Hi Daniel, Oops...way too much jargon in my reply. "Cut the resolution" means "Use a lower resolution cutoff". Of course that could mean a bigger or smaller number. It means "Use a bigger number for the value of resolution" The idea here is to select a resolution range in which most of the data are reasonably accurate, where the estimated correlation of Fourier terms with the true ones is something like 0.5 (the standard half-map correlation cutoff of 0.143), but it is not obvious exactly what quality cutoff and therefore what resolution cutoff should be used, so a few tries never hurt. You might try boxing your map yourself, using varying values for the buffer around the model (from 5 to 10 A). Then tell resolve_cryo_em not to box the incoming map. This way you can actually vary the solvent content. Using the model in density modification could help but keep in mind that you should always also do it without the model so that you have some map that you know does not have any model bias. (The model-based density modification does not seem to have model bias in my experience but it always could). All the best, Tom T On Wed, Jun 8, 2022 at 11:04 PM Daniel Larsson wrote:

Hi Tom,

Thank you so much for the feedback.

By “cutting the resolution” does that mean that I should put a smaller or higher number for the density modification resolution? I guess I should try both lower and higher...

The solvent content using default settings is very high, more than 97%. My guess is that this is because there is a lot of nucleic acid and perhaps the density is being more localized (e.g. to the phosphate groups) than what is normal for proteins. I tried to force a lower solvent content, such as 90%, but that did not change the R-value. I will try with model-based density modification and see if that helps.

Regards, Daniel

On 8 Jun 2022, at 21:39, Tom Terwilliger < [email protected]> wrote:

Hi Daniel,

We don't have a standard way to interpret the R value for density modification...hence the rough guide of "0.25 is good and 0.5 is poor" but no documentation. This applies to both X-ray and cryoEM density modification.

This R value describes the mismatch between measured and map-based structure factor amplitudes. Usually these match pretty well in cases where the map is improved a lot and not when the map does not improve, but there is no obvious and simple relationship.

My recommendation would be to use the estimate of map resolution and visual quality as your guide as to whether it has improved your map, and the R value in density modification as a possible indication that some change of parameters might help if things are not going well (not that it tells you what to change, but a good one is always resolution and another is solvent content.)

In your case I might try cutting the resolution or randomly changing solvent content and testing...but as the map looks ok I would not try too hard.

I hope that helps! All the best, Tom T

On Wed, Jun 8, 2022 at 14:25 Daniel Larsson wrote:

...

Hi all,

I tried to do density modification using the resolve_cryo_em tool. The resolution improves quite a lot from around 2.60 to around 2.35 and the density looks significantly better than the autorefined original map. However, the R-values is close to 0.5, which seems very high. How should the R-value generated by this tool be interpreted? It is very poorly documented and not mentioned in the paper or on the manual page, but Tom Terwilliger mentioned in a workshop that “low number is good” and “point two-four is a very good number, point five is not a good number”, which makes me worried.

Regards, Daniel

När du har kontakt med oss på Uppsala universitet med e-post så innebär det att vi behandlar dina personuppgifter. För att läsa mer om hur vi gör det kan du läsa här: http://www.uu.se/om-uu/dataskydd-personuppgifter/

E-mailing Uppsala University means that we will process your personal data. For more information on how this is performed, please read here: http://www.uu.se/en/about-uu/data-protection-policy

_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]

-- Thomas C Terwilliger Laboratory Fellow, Los Alamos National Laboratory Senior Scientist, New Mexico Consortium 100 Entrada Dr, Los Alamos, NM 87544 Email: [email protected] Tel: 505-431-0010

-- Thomas C Terwilliger Laboratory Fellow, Los Alamos National Laboratory Senior Scientist, New Mexico Consortium 100 Entrada Dr, Los Alamos, NM 87544 Email: [email protected] Tel: 505-431-0010

Hi Pavel, This R-value is quite different from a refinement R. Basically the statistical or maximum-likelihood density modification procedure allows calculation of each structure factor based on the values of all the other structure factors and the expectations about the map (i.e, flat solvent, distribution of density values in protein, ncs, etc.) These "map-phasing" structure factors are (at least in the first cycle) independent of the starting structure factors and can be compared with the measured data and an R-value obtained. The values of these R-values range from about 0.25 to 0.5 in most cases, and when the map is very good usually the R values are smaller. These R values do not involve any atomic model so they are quite unlike a refinement R. As noted by Daniel, there is not much (or maybe any) discussion of these density modification R-values in the literature (I just looked quickly and didn't find anything really discussing this) although they are used in Resolve and related ones are used in dm and perhaps other software (if anyone knows of a relevant discussion, please let me know!). The closest I found was using a free R related to the one calculated by resolve for optimization of parameters: https://atbweb.stanford.edu/atb_publications/roberts_brunger_1995.pdf For discussion of how map-likelihood structure factors are calculated in Resolve, see "Map-likelihood phasing": https://journals.iucr.org/d/issues/2001/12/00/gr2165/ I hope that helps! All the best, Tom T On Fri, Jun 10, 2022 at 2:06 PM Pavel Afonine wrote:

Hi Tom and Daniel,

I'm confused about what R factor we are talking about here? You say "This R value describes the mismatch between measured and map-based structure factor amplitudes. ", then questions I'd ask are:

- what program was used to calculate it?

- Was all scaling between model and data generated Fs done properly? Not treating solvent region (or whatever that is in cryo-em) alone can result in large R factors.

- Was model refined properly (coordinates and B factors). Unrefined B factors even for perfectly fitting model (coordinates wise) can result in high R factors.

- Why we calculate and try to interpret R factors at all given it is cryo-EM case (no Fobs)?

Pavel On 6/9/22 05:56, Tom Terwilliger wrote:

Hi Daniel,

Oops...way too much jargon in my reply. "Cut the resolution" means "Use a lower resolution cutoff". Of course that could mean a bigger or smaller number. It means "Use a bigger number for the value of resolution" The idea here is to select a resolution range in which most of the data are reasonably accurate, where the estimated correlation of Fourier terms with the true ones is something like 0.5 (the standard half-map correlation cutoff of 0.143), but it is not obvious exactly what quality cutoff and therefore what resolution cutoff should be used, so a few tries never hurt.

You might try boxing your map yourself, using varying values for the buffer around the model (from 5 to 10 A). Then tell resolve_cryo_em not to box the incoming map. This way you can actually vary the solvent content.

Using the model in density modification could help but keep in mind that you should always also do it without the model so that you have some map that you know does not have any model bias. (The model-based density modification does not seem to have model bias in my experience but it always could).

All the best, Tom T

On Wed, Jun 8, 2022 at 11:04 PM Daniel Larsson wrote:

...

Hi Tom,

Thank you so much for the feedback.

By “cutting the resolution” does that mean that I should put a smaller or higher number for the density modification resolution? I guess I should try both lower and higher...

The solvent content using default settings is very high, more than 97%. My guess is that this is because there is a lot of nucleic acid and perhaps the density is being more localized (e.g. to the phosphate groups) than what is normal for proteins. I tried to force a lower solvent content, such as 90%, but that did not change the R-value. I will try with model-based density modification and see if that helps.

Regards, Daniel

On 8 Jun 2022, at 21:39, Tom Terwilliger < [email protected]> wrote:

Hi Daniel,

We don't have a standard way to interpret the R value for density modification...hence the rough guide of "0.25 is good and 0.5 is poor" but no documentation. This applies to both X-ray and cryoEM density modification.

This R value describes the mismatch between measured and map-based structure factor amplitudes. Usually these match pretty well in cases where the map is improved a lot and not when the map does not improve, but there is no obvious and simple relationship.

My recommendation would be to use the estimate of map resolution and visual quality as your guide as to whether it has improved your map, and the R value in density modification as a possible indication that some change of parameters might help if things are not going well (not that it tells you what to change, but a good one is always resolution and another is solvent content.)

In your case I might try cutting the resolution or randomly changing solvent content and testing...but as the map looks ok I would not try too hard.

I hope that helps! All the best, Tom T

On Wed, Jun 8, 2022 at 14:25 Daniel Larsson wrote:

...

Hi all,

I tried to do density modification using the resolve_cryo_em tool. The resolution improves quite a lot from around 2.60 to around 2.35 and the density looks significantly better than the autorefined original map. However, the R-values is close to 0.5, which seems very high. How should the R-value generated by this tool be interpreted? It is very poorly documented and not mentioned in the paper or on the manual page, but Tom Terwilliger mentioned in a workshop that “low number is good” and “point two-four is a very good number, point five is not a good number”, which makes me worried.

Regards, Daniel

När du har kontakt med oss på Uppsala universitet med e-post så innebär det att vi behandlar dina personuppgifter. För att läsa mer om hur vi gör det kan du läsa här: http://www.uu.se/om-uu/dataskydd-personuppgifter/

E-mailing Uppsala University means that we will process your personal data. For more information on how this is performed, please read here: http://www.uu.se/en/about-uu/data-protection-policy

_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]

-- Thomas C Terwilliger Laboratory Fellow, Los Alamos National Laboratory Senior Scientist, New Mexico Consortium 100 Entrada Dr, Los Alamos, NM 87544 Email: [email protected] Tel: 505-431-0010

-- Thomas C Terwilliger Laboratory Fellow, Los Alamos National Laboratory Senior Scientist, New Mexico Consortium 100 Entrada Dr, Los Alamos, NM 87544 Email: [email protected] Tel: 505-431-0010

_______________________________________________ phenixbb mailing [email protected]http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]

-- Thomas C Terwilliger Laboratory Fellow, Los Alamos National Laboratory Senior Scientist, New Mexico Consortium 100 Entrada Dr, Los Alamos, NM 87544 Email: [email protected] Tel: 505-431-0010