Re: [ccp4bb] Careless and negative Wilson B-factor

Sun, 08 Jan 2023 21:05:27 -0800

Dear Doeke,


Thank you for developing such excellent software! I will try the options you suggested and post any additional issues to GitHub.


Regarding the R-merge, R-meas, and R-pim values, I used the XDS CORRECT output (XDS_ASCII.HKL) as the "Unmerged data" and an mtz file obtained from each of the merging software as the "MTZ file" input in Phenix.table_one to calculate these values.

For the Aimless-refined model, I ran phenix.refine with the "Update waters" option, using the structure factor obtained from Aimless and a model previously refined with the Careless output structure factor. However, the refined electron density with the Aimless output was weaker than that with the Careless output, which resulted in the removal of most of the water during the phenix.refine cycles.


Thank you,


Gyuhyeok

--------- Original Mail ---------
Sender : "Hekstra, Doeke Romke" <[email protected]>
Recipient : <[email protected]>
Received Date : 2023/Jan/9(Mon) 12:21:46
Subject : Re: [ccp4bb] Careless and negative Wilson B-factor

Dear Gyuhyeok,

 

Please feel free to describe any challenges with settings for Careless at https://github.com/rs-station/careless/issues

We will be happy to help and improve our online documentation accordingly.

I assume some of the statistics you report below are simply duplicated from Aimless? We did not implement calculations for R-merge, R-meas, and R-pim. Careless can currently report CC1/2, CCanom (if Friedel mates were kept apart), Rsplit, and a new quantity, CCpred, based on prediction of expected intensities for a heldout test set of integrated intensities. To do so, make sure you use the “--merge-half-datasets” and “--half-dataset-repeats N” (with N some number of repeats) when running Careless. You can then use “careless.cchalf your_output_xval.mtz” to obtain CC1/2.

There are some curious differences between your refinements (for example, 203 versus 13 solvent molecules) that may or may not be directly related to Careless. If you are willing to share your PDB and MTZ files, we’d be very interested.

 

Thanks,

Doeke

 

 

From: CCP4 bulletin board <[email protected]> On Behalf Of "???"
Sent: Sunday, January 8, 2023 9:45 PM
To: [email protected]
Subject: Re: [ccp4bb] Careless and negative Wilson B-factor

 

Dear Eleanor,

 

I agree that the decrease in the B-factor is likely due to "Careless."

The author of this article seems to recognize Wilson B-factor issues during the peer review process (https://static-content.springer.com/esm/art%3A10.1038%2Fs41467-022-35280-8/MediaObjects/41467_2022_35280_MOESM2_ESM.pdf) and offers some options for addressing them (such as the --wilson-prior-b WILSON_PRIOR_B option).

It may take some time for me to get the software running correctly.

 

Gyuhyeok

 

P.S. I have attached some preliminary statistics using datasets from Careless or Aimless. Please note that both the Atomic B-factor and the Wilson B-factor are significantly reduced in Careless data.

 

Merging tool

Carelss

Aimless

Wavelength

1.0000

1.0000

Resolution range

37.98 - 2.60
(2.69 - 2.60)

43.06 - 2.60
(2.69 - 2.60)

Space group

I 2 3

I 2 3

Unit cell

161.128 161.128 161.128
90 90 90

161.128 161.128 161.128
90 90 90

Total reflections

886846 (79364)

886852 (79364)

Unique reflections

21520 (2135)

21522 (2135)

Multiplicity

41.2 (37.2)

41.2 (37.2)

Completeness (%)

99.95 (100.00)

99.94 (99.95)

Mean I/sigma(I)

14.2 (0.8)

14.2 (0.8)

Wilson B-factor

-0.91

74.5

R-merge

0.240 (4.566)

0.240 (4.566)

R-meas

0.243 (4.629)

0.243 (4.629)

R-pim

0.038 (0.757)

0.038 (0.757)

CC1/2

0.998 (0.362)

0.997 (0.362)

CC*

0.999 (0.729)

0.999 (0.729)

Reflections used in refinement

21520 (2135)

21519 (2134)

Reflections used for R-free

1118 (112)

1118 (112)

R-work

0.2027 (0.2547)

0.2093 (0.3422)

R-free

0.2601 (0.3282)

0.2564 (0.3634)

CC(work)

0.664 (0.708)

0.919 (0.622)

CC(free)

0.622 (0.569)

0.766 (0.478)

Number of non-hydrogen atoms

3930

3740

macromolecules

3655

3655

ligands

72

72

solvent

203

13

Protein residues

443

443

RMS(bonds)

0.008

0.009

RMS(angles)

1.030

1.110

Ramachandran favored (%)

95.4

93.2

Ramachandran allowed (%)

4.3

6.4

Ramachandran outliers (%)

0.2

0.5

Rotamer outliers (%)

2.0

3.3

Clashscore

8.95

8.41

Average B-factor

19.9

70.2

macromolecules

19.7

70.0

ligands

31.1

81.5

solvent

20.1

61.5

body,div,table,thead,tbody,tfoot,tr,th,td,p { font-family:"Liberation Sans"; font-size:x-small }a.comment-indicator:hover + comment { background:#ffd; position:absolute; display:block; border:1px solid black; padding:0.5em; }a.comment-indicator { background:red; display:inline-block; border:1px solid black; width:0.5em; height:0.5em; }comment { display:none; }

--------- 원본 메일 ---------

보낸사람 : Eleanor Dodson <[email protected]>
받는사람 : <[email protected]>
받은날짜 : 2023/Jan/8(Sun) 10:39:39
제목 : Re: [ccp4bb] Careless and negative Wilson B-factor

I suppose so, but I would refine your final model against the initial data too, and see what R factors you get..

Eleanor

 

On Fri, 6 Jan 2023 at 14:46, "조규혁" <[email protected]> wrote:

Dear Eleanor,

 

I was wondering if the model and structure factors with the negative Wilson B-factor can be deposited in the PDB or made interpretable for biochemical experiments.

The data were initially processed in XDS and the CORRECT output (XDS_ASCII.HKL) was provided to Careless.

 

Sincerely,

 

Gyuhyeok

--------- 원본 메일 ---------

보낸사람 : Eleanor Dodson <[email protected]>
받는사람 : <[email protected]>
참조 : <[email protected]>
받은날짜 : 2023/Jan/6(Fri) 20:09:57
제목 : Re: [ccp4bb] Careless and negative Wilson B-factor

The tool must have sharpened your data - sometimes enhancing the outer data does improve the electron density..

So I wouldnt worry about that "negative" B factor - it is doubtless a result of the "careless" run, and not presumably the one you got at the data processing step..

Eleanor

 

 

On Fri, 6 Jan 2023 at 08:27, "조규혁" <[email protected]> wrote:

Dear all,

I recently tested a novel merging tool for crystallography data called Careless (https://github.com/rs-station/careless).

When I used it on our 2.6 A-resolution structure in the I23 space group with a lipid bound structure, I noticed that the electron density surrounding the acyl chain was significantly improved.

However, I am concerned about the Wilson B-factor. The Phenix.table_one tool reports that the Wilson B-factor of the merged mtz file is -0.91. Phenix.Xtriage warns that a negative Wilson B-factor can be an indication of unusual pathology or artificial manipulation.

Has anyone else encountered similar conditions? Are there any circumstances in which a negative Wilson B-factor can be rationalized?

Thank you,
Gyuhyeok

 

P.S. Here is the part of the output of Phenix.table_one.

 

Wavelength

1.0000

Resolution range

37.98 - 2.60
(2.69 - 2.60)

Space group

I 2 3

Unit cell

161.128 161.128 161.128
90 90 90

Total reflections

886846 (79364)

Unique reflections

21520 (2135)

Multiplicity

41.2 (37.2)

Completeness (%)

99.95 (100.00)

Mean I/sigma(I)

14.2 (0.8)

Wilson B-factor

-0.91

R-merge

0.240 (4.566)

R-meas

0.243 (4.629)

R-pim

0.038 (0.757)

CC1/2

0.998 (0.362)

CC*

0.999 (0.729)

 

 

 

 

PhD student

Department of Chemistry

Gwangju Institute of Science and Technology (GIST)

123 Cheomdangwagi-ro, Buk-gu, Gwangju, 61005

Republic of Korea

Tel. +82 62-715-4633

e-mail: [email protected]

 


To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1


 

 

 

 

PhD student

Department of Chemistry

Gwangju Institute of Science and Technology (GIST)

123 Cheomdangwagi-ro, Buk-gu, Gwangju, 61005

Republic of Korea

Tel. +82 62-715-4633

e-mail: [email protected]

 


To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1


 

 

 

 

PhD student

Department of Chemistry

Gwangju Institute of Science and Technology (GIST)

123 Cheomdangwagi-ro, Buk-gu, Gwangju, 61005

Republic of Korea

Tel. +82 62-715-4633

e-mail: [email protected]

 


To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1


To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1






PhD student

Department of Chemistry

Gwangju Institute of Science and Technology (GIST)

123 Cheomdangwagi-ro, Buk-gu, Gwangju, 61005

Republic of Korea

Tel. +82 62-715-4633

e-mail: [email protected]




To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1

Reply via email to