Phenix (Python-based Hierarchical ENvironment for Integrated Xtallography) has been developed for the determination of macromolecular crystallographic structures. Phenix is a leader in automating the structure determination process.
I am going to start playing around with Phenix, but first wanted bring together helpful resources about the program suite:
This is a follow up to the post Is the Beam Center Correct? which shows how to determine if your beam center is correct. Today, we will be looking at how correct the beam center if it is wrong.
The computer screen is not as clear as I would like, but think you will be able to follow along (watching the video full screen should help). The audio is a little hard to hear at the end since I am speaking directly behind the camera. Anyway, I would love to hear your thoughts.
Here is the post on the format of a comm file. We also had a post on the overview of using ipmosflm, which should help you follow along with this post if you get stuck. When I bring up ‘previous set’ am referring to the offset seen in the Is the Beam Center Correct? video.
I would love to hear some feedback on this! Can you follow along? Do you like having videos? I would also love hear suggestions on screen recording software (linux or windows).
We have had a couple of instances when the beam center was incorrect in the header of our images. The header of an image is written out when you collect a diffraction image. The header may contain the wavelength, collection time, beam center and oscillation step size, which are read by data processing software (in this case Ipmosflm).
Unfortunately, the information contained within the header may not be correct.
How can you look for to tell if your beam center is incorrect?
One way is by looking at where the Bragg reflections are predicted vs. where they actually are located. It is important to check if the orientation of the lunes and the spacing between Bragg reflections looks reasonable. If they are not then you may an issue with your unit cell and/or space group and not a beam center problem.
If you are having trouble seeing the offset in the video here is a screen shot:
The blue arrow is pointing toward the Bragg reflections while the red arrow indicates the predicted reflection locations. Also you can see that there are fewer Bragg reflections (dark spots) than what is predicted (yellow squares) which is a sign that there is a problem.
CNS (Crystallography and NMR systems) is able to perform simulated annealing to get started, one must first create a generate file.
1) Input then scroll down to Refinement, refine.inp and Edit
2) amy.pdb needs to be replaced with your pdb file
3) The space group, unit cell, angles and amy.cv need to be updated
4) Adjust the resolution to your desired range. The overall B-factor correction should be set to isotropic unless you are dealing with very high resolution data (~1 Angstrom). Set Bulk solvent correction needs to be set to False
5) Change annealing schedule to slowcool
Note: Not shown, but I usually set the map grid to 0.25 for better viewing
Save an updated file
In your terminal:
type: cns < refine.inp > refine.out &
Note: if you renamed your generate files then use them as your .inp
The ‘&’ symbol allows your cursor to be free
type: tail -f refine.out
This will allow to see the progress of the processing in your terminal
This allows you to quickly see if the inputs have generate an error
CNS (Crystallography and NMR systems) is able to generate a composite omit map. In getting started, one must first create a generate file.
1) Input then scroll down to Refinement, composite_omit_map.inp and Edit
2) Three ‘amy’ files need to be placed with your appropriate files
The space group, unit cell and angles also need to be updated (I find this slightly annoying since the information is contained in the files you are submitting)
3) Bulk solvent correction needs to be set to False
4) I suggest putting the map grid at 0.25, raising the starting temp to 1500 and 50 K steps. I have come across those that adjust the random number generator, but haven’t noticed a huge difference.
Save an updated file
In your terminal:
type: cns < composite_omit_map.inp > composite_omit_map.out &
Note: if you renamed your generate files then use them as your .inp
The ‘&’ symbol allows your cursor to be free
type: tail -f composite_omit_map.out
This will allow to see the progress of the processing in your terminal
Doing this has allowed me to quickly see if my inputs have generate an error
Artemcomments:
late graduate students and Postdocs are vulnerable…
the challenge is to stick out from the background of young researchers…
Vulnerable in the sense of being attacked
The logical response is to defend, but be careful how you do so
The reflex is to follow protocol, keep your head down and get a paper published. What other options do you have? You aren’t in charge, it’s not your lab.
Getting ready to move on, you search for job openings, ask your boss who they know, polish the CV and get ready for interviews.
The job opening appears and 300 people apply. A mountain of applications, all saying the same thing.
Maybe the riskiest thing you can do, is not take any risks at all.
This is the part of the post where you want 10 things to try
This is the part of the post where we list what is remarkable
This is the part where I don’t tell you what you are capable of
1) Open the desired coordinate files in Coot (click here if you need some help) (You know you have been looking at structures too long when they start to look like faces.)
2) Under Calculate you have two methods of superimposing:
SSM Superpose (we will go with this option in this example: Calculate -> SSM Superpose…) or LSQ Superpose Note: SSM Superpose stands for secondary-structure matching and if you need to do it outside of Coot there is a server.
3) Select which PDB you would like to move and apply:
The structures should now be superimposed:
CNS (Crystallography and NMR systems) is able to perform simulated annealing as well as generate a composite omit map, which is nice compliment to the CCP4 suite of programs (Phenix has similar features). In getting started, one must first create a generate file which is what the post will be covering via their website.
1) In your terminal type: cns_web (note: admin may have set up this command different)
Web browser should launch and select input files at middle left
2) Scroll down to generate.inp and click edit
A new window will open
3) Change ‘convert chainid to segid if chainid is left blank’ to True
Scroll down a ways to ‘general parameters’
4) Change: set bfactor flag to True AND set occupancy flag to True
Save and Exit
In your terminal
type: cns < generate.inp > generate.out &
Note: if you renamed your generate files then use them as your .inp
The ‘&’ symbol allows your cursor to be free
type: tail -f generate.out
This will allow to see the progress of the processing in your terminal
Doing this has allowed me to quickly see if my inputs have generate an error
Note: Depending on your needs using generate_easy.inp may be sufficient
Additional information can be found in the tutorial section of the CNS website.
When giving presentations, scientists typically don’t have trouble making their complicated work seem…well, complicated. The challenge is in making your message clear and audience appropriate.
Example:
How does life expectancy at birth and the number of children a woman has change by country over a period of one hundred years?
I felt a table with 8 point font coming on, but instead was amazed by this.
Hint: Hit play
Would making a graph like this be helpful in crystallography? Perhaps showing PDB entries by country over the last 60 years? I realize this tool may not be the most useful to our community, but it’s helpful in the sense that it inspires creativity. I now find myself contemplating how I can present research in a way that is clear, concise and creative. How can I help my data tell a story?
Tableau allows for the searching of protein folding patterns of substructures in the PDB structural database. This type of searching can be helpful in understanding protein structure, function and evolution.
The server searches using secondary structure elements and is capable of finding either an entire structure or a substructure of a larger structure (ref).
Tableau has a number of simple inputs (title, email, desired structure, output), however, you want to read the suggested tips. You can submit structures that contain multiple chains or domains, but it is not recommended. I submitted a number of structures and had a response time of about 3 minutes.
(You know you have been looking at structures too long when they start to look like faces.)
Note: SSM Superpose stands for secondary-structure matching and if you need to do it outside of Coot 
