VMD-L Mailing List

From: Ryan McGreevy (ryanmcgreevy_at_ks.uiuc.edu)
Date: Wed Jan 17 2018 - 14:49:09 CST

Sorry, it looks like the info on the -threshold option wasn't updated when
mdff check was updated to use the new algorithm. -threshold is no longer X
sigmas, but instead a direct voxel value cutoff, i.e., any value in the
density below X will be ignored. The default is no threshold, so any voxels
within the envelope will be included. Generally, 0.1 is the recommended
threshold.

On Wed, Jan 17, 2018 at 3:36 PM Rune Thomas Kidmose <rtk_at_mbg.au.dk> wrote:

> Hi again,
>
>
> After yet again playing with MDFF check for calculating CCC values, I was
> just wondering what the default sigma threshold value is, if none is given
> in the command?
>
>
> e.g.
>
>
> mdff check -ccc -map emd_6826.mrc -res 5.8 -cccfile ccc_f001.txt
>
>
> vs.
>
>
> mdff check -ccc -map emd_6826.mrc -res 5.8 -cccfile ccc_f001.txt
> -threshold 0.5
>
> I have not been able to find the answer online, so apologies if it is
> clearly stated some where and I missed it.
>
>
> Any standards on which sigma value one should use when calculating
> the CCC via mdff check?
>
>
> /Ruki
>
>
>
> ------------------------------
> *Fra:* Ryan McGreevy <ryanmcgreevy_at_ks.uiuc.edu>
> *Sendt:* 15. september 2017 15:56
> *Til:* Rune Thomas Kidmose; vmd-l_at_ks.uiuc.edu
> *Emne:* Re: vmd-l: MDFF Check -ccc in different VMD versions
>
> mdff check -ccc in 1.9.3 uses a newer, faster algorithm for calculating
> the cross correlation. The discrepancy you see is because the new algorithm
> calculates the simulated density map differently than the old algorithm.
> The new algorithm is also more accurate than the old one, so the value you
> get in 1.9.3 is more "correct". The higher the resolution of your map, the
> greater the discrepancy will be. However, if for some reason you really
> want to use the old algorithm in 1.9.3, you can use the -deprecate flag in
> the mdff check command.
>
> On Fri, Sep 15, 2017 at 6:37 AM Rune Thomas Kidmose <rtk_at_mbg.au.dk> wrote:
>
>> Hi,
>>
>>
>> I have been playing around with MDFF in different versions of VMD (1.91
>> &1.93).
>>
>>
>> In VMD v.1.91 the MDFF plugin I use is v. 0.2
>>
>>
>> In VMD v.1.93 the MDFF plugin I use is v. 0.5
>>
>>
>> When running the following CCC calculations on the same input pdb and map
>> I get different results in one of the cases:
>>
>>
>> mdff ccc (atomselection of whole protein) -i map.mrc -res 2
>>
>>
>> VMD 1.91: = 0.649038
>>
>>
>> VMD 1.93: = 0.649038
>>
>>
>> mdff check -ccc (whole protein) -map map.mrc -res 2 -cccfile output.txt
>>
>>
>> VMD 1.91: = 0.646135
>>
>>
>> VMD 1.93: = 0.5732
>>
>>
>> I state no thresholds in either case so I believe I am calculating global
>> CCC values.
>>
>>
>> So I was wondering why I see this discrepancy in CCC values between the
>> different VMD (and MDFF) versions, when using mdff check -ccc, and which I
>> should assume to be "correct"?
>>
>> Is the easy answer to just use mdff ccc instead of mdff check -ccc?
>>
>> Cheers
>>
>> Ruki
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>