If you're using an older version of Phenix with TLS refinement, this may be simply due to the issue of the radius for B-factor restraints being less than the length of the disulfide bond (so the two sulfurs aren't restrained to similar B-factors). Looking back at some of my older refinements, I see many disulfides with negative density blobs that are probably attributable to this.

From: [email protected] <[email protected]> on behalf of Pavel Afonine <[email protected]>
Sent: Monday, 15 February 2016 6:46 PM
To: Ezequiel Noguera; [email protected]
Subject: Re: [phenixbb] Update: Check for radiation-induced reduction of a disulfide bond - PS

P.S.:

If atom is misplaced normally you would see both: positive and negative residual density in close proximity to each other.

If you only see negative density at SG centers this may mean the site is partially occupied.

Pavel

Sorry, my previous email was incomplete. My structure has negative density in the center of some disulfides (the magnitude of this effect vary between different monomers in the ASU). I'd like to refine the structure in an "artificial reduced form", where cysteines involved in disulfides are treated in reduced form, but repulsions originating from neighboring sulfur atoms are turned-off, to monitor distances (I expect that SG atoms remain close to 2 A), guided only by data. Thank you,