[Cod-bugs] COD conversion with HighScore

[Thomas Dortmann]

Hi Saulius,

a) concerning atom-labels:

we fixed the "Wat" atom-label in our conversion and now the number of pattern containing the (wrong!) element Astatine is down from 1180 to  8!
These eight remaining patterns are all minerals, where water is coded as "OWat(n)" instead of "Wat";
these are the corresponding CIF's:
9006364 - 9006368
9014246
9014312
9016377

b) concerning large Baniso and Biso values:

Both Thomas Degen and myself, we have much more experience with inorganic crystal structures than with organic crystal structures. So it is well possible, that our warning threshold for  B > 10 is too low. On the other hand I thought many organic structures are solved from low temperature data (< 100 K), with thermal vibrations being lower than at room temperature?
Anyway, we are trying to get better statistics about this point from other databases, but this could take a while.

I will come back on this issue when we have news about it.

Best regards,
Thomas Dortmann




-----Original Message-----
From: Saulius Gražulis <grazulis at ibt.lt>
Sent: 13 January 2020 08:23
To: Thomas Degen <thomas.degen at panalytical.com>; Thomas Dortmann <thomas.dortmann at panalytical.com>; cod-bugs at ibt.lt
Cc: Thomas Dortmann <thomas at tdsonline.nl>
Subject: Re: [Cod-bugs] COD conversion with HighScore

Dear Thomas,

I have finished a short inspection of the COD Uij problems. I attach the file with possible large Uij reasons identified, and a summary of the reason frequencies (files REASONS.lst and SUMMARY.txt, respectively. The REASONS.lst file can be read as CSV with TAB (ASCII 9) characters as column separators.

I have only inspected files with some Bij > 300. There are 40 such files. Lowering threshold Bij > 200 would add extra 10. Lowering to 150 adds ~320 extra files (376 total), and at Bij > 100 we have 1106 total.
Thus, Bij < 100 are very common, and Bij > 150 are relatively rare.

On 2020-01-11 18:47, Thomas Degen wrote:
> Concerning these many pattern having so many big displacement
> parameters (which we don't see in other databases) My guess is that
> the Units got confused. So it wasn’t U but the data was given as B or
> Beta instead (and simply wrongly flagged as U).

The main reasons for large Bij values, as I see them, are these:

> saulius at koala Uiso/ $ head SUMMARY.txt
>      15 Biso instead of Uiso
>       6 Uij multiplied by 1E4?
>       5 Digits missing from some Uij values?
>       4 Bij instead of Uij
>       3 Bij instead of Uij for just one atom? (???) Or refinement problems?
>       2 Two Uij values stand out. Manual data entry error? Or refinement problems?
>       2 Bij instead of Uij?
>       1 Values for one atom ('C(1)') very large. Problems with refinement?
>       1 Two Uij values stand out. Manual data entry error? Uij multiplied by 1E4?
>       1 Two Uij values stand out. Manual data entry error? (Typed "9"
> instead of "0"?)

I think we can reasonably fix the first four lines, which gives 15 + 6 +
5 + 4 = 30 corrected COD records. That's doable and most probably will be correct, but very little. The rest, IMHO, starts getting dubious. In many cases (say for "Uij multiplied by 1E4?") we should probably contact authors to verify that my interpretation of their files is correct.

As for the bulk of the Bij>10 structures, I would say most are organics and have naturally higher Bij values than minerals. I'll discuss this on the COD AB, there the people have much more experience with small molecule crystals than me.

Taking a random structure from the COD_Conv_Warnings.csv list:

> #@ CODID AtLabelUij data name  Uij      Bij> 4078519C21_atom_site_aniso_u_11  0.17300000 13.65953249
> 4078519C21_atom_site_aniso_u_33  0.16900000 13.34370515
> 4078519C17_atom_site_aniso_u_22  0.14600000 11.52769794
> 4078519C19_atom_site_aniso_u_22  0.13000000 10.26438858

> #@ label Uiso(CIF)Ueq(comp)Beq(comp)Uiso-Ueq
> C21       0.115    0.114728  9.05854000  0.00027223
> C19       0.096   0.0964213  7.61312000 -0.00042125
> C65B       0.085   0.0852352  6.72990000 -0.00023515
> C17      0.0741   0.0744293  5.87670000 -0.00032931
> C16      0.0652   0.0651477  5.14386000  0.00005231

So, the largest Bij value (U11->B11 for C21) is 13.7, the largest Biso (again, for C21) is 9.1, and the structure, although it has some mild disorder, looks pretty normal to me. Ueqiv computed from the Uij are consistent, within error, with the values provided in Uiso. Do you think Bij > 10 indicates a problem here?

Sincerely yours,
Saulius

--
Dr. Saulius Gražulis
Vilnius University Institute of Biotechnology, Saulėtekio al. 7
LT-10257 Vilnius, Lietuva (Lithuania)
fax: (+370-5)-2234367 / phone (office): (+370-5)-2234353
mobile: (+370-684)-49802, (+370-614)-36366

--
This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean.

This email and any files transmitted with it are confidential and maybe legally privileged. Such message is intended solely for the use of the individual or entity to whom they are addressed. Please notify the originator of the message if you are not the intended recipient and destroy all copies of the message. Please note that any use, dissemination, or reproduction is strictly prohibited and may be unlawful.

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.