How to analyse a given molecule's binding affinity to DNA

Question

[Min Dai]

Hi,

I want to know whether a given molecule can bind to DNA or not, is there any recommend software or tool? The DNA here can be single-stranded, double-stranded or some with complex structure, such as G-quadruplex.

Thanks in advance!

BP

Answer 1

[matteoferla]

A good programmer is a lazy one. First load your protein structure in PyMOL, look at the APBS surface, if you see positive patches (blue) that is potentially a bunch of arginines that coordinate the bound phosphate backbones —often as a very shallow groove. Then on the other hand they could be phospholipid binding...

Location

If Eukaryotic, is your protein properly known to be nuclear?

What does Uniprot say? (Caveat: propagated results often became erroneous)

If nothing, what does ELM DB predict?

A membrane associate protein might have a patch of hydrophobicity with a charged ring around it on a spot that is most elevated. But not always, DOCK180 DHR1 is one example, where it pulls the membrane in a weird way. Transmembrane proteins are obvious, but their di-arginine motif close to the membrane region is quite studied.

DNA-binding protein are often crystallised with DNA in them as they are very flexible unbound. What does the B-factor putty look like in the "blue groove"?

Docking DNA

There are many docking tools. Each different force-fields with different applications. So no answer for this will be the same....

I use the features within Rosetta and its Python module, Pyrosetta. There is a tutorial on how to dock two macromolecules.

Caveats

However, if either of your structures is an threaded+ab initio model (from say Phyre, I-Tasser etc.) the loops and some other bits will be in wild places, which will throw off the calculations. So docking in general is a no go.

It does not apply here, but it should be said (for other readers) that predicting specific nucleobases is not possible as the difference between the bases result in only a few kcal/mol difference.