Lecture 17 Table 2

Question 1

For resolution range what do we expect for the low resolution and high res shell

Answer 1

We expect the lowest value for refold ion to be the same for both high and low wheels

Ex. 43.5-1.7 (1.8-1.7) both have 1.7 as lowest

Question 2

Completeness in brackets means what

Answer 2

It has that completeness for data points in the high resolution Range

Question 3

What is sigma cutoff

Answer 3

The number of reflections that were rejected

Question 4

What is working and test set

What is Rcryst and Rfree

What is the hand wavey trick

Answer 4

Working is the unique reflections they use in analysis

Test is data they took out (want at least 1000)

Rcryst is calculated from working set

Rfree is calculated from Test set

Given as percent

If highest resolution they could get is 1.7 , the Rcryst they should get is 17%

Question 5

For each non hydrogen atoms (protien,ion, water) what parameters are being refined

Answer 5

4 parameters:
- The xyz postions, and b factor

So multiply 4 by the total non H atoms, that value should be 3x less than the number of reflections in the working set

Question 6

Rms deviations should be what

Answer 6

0.02 for bonds

2 for angles

Question 7

What is there to note about the average b factors for the protein

What does this not apply to

What is special about the hydrogens

Answer 7

The avg B factor for the protien is restrained based on connectivity :
- c beta bound to alpha, they’d have similar b factors because bonded and the bond angles are same

Doesn’t apply to ions and water :
- ions are single ions like sulphate and phosphate which has their own b factors values

Hydrogens don’t scatter so we don’t have them in the model meaning we don’t refine them. We put them in the model as “riding hydrogens” where we calculate the spots they should go in.

Question 8

Overall avg b factors and Wilson b factor comparing

Answer 8

If compared the two value should be pretty close for us to say it’s a good model

Also overall avg should be similar to protiens avg b factors for

Question 9

For the ramachandran values, if percent is missing Whag does this mean

Answer 9

If don’t add to 100% this mean some percent of aA were neither in most favoured or allowed

Also many sure is not a high percent that is in the neither catergory

Question 10

What is the relation between higher res data and unique reflections

Answer 10

Higher resolution data gives us more unique reflections

Question 11

What does it mean if there is a 0 for number of water molecules

Answer 11

The couldn’t find electron density in the protien for hydrogen bonding interactions with water so they didn’t model the water

Question 12

What is the limit for Rwork and Rfree

Answer 12

34%/ 3.4 A

Anything Higher than this is bad

Question 13

What are rotamer outliers

Answer 13

The data for aA whos bond angles are highly improbable

Question 14

If you chose a higher value for sigma how much electron density would you see

Answer 14

Less electron density

Question 15

Explain the first example of when things go wrong (2001 paper)

Answer 15

They added OsCl3 to make the crystal diffract better and get better resolution

Since their resolution was low their Rfree was bad so they decided to model 16 models overtop of each other which lowered the R values

Basically did a lot of operation to improve electrons density so they could model it

2006 paper:
- another paper used the 2001 model and compared it to their model, they didn’t match
- they inverted the 2001 model, changing the chirality, and it matched
- so 2001 structure had wrong chirality due to the program they used which inverted the intensity
- this is what made their R values bad so they needed 16 copies

Question 16

Explain the second example of when things go wrong

Answer 16

They had good data but they could not solve their structure

This is because they had the wrong space group which mixed up the unit cell dimensions

Messed up interpretation of data

Question 17

Explain the third example of when things go wrong

Answer 17

The modelled a protien that Diffracted to 2.8A resolution

Someone went in the PDB and tested if they could refine that data better

He found they the r values that were reported were different from the one he got. He got much lower values.

Also the wilson B factor and avg B factor for protien were not the similar to each other in the original model

They retracted their paper because of scientific misconduct because they made up the numbers

Question 18

Explain. Retractions

Answer 18

2 of every 1000 papers get retracted (low) and it not always cause of scientific misconduct (can be due to mistakes)