 \). The compound is about 99% protonated at pH 7.4. Option (a) ignores ionisation; option (d) applies the acid form of the correction.

Practice Q2. Which is a non-classical bioisosteric replacement for a carboxylic acid?

(a) An ester (b) A tetrazole (c) An aldehyde (d) A primary alcohol

Answer: (b) Tetrazole

The tetrazole is acidic in a similar range, planar and anionic at physiological pH, but generally more lipophilic. It is the standard non-classical replacement; an ester is a different functional group (and usually a prodrug).

Practice Q3. For the fit \( \log(1/C) = -0.25(\log P)^2 + 1.50\log P + 2.0 \), the optimum \( \log P \) is:

(a) 1.5 (b) 3.0 (c) 4.0 (d) 6.0

Answer: (b) 3.0

\( \log P_0 = k_2/(2k_1) = 1.50/0.50 = 3.0 \). Option (d) forgets the factor 2.

Practice Q4. A compound has mass 620, \( \log P \) 6.2, 3 H-bond donors and 7 acceptors. The number of Lipinski violations is:

(a) 0 (b) 1 (c) 2 (d) 3

Answer: (c) 2

Mass > 500 and \( \log P \) > 5 are violations; donors and acceptors are within limits. Two violations flag probable poor oral absorption.

Level 3: Part C Style

Practice Q5. A series of para-substituted analogues gives \( \log(1/C) = 1.60\,\sigma + 0.80\,\pi + 3.00 \) (\( n = 14 \), \( r = 0.96 \)). Using \( \sigma_p \) and \( \pi \) values of 0.78 and −0.28 for NO\(_2\), 0.23 and 0.71 for Cl, −0.27 and −0.02 for OCH\(_3\), which statements are correct?

  1. The predicted \( \log(1/C) \) for the p-NO\(_2\) analogue is about 4.02.
  2. The p-Cl analogue is predicted to be more active than the p-OCH\(_3\) analogue.
  3. Electron-donating substituents increase activity.
  4. The absence of a \( (\log P)^2 \) term means no optimum lipophilicity was detected over the range studied.

(a) 1, 2 and 4 only (b) 1 and 3 only (c) 2, 3 and 4 only (d) 1, 2, 3 and 4

Answer: (a) 1, 2 and 4 only
  1. \( 1.60(0.78) + 0.80(−0.28) + 3.00 = 1.248 − 0.224 + 3.000 = 4.02 \). Correct.
  2. Cl: \( 1.60(0.23) + 0.80(0.71) + 3.00 = 3.94 \); OCH\(_3\): \( 1.60(-0.27) + 0.80(-0.02) + 3.00 = 2.55 \). Correct.
  3. The positive \( \sigma \) coefficient means electron-withdrawing groups increase activity. Wrong.
  4. A linear \( \pi \) term means activity was still rising at the top of the range sampled; an optimum may lie outside it. Correct.

Practice Q6. Three leads against the same enzyme:

LeadIC\(_{50}\)Heavy atoms\( \log P \)
X1.0 µM151.5
Y10 nM253.0
Z50 nM385.2

Which statements are correct? (Use LE \( \approx 1.37\,\mathrm{pIC}_{50}/N \) kcal mol⁻¹ per heavy atom.)

  1. Y has the highest ligand efficiency of the three.
  2. Z has the lowest lipophilic ligand efficiency.
  3. X has the highest ligand efficiency.
  4. Z would fail a Lipinski \( \log P \) criterion.

(a) 1, 2 and 4 only (b) 2, 3 and 4 only (c) 1 and 4 only (d) 2 and 3 only

Answer: (b) 2, 3 and 4 only

LE: X = \( 1.37 \times 6.0/15 = 0.55 \); Y = \( 1.37 \times 8.0/25 = 0.44 \); Z = \( 1.37 \times 7.30/38 = 0.26 \). So X is highest (statement 3 correct, 1 wrong). LLE: X = 6.0 − 1.5 = 4.5; Y = 8.0 − 3.0 = 5.0; Z = 7.30 − 5.2 = 2.1 — Z lowest. Correct. Z has \( \log P = 5.2 > 5 \). Correct.

Practice Q7. An acid (\( \log P = 3.5 \), \( \mathrm{p}K_a = 4.5 \)) is highly potent in an isolated-enzyme assay but weak in whole cells. A tetrazole analogue (\( \mathrm{p}K_a \) similar, \( \log P = 4.1 \)) is more potent in cells. Which statements are consistent?

  1. At pH 7.4 both compounds are >99% ionised.
  2. The tetrazole’s higher \( \log P \) gives it a higher \( \log D_{7.4} \), aiding membrane passage.
  3. The cell-assay discrepancy for the carboxylic acid can be explained by poor permeability rather than by weaker target binding.
  4. Measuring \( \log P \) alone would have predicted the cell potencies correctly.

(a) 1, 2 and 3 only (b) 1 and 4 only (c) 2, 3 and 4 only (d) 1, 2, 3 and 4

Answer: (a) 1, 2 and 3 only
  1. \( 1/(1 + 10^{4.5-7.4}) = 0.9987 \) for both. Correct.
  2. \( \log D_{7.4} = \log P - \log(1 + 10^{2.9}) = \log P - 2.90 \): 0.60 versus 1.20. Correct.
  3. A compound that cannot reach the intracellular target is weak in cells however good its \( K_i \). Correct.
  4. \( \log P \) ignores ionisation; \( \log D \) at the relevant pH is the predictive quantity. Wrong.

Quick Revision Table

IdeaFormula / fact
Partition\( P = [\mathrm{D}]_{\text{oct}}/[\mathrm{D}]_{\text{aq}} \) (neutral form); \( \pi_X = \log P_{\mathrm{RX}} - \log P_{\mathrm{RH}} \)
Distribution (acid)\( \log D = \log P - \log(1 + 10^{\mathrm{pH} - \mathrm{p}K_a}) \)
Distribution (base)\( \log D = \log P - \log(1 + 10^{\mathrm{p}K_a - \mathrm{pH}}) \)
Ionised fraction (acid)\( 1/(1 + 10^{\mathrm{p}K_a - \mathrm{pH}}) \)
Lipinskimass ≤ 500, \( \log P \) ≤ 5, donors ≤ 5, acceptors ≤ 10
Veberrotatable bonds ≤ 10, PSA ≤ 140 Å\(^2\)
Hansch\( \log(1/C) = -k_1(\log P)^2 + k_2\log P + k_3\sigma + k_4E_s + k_5 \)
Optimum\( \log P_0 = k_2/(2k_1) \)
Design toolsCraig plot (π vs σ), Topliss tree, Free–Wilson, CoMFA
Ligand efficiency\( \mathrm{LE} \approx 1.37\,\mathrm{pIC}_{50}/N_{\text{heavy}} \) kcal mol⁻¹ per atom
Lipophilic efficiency\( \mathrm{LLE} = \mathrm{pIC}_{50} - \log P \)
BioisosteresCOOH → tetrazole; H → F; phenyl → pyridyl; amide → oxadiazole

Key Takeaways

What’s Next