in silico methods in drug discovery

In the simplest form of quantum mechanics calculations, what is the scaling factor for the calculations when the number of electrons is doubled?

• 16 fold

Which was the first true GPCR to have its crystal structure determined?

bovine rhodopsin

A technique that uses a training set of 3D structures, aligned in a grid is:

• CoMFA

emerging fields for computer based methods in later stages of drug discovery

• emerging fields

  • prediction of pharmacokinetics

  • prediction of toxicity/side effects

  • save time, money and reduce the number of animals used

SAR

structure activity relationship

• relationship between activity and some feature of its structure

• qualitative

QSAR

• mathematical relationship between physiochemical properties of a compound and its activity

• quantitative and predictive

QSAR properties

• quantitative (and predictive)

• need to do series of compounds where biological activity and properties is known, standard curve

  • then extrapolate for unknown and predict biological activity

• can be a non linear relationship

• can be more than one property

hammett constant

• qsar first done on antibiotics with different ring substituents

• tendency for electrons being withdrawn from ring meant higher antibiotic activity

log p

• analysis of lipophilicty of compounds

• higher log p, the more lipophilic

how is log p calculated

• octanol and water wont mix

• test compound

• shake

• measure how much is partitioned and how much is dissolved

• take log of these values divided

Examples of properties examined by QSAR

• molecular mass

• Log P

• Hammett constant

how are qsar properties measured

• fragmentation on paper, not physically breaking apart

• uses physiochemical parameters derived from fragments derived from experimental data

• sum of each fragment

computers in fragment based QSAR

• computers derive the QSAR equations

• databases to calculate molecular properties eg log p, using databases helps to speed the process up!

3 molecular modelling approaches for small molecules?

• quantum mechanics

• semi-empirical methods

• molecular mechanics

Quantum mechanics

• based on quantum physics

• describes positions and energies of electrons and nuclei

• lots of maths, computationally expensive so implemented for small molecules only

• doubling number of electrons 16x more calculations

Semi-empirical methods

• simplification of quantum mechanics

  • can be used for slightly larger molecules

• experimentally derived parameters used in parts of equation

molecular mechanics basic principles

• treats molecules as a collection of spheres and springs

  • springs are bonds

  • spheres are atoms

what influences energy and structure in molecular mechanics

Bond stretching

Bond torsion

Bond angle

Non bonded interactions

total energy in molecular mechanics

Total energy = stretch + torsion + angle + van der waals + electrostatic

• this has to be done for every atom

• molecular mechanics programs try to minimise energy of molecule

advantages of molecular mechanics programs

• relatively fast

  • can be used for:

  • small molecules

  • macromolecules eg peptides

  • docking to proteins

  • solvent effects

• derived energies have no absolute meaning but can be compared

• reasonably easy to understand

• cheap to implement

determination of target structure for larger molecules

• experimental methods

  • x ray crystallography

  • NMR ditto

  • cryo-electron microscopy - easier but still not trivial

• can predict secondary structure of small parts of protein

• can compare unknown structures with known structures

homology modelling

• use known crystal as template, align sequence of target

• use molecular mechanics to improve structure

examples of gpcr in homology modelling

• bovine rhodopsin

• human b2 adrenoreceptor

• human d3 dopamine receptor

homology modelling of ligand gated ion channels

• acetylcholine binding protein - extracellular domain of LGIC, water soluble so easier to crystallise

  • lots of studies use this as a starting point, allowing modelling of nicotinic and related receptors

• ELIC and GLIC for bacteria which then allowed more discovery

What can modelling tell us?

• mechanistic understanding

  • where does a ligand bind

  • how does it bind

• predictive models

  • provide data for conventional QSAR

  • provide data for 3D QSAR

  • docking experiments

CoMFA

• align set of similar, characterised molecules

• place them into a 3d grid

• calculate steric and electronic properties

• correlate with activity

• use relationships to predict activity of similar but uncharacterised compounds

• can take into account flexibility

example of use of comfa

• human steroid 5a reductase inhibitors

• all had a steroid backbone

• observed activity and predicted activity showed positive correlation

structure based in silico screening

• test compound docking to target

• work out energy between molecule and target

• can screen molecules that have never been synthesised

Where has structure based screening been used?

• inhibitors of anthrax toxin

• x ray structure

Building within the binding site for structure based in silico screening

• add components to the molecule to try and occupy and form interactions within the binding site

• uses DISCO

• understand dynamic protein structures such as p glycoprotein against chemotherapy resistance

Current status of in silico screening

• still retain biological assays, in vitro and in vivo

• filter down large chemical libraries - virtual screening

  • priortisation of predicted better assays