[buster-discuss] test flag number in BUSTER

[Keitaro Yamashita]

Dear David (and BUSTER developers),

Thank you for your reply.

Do you mean BUSTER always use flag number 0 if FreeR_flag column
present in the mtz file, right?

If so, in my case the flag was correctly chosen.


I have still some questions:

- If superior bulk solvent model helped, I think it should improve
R-factors in low resolution shell.
  But actually, rather high resolution shell got improved.

- Sorry I correct these values 42% and 35%, respectively:
> Looking at free-R value given by autoBUSTER, I noticed it is very
> smaller than by other softwares (phenix.refine and refmac5 say 43%,
> but BUSTER says 36%).

And here's summary:

              Rwork/Rfree    after refinement (5 macro cycles)
BUSTER        30/35       -> 34/35
phenix.refine 35/42

- It means R-free by BUSTER is approximately equal to R-work by phenix.refine.
This is why I doubted test flag was wrongly chosen.

- And after BUSTER refinement cycles, R-work is getting higher,
finally, R-work and R-free reached similar values.
Since difference between R-work and R-free is very small -- I think
something is wrong with my input.


I will appreciate any comment,

Keitaro



2011/3/27 David Veesler <david.veesler at afmb.univ-mrs.fr>:
> Hi Keitaro,
> I am not a Buster developer but a thorough user and can provide you with some answers.
> When you start a Buster job, you can see on the shell that Buster carries on a basic check of your pdb and mtz and directly after you can see something like that
>
> ##############################################################################
> # refining against FP,SIGFP with: FreeR_flag
> ##############################################################################
>
> In that case it means that Buster runs with a FreeRflag:
> -the one present in your mtz if you have one with the correct name and 0 label
> -or with an automatically generated one with 5% of your reflections if you don't have one matching these criteria
>
> Regarding your Rfree values, there is nothing surprising at all, Buster always provide better Rfree values than other refinement software and this is due in a large extent to the superior solvent model. Accordingly, if you compare the electron-density maps you will see that they are of superior quality with Buster.
>
> HTH
> Cheers
> David
>
> Le 26 mars 2011 à 08:02, Keitaro Yamashita a écrit :
>
>> Dear BUSTER developers,
>>
>> I have questions about running autoBUSTER 2.8.
>> I'm very new to this software.
>>
>> How can I know which FreeR_flag number is used as test set?
>> And how can I tell BUSTER the correct flag number?
>>
>> Looking at free-R value given by autoBUSTER, I noticed it is very
>> smaller than by other softwares (phenix.refine and refmac5 say 43%,
>> but BUSTER says 36%).
>> So I'm afraid that free-R flag number was wrongly chosen.
>>
>>
>> Thanks in advance,
>>
>> Keitaro
>> _______________________________________________
>> buster-discuss mailing list
>> buster-discuss at globalphasing.com
>> https://www.globalphasing.com/mailman/listinfo/buster-discuss
>>
>> --
>> This message has been scanned for viruses and
>> dangerous content by MailScanner, and is
>> believed to be clean.
>>
>
>
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
>
> _______________________________________________
> buster-discuss mailing list
> buster-discuss at globalphasing.com
> https://www.globalphasing.com/mailman/listinfo/buster-discuss
>