Validation

Finding these really bad errors in the previous step in a very small protein was already difficult. Try to imagine finding much more subtle errors in big proteins the same laborious way... Fortunately, there are some facilities to help you. Take the four proteins listed above and send them to the WHAT IF server for protein structure validation. See if the report would have helped you find the errors quicker .

I am aware that the C-terminal extra oxygen is missing, and that the backbone of all four proteins is terrible. That is because I made these structures by hand.

Figure 54. Normal distribution. Validation is very much based on statistics, and the normal distributions is the central vehicle in those statistics.

While we are at these validation reports, there is a database of errors in protein structures. This database is called PDBREPORT, and it is made with the WHAT_CHECK software that is also available as a WHAT IF based web-server. Find the PDBREPORT database and the server and use them (and their documentation) to answer some of the questions listed below.

Question 62: Which section(s) in the report for hypo_bad_01.pdb was most informative about the error in this structure?

Question 63: Which section(s) in the report for hypo_bad_02.pdb was most informative about the error in this structure?

Question 64: Which section(s) in the report for hypo_bad_03.pdb was most informative about the error in this structure?

Question 65: Which section(s) in the report for hypo_bad_04.pdb was most informative about the error in this structure?

Question 66: If you call things wrong when they deviate more than 3σ from the theoretical mean, how big is then the risk that you call something wrong that is actually right?

Question 67: What is a Z-score?

Question 68: Most checks compare aspects of one structure with the average of a series of well-solved structures. But for bond-lengths and bond-angles the PDBREPORT seems to have better values. How come we know the perfect bond-length?

Question 69: Which checks are useless for homology models?

Question 70: What is packing quality ?

Question 71: What are side-chain flips? And how come we find so many of those in the PDB?