These pages are not displaying properly because the Compatibility View in your Internet Explorer is enabled. We suggest that you remove 'fu-berlin.de' from your list of sites that have Compatibility View enabled.

In Internet Explorer, press the 'Alt' key to display the Menu bar, or press and hold the address bar and select 'Menu bar'.

Click 'Tools' and select 'Compatibility View settings'.

Select 'fu-berlin.de' under 'Websites you've added to Compatibility View'.

Click 'Remove'.

Computer aided drug design (Structural Bioinformatics)

Fill in the following text by selecting the correct solutions in the drop down menus. Afterwards you can check your answer with the 'Result' button.

Imagine you want to find or even design a new drug for example against a certain cancer. A good starting point is then to investigate which metabolic pathways are associated with that disease (for example in KEGG) and choose a compound in that pathway which can be influenced to change its function without having adverse effects elsewhere.
Take for example the epidermal growth factor receptor (EGFR) which regulates cellular proliferation and when dysregulated can cause lung cancer. EGFR is a tyrosine kinase which is located in the

and upon binding a ligand from outside can dimerize and

molecules inside thereby starting a signalling cascade. Several drugs which can bind and thereby block the ATP binding site of EGFR have been developed and approved since 2003 but effectiveness can still be improved.
Now the number of available compounds which could be tested as a new drug are in the billions but only a small fraction of them could actually bind to our protein of interest and even less of those bind with the desired effectiveness. Testing new drugs takes years and millions of Euros (or any other currency you like) so here you need bioinformaticians to step in and provide a manageable set of

which are most likely to perform the required activity.
The first step when looking for ligands for a certain protein is to determine the structure of that protein and ligands. This includes primary structure (

of amino acids), secondary structure (e.g. alpha-helices, beta-sheets,

), tertiary structure (complete 3D folding) and quaternary structure (3D formation of complexes containing several monomers). When the 3D folding can't properly be measured in physical experiments (e.g. X-ray crystallography or NMR), for example if the protein can't be properly crystallized or the

is too low, we can often fall back on homology modelling (predicting structure via homologous protein).
Identifying binding sites can be done via geometry-based methods (grid-based, sphere-base) or

(carbon probe or docking).
Since there are already enough compounds which have been tested for their binding affinity we now can train a machine learning model which can predict active versus inactive compounds. Another filtering step can be to test if the predicted active compounds have physical and chemical properties like

. Here we can check the molecular weight, number of hydrogen bond donors or acceptors and hydrophobicity.
If we don't want to fall back on prior knowledge we can still perform docking to predict the

between compound and protein of interest. The binding affinity tells us if the compound can bind at all and for how long it can block the binding site for other molecules thus how effective the drug can be.
In the end we shouldn't forget to look for possible

, proteins with similar binding-sites like the tested protein, which can cause side-effects in the body.

"Imagine you want to find or even design a new drug for example against a certain cancer. A good starting point is then to investigate which metabolic pathways are associated with that disease (for example in KEGG) and choose a compound in that pathway which can be influenced to change its function without having adverse effects elsewhere.

Take for example the epidermal growth factor receptor (EGFR) which regulates cellular proliferation and when dysregulated can cause lung cancer. EGFR is a tyrosine kinase which is located in the membrane and upon binding a ligand from outside can dimerize and phosphorylate molecules inside thereby starting a signalling cascade. Several drugs which can bind and thereby block the ATP binding site of EGFR have been developed and approved since 2003 but effectiveness can still be improved.

Now the number of available compounds which could be tested as a new drug are in the billions but only a small fraction of them could actually bind to our protein of interest and even less of those bind with the desired effectiveness. Testing new drugs takes years and millions of Euros (or any other currency you like) so here you need bioinformaticians to step in and provide a manageable set of good candidates which are most likely to perform the required activity.

The first step when looking for ligands for a certain protein is to determine the structure of that protein and ligands. This includes primary structure (sequence of amino acids), secondary structure (e.g. alpha-helices, beta-sheets, loops), tertiary structure (complete 3D folding) and quaternary structure (3D formation of complexes containing several monomers). When the 3D folding can't properly be measured in physical experiments (e.g. X-ray crystallography or NMR), for example if the protein can't be properly crystallized or the resolution is too low, we can often fall back on homology modelling (predicting structure via homologous protein).

Identifying binding sites can be done via geometry-based methods (grid-based, sphere-base) or energy-based (carbon probe or docking).

Since there are already enough compounds which have been tested for their binding affinity we now can train a machine learning model which can predict active versus inactive compounds. Another filtering step can be to test if the predicted active compounds have physical and chemical properties like approved drugs. Here we can check the molecular weight, number of hydrogen bond donors or acceptors and hydrophobicity.

If we don't want to fall back on prior knowledge we can still perform docking to predict the binding affinity between compound and protein of interest. The binding affinity tells us if the compound can bind at all and for how long it can block the binding site for other molecules thus how effective the drug can be.

In the end we shouldn't forget to look for possible off-targets, proteins with similar binding-sites like the tested protein, which can cause side-effects in the body."