Structure prediction & homology modelling

12 important questions on Structure prediction & homology modelling

Explain how the concept “Structure is more conserved than sequence” is used.

If two sequences are similar, most likely these two proteins will have a very similar structure. ---> implies that --> if our sequence of interest is similar to a protein sequence with a known structure, we roughly know what its structure will be.

Give the 4 steps of homology modelling.

  1. Find a template
  2. Make an alignment between your sequence and the template
  3. Create a homology model (modeller)
    1. loop modelling
    2. rotameroptimisation
  4. Validate your structure

What is threading/fold recognition?

  • Fit your sequence onto existing structures (may not indicate homology):
  • We need a score to evaluate how well a sequence fits onto a structure
  • we use a “potential”
  • Note that threading is effectively an alignment problem between a sequence and a structure: Gaps may be introduced
  • Higher grades + faster learning
  • Never study anything twice
  • 100% sure, 100% understanding
Discover Study Smart

What is a pairwise residue contact potential?

  • Gives an energy between two residues
  • Can be calculated by considering the contacts of a residue
  • A contact is a  pair of residues that are close in 3D space
  • Two hydrophobics would give a low energy in such a potential (they are favourable)

When creating an alignment, what to look out for?

  • Is anything about the function of the target & template known?
  • Can we recognise any functional motifs? –align by hand!(or Motif Aware Praline)
  • general: MSA of profile alignment

What are the constraints in MODELLER based on?

MODELLER uses “knowledge based” constraints. The constraints are based on the template, but also on knowledge based energy functions. The constraints are optimised using molecular dynamics with simulated annealing.

What is simulated annealing?

  • Simulated using the constraints and knowledge based potential (simulated as E)
  • Lower the temperature (T) during the simulation (constraints and potential become more important)
  • This is a way to minimize the energy (potential) of a system


(entropy is left out)

What is done for structure refinement?

Given a rough model for the backbone, find set of best rotamers–hydrogen bonds etc.

How are all models by modeller evaluated and how can you pick the best one?

Score models on several properties:
  • (experimental structure validation)
    • hydrogen bonds
    • distances
    • phi/psi angles
  • for evaluating models:
    • globularity (compactness)
    • amount of secondary structure
    • amount of hydrogen bonds
    • protein "like" features


molprobity = good example

Meta-servers: what are ensemble methods?

Take results from several prediction method, and look for agreement

How accurate is Secondary Structure Prediction?

  • Secondary structure prediction is relatively accurate
  • Predict secondary structure for your target sequence
  • Compare to assigned structure for your (suggested) template

What is global distance test?

Alternative to RMSD, as this is sensitive to outliers.

GDT_TS --> total score from global distance test.
why less sensitive to outliers?

The question on the page originate from the summary of the following study material:

  • A unique study and practice tool
  • Never study anything twice again
  • Get the grades you hope for
  • 100% sure, 100% understanding
Remember faster, study better. Scientifically proven.
Trustpilot Logo