Protein structure: Reality check
The computergame fold.it is a very successful citizen science project to inquire the complicated structures of proteins. But It can propose some protein structures, "in silico", which are not found in the real world. Summerstudent Rafal, who worked in the MX-team during summer, told me: "fold.it is fine, and I am sure that such simulations will become better in the future. But the reality check are always the x-ray experiments!" And he mentioned some examples, where results of "in silico" modeling were quite different from structures found with X-Ray diffraction.
So we still need experiments to find out about the real protein structure. This is not trivial: You need to grow a protein crystal first, which is already an art. Only then, you can submit them at the MX-Beamlines at BESSY II to x-ray diffraction and find out how they are folded. Hundreds of samples can be put through, so this process is already highly automatised. But this creates a lot of data....
Now, a new paper by the MX-team and their partners at Phillips-University in Marburg demonstrated how a newly developed expert system might help with screening hundreds of data sets: it can even automatically identify promising molecules which are binding at a specific protein. This might accelerate drug design.
The newly developed expert system did analyse 364 data sets and found several good candidates for drug development. Credit:Structure 2016: doi: 10.1016/j.str.2016.06.010