[phenixbb] Tantalum anomalous signal of a low-resolution SAD dataset with lattice-translocation disorder

[Francis Reyes]

Ta6Br14 is a great compound to use for low resolution phasing problems, but not for its anomalous signal!

For example, a single cluster in the background of a 300kDa protein gives an a SAD signal of roughly 5% at 5A[1].  These clusters are almost always rotationally disordered so I used the pessimistic case. This is an extremely weak signal compared to Rmerge at this resolution, which assuming that this is the highest resolution bin in your case, I imagine your Rmerge to be somewhere in the 20-30s. You can't estimate a 5% signal when your reflections are as much as 30% inconsistent. 

The largest phasing power for these clusters comes from isomorphous differences at low resolution[2].  At 5A resolution, the contribution of a *single* TaBr cluster to scattering is less than 1% (not helpful to you), but at 25A, its as much as 10% [3]!  Furthermore, your Rmerge at this resolution are probably in the low 3-5%'s and you've likely counted thousands of photons so the counting error is low. 

I've solved many large structures with tantalum clusters, the phases at low resolution are *excellent*. 

Cheers,

F

[1]  Ethan Merrit's MAD power analysis page (http://skuld.bmsc.washington.edu/scatter/AS_signal.html). 
[2] http://www.nobelprize.org/nobel_prizes/chemistry/laureates/2009/steitz_lecture.pdf
[3] Using Holton assumptions to Magdoff Crick, http://proteincrystallography.org/ccp4bb/message9404.html. 

On Oct 21, 2015, at 12:35 AM, Shun Liu <sliu.xtal at gmail.com> wrote:

> Dear Phenix colleagues,
> 
> We are working on a Ta6Br14 cluster-SAD dataset (4.3 angstroms) with lattice-translocation disorder (with a total Rmerge of 0.16). Both SHELXC and Xtriage gave the similar positive result about the anomalous signal (See below). 
> 
> -------------------------------------------------------------------------------------------------
> Resl.   Inf.  9.58  7.61  6.65  6.04  5.61  5.27  5.01  4.79  4.61  4.45  4.31
> N(data)    1427  1410  1410  1409  1385  1452  1367  1399  1367  1395  1385
> <I/sig>    76.3  52.1  26.7  13.7   8.7   6.1   5.1   5.0   3.9   2.7   1.8
> %Complete  98.8  99.9 100.0  99.9 100.0 100.0 100.0  99.9  99.7  99.6  95.8
> <d"/sig>  11.67  9.67  6.12  3.73  2.50  1.59  1.21  0.99  0.84  0.79  0.68
> ------------------------------------------------------------------------------------------------ 
> unused:         - 49.7669 [   0/29  ]
> bin  1: 49.7669 -  9.2629 [3086/3100]  0.8212
> bin  2:  9.2629 -  7.3599 [3031/3031]  0.7689
> bin  3:  7.3599 -  6.4318 [3073/3073]  0.5380
> bin  4:  6.4318 -  5.8447 [3071/3075]  0.2933
> bin  5:  5.8447 -  5.4263 [3054/3054]  0.1230
> bin  6:  5.4263 -  5.1067 [3070/3071]  0.0363
> bin  7:  5.1067 -  4.8512 [3038/3039]  0.0126
> bin  8:  4.8512 -  4.6402 [3028/3048]  0.0074
> bin  9:  4.6402 -  4.4617 [2955/3037]  0.0021
> bin 10:  4.4617 -  4.3078 [2730/3059]  0.0000
> unused:  4.3078 -         [   0/0   ]
> --------------------------------------------------------------------------------------------------
> However, the tantalum sites we found and the initial map seem ambiguous, with which automatic model-building failed. 
> I am wondering whether the lattice-translocation disorder of the dataset impacts the reliability of the anomalous signal, tantalum sites and the initial map. If it does, how can we decrease its impact? If it doesn't, is it possible to find the accurate Ta sites and generate an interpretable map suitable for model-building with this dataset? (After all, it has been reported that Ta sites can be found at 6A resolution.)
> 
> Any suggestion and comment will be highly appreciated! 
> 
> Best,
> Shun
> 
> 
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
> Unsubscribe: phenixbb-leave at phenix-online.org