Estimating isomorphism by eyeballing the difference in unit cell lengths may mask significant non-isomorphism that is present. Iodine is a large atom and it is notoriously difficult to get an iodine derivative that is isomorphous to a native protein.



Depending upon the chemistry of your mother liquor, it is very possible that you no longer have a significant amount of I3C around, so not seeing a triangle of iodines does not mean that your data is worthless.



Instead, I would try MR-SAD again with Phaser, and try on individual and merged datasets. If you are merging datasets from multiple crystals, this might be problematic if the level of heavy atom substitution is different between crystals. You can also try as input the heavy atom solution from HySS or SHELX. Try to limit the resolution of your heavy atom search to the resolution of measurability for the anomalous signal.



Iodine soaks, especially quick soaks, might not be the best HA derivative in your particular case. Try tantalum bromide, that cluster is less likely to degrade and will give a stronger signal at lower resolution.



Diana



**************************************************

Diana R. Tomchick

Professor

Departments of Biophysics and Biochemistry

University of Texas Southwestern Medical Center

5323 Harry Hines Blvd.

Rm. ND10.214A

Dallas, TX 75390-8816

[log in to unmask]

(214) 645-6383 (phone)

(214) 645-6353 (fax)



On Apr 30, 2018, at 11:28 AM, Abris Bendes <[log in to unmask]> wrote:



Dear all,



I’m trying to solve the structure of a putative homodimer, but despite having collected good quality data, it has proven to be rather difficult.



I have collected native and derivative datasets (2.2-4Å, C2 SG, 165/35/115Å//104° unit cell) with acceptable merging statistics. Xtriage, Aimless does not report any apparent twinning, tNCS or significant anisotropy. Based on Matthews coefficient, two monomers should be present in the ASU. Self-rotation function doesn’t give a clear idea on NCS, I see a few weak peaks at Chi=180°. Crystal was checked against contaminant with SDS-PAGE, MS and ContaMiner.



MR approaches so far have failed due to low sequence homology (~20%). I have tried MoRDa, MrBUMP, Balbes or ensembles of different models as homodimer, monomer or divided into smaller domains. I can place the dimer or a monomer with high TFZ score (6-30), but with smaller domains MR usually thrashes. The solution always has a rather high R factor (over 0.5), unclear electron density and does not improve on subsequent intensive rebuilding. Trials in P1 (or in other possible spacegroups) have failed too.



I have recently collected multiple good datasets from a single I3C soaked crystal with measurability extending to 4-7Å on individual datasets or 3-4Å on merged multi-dataset data. No apparent radiation damage or other pathologies present. Moderate isomorphism with native data (<0.4% unit cell difference).



Any attempt to solve the substructure or get interpretable maps has so far failed even with exhaustive SHELX, CRANK2, HySS, Autosol, Phaser or AutoRickshaw runs. MR-SAD or SIRAS approaches have so far failed too. Despite promising HySS solutions (CC>0.35) I can´t see the expected triangle arrangement of the I atoms, can’t discern the hands and it always result in uninterpretable low FOM maps.



I have a few weaker datasets (not as isomorphous though) with different HA soaks (signal up to 7Å), so I could try MIR or multi-crystal averaging, though I have no experience with these approaches.



I can provide more details if needed and I would appreciate any kind of help with my problem.



Thank you very much.



Best regards,

Abris





________________________________



UT Southwestern





Medical Center







The future of medicine, today.