Dear Peng,
The choice of asymmetric unit is somewhat arbitrary. Please forgive me and ignore this if I am saying something obvious and you already know this does not apply in your case, but I see most of your previous posts concern EM structures, so perhaps you are relatively new to solving x-ray structures?

I assume that you are looking at your structure in coot (or similar), and you have turned on the option to show symmetry mates, and you found two places where your principal model ends but is continued by a symmetry mate.  If a portion of the molecule is visible in symmetry mates, that means it is present in your principal model but in a different place- in other words you have built parts of two separate molecules.  This is a common result. So you want to find out which residues they are, delete them from your main model, and copy then from the adjoining symmetry molecule into your main model. You may have to do this twice, if your longest stretch is 6BP and there should be 20. But bear in mind the second place you observed may be symmetry related, and disappear after you fix the first. Possibly some of the molecule is disordered.

There is probably a way to do this all in coot, using things like copy-fragment and merge-molecules, but I am a beginner in coot. What you can do is:

Find out which residues you want to transfer from the symmetry mate to main model

file:save symmetry coordinates, click on an atom in the symmetry mate where it abuts, and give a filename to save it as.

Use a text editor to replace those atoms in your main model with the atoms from the symmetry mate (save with a new name just in case)

Load into coot and see if it looks OK

Refine the new model and see if you get essentially statistics equal to before and no clashes

The achesymm web server (http://achesym.ibch.poznan.pl/) might be able to do this automatically.

(the blind leading the blind - always glad to hear better solutions)

On 01/30/2018 05:41 PM, Peng wrote:
>
> Dear Nicolas,
> Thank you for your help.
> Our DNA is not a palindromic sequence, and I think it should not be degraded easily.
> The data can be processed in the space groups C2221, P41212 or lower, with 12bp, 6bp ,24bp or longer in one NCS. I can see the connection between two symmetric DNA molecules in all these space groups.
> Actually, I am really confused about that because 20pb-DNA was used in crystallization.
> Peng
>
>
>
>
> 在2018年01月30 16时58分, "Nicolas FOOS"<[log in to unmask]>写道:
>
>
>     Dear Peng,
>
>     to me your problem sound a bit strange, except if it's a      palindromic sequence. I don't understand how you can have one part      of the DNA in one asymmetric unit and one in another one. My      question are : maybe you considering a NCS as a true symmetry and      underestimate the unit-cell dimensions? Or you DNA has been      degraded by DNase and the current size is not 20pb, in this case      you are not supposed to create an artificial  connection. Is the      resolution good enough to be certain of the sequence ?
>
>     Sorry, I don't provide any answer, but I am curious and try to      understand what is going on.
>
>     Hope this finally help.
>
>     Nicolas
>
>     Nicolas Foos
>     PhD
>     Structural Biology Group
>     European Synchrotron Radiation Facility (E.S.R.F)
>     71, avenue des Martyrs
>     CS 40220
>     38043 GRENOBLE Cedex 9
>     +33 (0)6 76 88 14 87
>     +33 (0)4 76 88 45 19
>
>     On 29/01/2018 20:28, Peng wrote:
>>
>>     Hello, everyone，
>>
>>      Recently, we solve a protein-DNA complex.            20bp-DNA was used for crystallization, but only 6bp was            found in one symmetric unit.
>>
>>     My question is：
>>
>>     How to define the DNA bond between P and O3’            from two different symmetric units during my refinement?
>>
>>     Peng
>>
>>
>>
>>
>>
>>
>
>
>