Content:


Introduction

Here are some maybe more site-specific notes for the CCP4/Shanghai Workshop held in October 2019.


Running jobs

Our software is usually driven by commands within a terminal (shell) - although there is a task for running autoSHARP within the CCP4i (not CCP4i2) interface. We recommend using that command-line interface, since it gives you full access to all features and tools of our software packages. Once you get used to this way of running programs it will become a very powerful way of running different programs (or different trials) on a large number of datasets.

You should be able to run jobs exactly as they are presented on the tutorial pages (for SHARP/autoSHARP or autoPROC) and described on the reference cards (autoSHARP, autoPROC) and in the manuals (SHARP/autoSHARP, autoPROC).


Tutorial

For the tutorial we will use one additional settings that should not necessarily be used under normal conditions: it is only used to provide faster execution (so we have results to look at within the short amount of time available). This means setting the -fast flag:

    run_autoSHARP.sh -fast ...

We are going to use some Se-MET MAD data for this tutorial: 3ISY. We have the following files:

Please download those files onto your computer: right mouse click and then "Save Link As...". This should save these 3 files into your ~/Downloads folder.

Next we want to create a directory via

    mkdir ~/autoSHARP

and go there:

    cd ~/autoSHARP

It is easiest to move the just downloaded files over here using

    mv ~/Downloads/3isy* .

Now we can decide what kind of phasing computation we want to do: the best one should be using 2-wavelength MAD. For this we are going to run

      run_autoSHARP.sh \
        -fast \
        -seq 3isy.pir -ha "Se" \
        -wvl 0.97934 infl -11 3.3  -sca 3isy_aimless_0.97934A.sca \
        -wvl 0.91162 hrem -1.8 3.3 -sca 3isy_aimless_0.91162A.sca \
        -id autoSHARP_MAD-1 | tee autoSHARP_MAD-1.lis

You could just cut-and-paste the above command into your terminal (mark with left mouse button, move into your terminal and press the middle mouse button, then hit return). Or type it at the prompt - in which case it would be all on one line without the "\" (continuation) characters.

When running, it will report something like

running autoSHARP (see /home/ccp4/autoSHARP/autoSHARP_MAD-1/LISTautoSHARP.html) ...

and you could open file:///home/ccp4/autoSHARP/autoSHARP_MAD-1/LISTautoSHARP.html) in your (firefox) browser. Remember to reload that page from time to time to refresh the content.

Once the jobs is finished, you could visualise results e.g. via

    cd autoSHARP_MAD-1/Results
    coot --script 05_coot.scm

Because we have 2 wavelength data here, we could also run SAD on one of the single wavelengths, e.g.

      run_autoSHARP.sh \
        -fast \
        -seq 3isy.pir -ha "Se" \
        -wvl 0.97934 infl -11 3.3  -sca 3isy_aimless_0.97934A.sca \
        -id autoSHARP_SAD-infl | tee autoSHARP_SAD-infl.lis

or

     run_autoSHARP.sh \
        -fast \
        -seq 3isy.pir -ha "Se" \
        -wvl 0.91162 hrem -1.8 3.3 -sca 3isy_aimless_0.91162A.sca \
        -id autoSHARP_SAD-hrem | tee autoSHARP_SAD-hrem.lis

It might actually be interesting to see how these different approaches behave in terms of phasing: which of the two wavelengths works on its own (as SAD)? Does 2-wavekength MAD give us better experimental phasing as expected?

You could now re-run these examples without the -fast flag: this would produce more - probably interesting - analysis and possibly a better (e.g. more complete) inital model. Please remember to change the output identifiers in this case (argument to the -id flag and the file name given after tee).


Examples

Have a look at the examples we provide here: just pick one of those examples - maybe one that will run quite quickly - and see what steps the program is performing. You could also pick just one of the various wavelength datasets for the MAD examples: some will work even with a single wavelength (SAD), while some might not work as well. Why is that?

If you pick one of those examples and download the relevant files (right mouse-click the file name/link and select "Save Link As..."), remember that they will be saved by default in your ~/Downloads folder. You might need them over to your current working directory with a command like

mv ~/Downloads/1o22* .

A good way of getting started is to run

run_autoSHARP.sh -h

to get a help message. At the end it will show some example commands for typical situations (SAD, MAD, SIRAS, partial model etc) that you should be able to easily adapt to your situation. Please note:

  • If you don't have a sequence file handy, just give it the number of amino-acid residues per monomer (via -nres <N> flag). Remember that you will have to specify the number of Se or S atoms (per monomer) if doing Se-MET or S-SAD in that case (since autoSHARP won't be able to determine the number of expected sites from the sequence)
  • If you didn't do a fluorescence scan and you are not collecting close to the edge, just giving the wavelength value should be ok.
  • To speed things up - or for testing different approaches - you could add the -nobuild flag to the command: autoSHARP will then stop after the initial density modification step. If everything worked well, you should have some interpretable electron density maps to see if the structure is solved. And if something (HA detection, hand determination etc) didn't work, no time will be wasted trying to build into an uninterpretable map. You can always re-run autoSHARP without this flag for successful jobs.
    • Another way for speeding it up is adding the -fast flag
    • Or prepend your command with i_can_run_arpwarp=0 (to pretend that ARP/wARP is not installed we need to set this environment variable)
  • Remember to load the LISTautoSHARP.html file (inside the directory created by the autoSHARP run - usually something like autoSHARP/LISTautoSHARP.html) into a browser (e.g. firefox) to follow the progress of the job - hitting the reload button from time to time.

When working with your own data, please remember:

  • The space group of your processed data might not be correct/complete: e.g. processing might have given you P21212 data when it actually is P212121. In this case you will need to add the -spgr argument/flag to your command to force use of a specific space group. Remember that you don't need to test enantiomorphs, ie. only one of P41 or P43 needs to be checked (but you might want to test P4 and P42)
  • I222 is a bit of an odd space group since the diffraction pattern is indistinguishable from I212121 and both need to be tested.
  • If you have already some placed (partial) model, this can be used as a starting point (for site determination) - as long as it is consistent with the phasing data sets given to autoSHARP.
  • You can provide a set of initial heavy atom sites autoSHARP should start from (instead of letting it run SHELXC/SHELXD internally) - which would be a good idea when such a substructure solution is only possible through some advanced fine-tuning of search parameters.
  • The phases from SHARP (eden.mtz in the SHARP subdirectory) can also be used as a starting point for model building with SHELXE.