Re: [ccp4bb] A new capability on the STARANISO server: "PDBpeep"
Dear Jacob and Gerard, we performed a statistical analysis of anisotropy of the entire PDB, with a special focus on membrane proteins (Jacb, your other post). You can find more details on this article: https://www.nature.com/articles/s41598-017-17216-1 We also performed an analysis by space group, not reported in the article, but not seeing anything coming out. The data used in the article is being reviewed at the moment to be available for you to play with. Hopefully, the reviewer is going to read your posts and is going to agree on the importance of analyzing this phenomenon. all the best Vincent On 28/03/2018 20:35, Gerard Bricogne wrote: Dear Jacob, Thank you for the appreciative comment. I am not sure that there is any such thing as an up-to-date estimate of the prevalence of anisotropy in the PDB - but now you can get a feel for it yourself by looking at any entries you want. However please do not submit the whole PDB to the server - yet ;-) . From looking at anisotropy as visible through the overall scaling Debye-Waller factor, I would say that whenever anisotropy is allowed by the Laue group, it will be present, even if mild. Symmetry lower than cubic means that intermolecular contacts along directions that are not symmetry-equivalent will be different, and there is no reason why different contacts should create identical degrees of long-range order. Have fun! Gerard. -- On Wed, Mar 28, 2018 at 06:13:29PM +, Keller, Jacob wrote: Wow, this is really cool--just tried a quick look at a recent membrane protein (5eqi) and you can see right away that there is anisotropy. I would guess this can be found in the literature, but how prevalent is anisotropy in the PDB? Jacob Keller + Jacob Pearson Keller Research Scientist / Looger Lab HHMI Janelia Research Campus 19700 Helix Dr, Ashburn, VA 20147 Desk: (571)209-4000 x3159 Cell: (301)592-7004 + The content of this email is confidential and intended for the recipient specified in message only. It is strictly forbidden to share any part of this message with any third party, without a written consent of the sender. If you received this message by mistake, please reply to this message and follow with its deletion, so that we can ensure such a mistake does not occur in the future. -Original Message- From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Gerard Bricogne Sent: Wednesday, March 28, 2018 11:56 AM To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] A new capability on the STARANISO server: "PDBpeep" Dear all, Ever since the WebGL viewer became available on the STARANISO server, we have found ourselves using it with increasing frequency to take a quick look (a "peep") at the diffraction data deposited with various PDB entries - for example, to try and identify a root cause for some sub-optimal refinement results, or, quite often, just out of sheer curiosity! This involved a totally straightforward procedure whereby the diffraction data file associated with a given PDB entry was downloaded from the PDB and subsequently uploaded to the STARANISO server. Gradually, however, this operation became so popular among some of us that we thought it would be useful to implement this simple procedure as an autonomous capability - and thus was born "PDBpeep" ! You can access this new feature by connecting to http://staraniso.globalphasing.org/cgi-bin/PDBpeep.cgi and enter a PDB code into the box. As indicated on that page, this provides only a cursory look at the overall quality of each dataset, and any further analysis or output can only be obtained by submitting the datafile to the STARANISO server. Better results would clearly be obtainable if the raw images for these datasets had been deposited and could be reprocessed, with the untruncated output of that processing then being submitted to the STARANISO server (reprocessing the images with autoPROC would combine those two steps into a single one). Most of the deposited datasets have been isotropically truncated, and their 3D view in WebGL often suggests that this truncation was too drastic. A number of entries will show infelicities - such as cusps and/or missing angular ranges, or even stripes caused by gaps between the modules of pixel detectors if the beam centre is at a position symmetric relative to those gaps - all marked up in blue. Our purpose in sharing this capability with the community is to bring a further contribution to the process of making everyone more "data quality aware" and keen to scrutinise more closely the protocols by which they collect diffraction data, or have such data collected on their behalf. We will be grateful to receive feedback about PDBpeep, just as we have
Re: [ccp4bb] A new capability on the STARANISO server: "PDBpeep"
>>>These simple things are easily forgotten if there isn't direct visual >>>evidence for their existence. Maybe someday we scientists will collectively realize the huge importance of intuitive data presentation and focus more on it, scorn it less as "mere presentation"? When one thinks of the most powerful breakthroughs in the history of science, one is struck by the advances of simply presenting data (or collecting data) in a way that speaks directly to the intuition, e.g., heliocentrism, the DNA double helix, structural bio in general, cartesian graphing. What about things like molecular graphics, or the ribbon diagrams and other representations which underlie them? All of these are "mere presentations" but have driven science forward incredibly. Shouldn't we focus on making more data-to-intuition translators? For example, what there was a VR interface in which the user could actually feel physical forces in protein molecules, get a touch-intuitive sense of what proteins are like? We would already "know" so much more about proteins! Jacob + Jacob Pearson Keller Research Scientist / Looger Lab HHMI Janelia Research Campus 19700 Helix Dr, Ashburn, VA 20147 Desk: (571)209-4000 x3159 Cell: (301)592-7004 + The content of this email is confidential and intended for the recipient specified in message only. It is strictly forbidden to share any part of this message with any third party, without a written consent of the sender. If you received this message by mistake, please reply to this message and follow with its deletion, so that we can ensure such a mistake does not occur in the future.
Re: [ccp4bb] A new capability on the STARANISO server: "PDBpeep"
Dear Jeffrey, Thank you for sharing your observations and the questions they bring up. I can see two things in the 3D displays for your two PDB entries, none of which seems new or surprising. The first is that there is a peppering of blue dots within the volume that otherwise seems occupied with colours indicating the presence of significant data. This can be created by the rejection of individual misfits, or by the folding over by symmetry of detector gaps (see 5lgk for a spectacular example of the latter!). The second is indeed a waviness of the boundary surfaces between different colours. This is because the local average of I/sig(I) is modulated by the distribution of I itself (if there were only Poisson statistics errors entering sig(I), then each I/sig(I) would just be sqrt(I)), and also by the multiplicity of measurement through the 1/sqrt(multiplicity) factor. When you combine those two distinct and independent influences, you can get quite a lot of directional modulation in the local average of of I/sig(I), the first kind coming largely from the tertiary structure of the molecule and the second from the way the data were collected. These simple things are easily forgotten if there isn't direct visual evidence for their existence. Thank you again: this feedback is very useful in pointing out what needs commenting upon to makes these plots understandable. With best wishes, Gerard. -- On Wed, Mar 28, 2018 at 08:22:23PM +, Jeffrey B Bonanno wrote: > Hi Gerard, > > Very nice server indeed. I reviewed the 3D plots of some recent high res > structures and am struck by the granularity which is not necessarily globally > directional. For instance, I ran 4QRN and 5EHI and see "ripples" in the > I/sigI plot between the various "layers" (higher res yellow to orange > "transition" features are most easily visible but the purple to blue low res > are very interesting as well). Can you comment on the significance of these > intensity-weighted reciprocal lattice features as pertains to real space? It > is tempting to see a manifestation of the solvent mask in the color coding :-) > > best, > Jeff > > Jeffrey B. Bonanno, Ph.D. > Department of Biochemistry > Albert Einstein College of Medicine > 1300 Morris Park Avenue > Bronx, NY 10461 > off. 718-430-2452 fax. 718-430-8565 > email jeffrey.bona...@einstein.yu.edu > > -Original Message- > From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Gerard > Bricogne > Sent: Wednesday, March 28, 2018 2:35 PM > To: CCP4BB@JISCMAIL.AC.UK > Subject: Re: [ccp4bb] A new capability on the STARANISO server: "PDBpeep" > > Dear Jacob, > > Thank you for the appreciative comment. > > I am not sure that there is any such thing as an up-to-date > estimate of the prevalence of anisotropy in the PDB - but now you can > get a feel for it yourself by looking at any entries you want. However > please do not submit the whole PDB to the server - yet ;-) . > > From looking at anisotropy as visible through the overall scaling > Debye-Waller factor, I would say that whenever anisotropy is allowed > by the Laue group, it will be present, even if mild. Symmetry lower > than cubic means that intermolecular contacts along directions that > are not symmetry-equivalent will be different, and there is no reason > why different contacts should create identical degrees of long-range > order. > > > Have fun! > > Gerard. > > -- > On Wed, Mar 28, 2018 at 06:13:29PM +, Keller, Jacob wrote: > > Wow, this is really cool--just tried a quick look at a recent membrane > > protein (5eqi) and you can see right away that there is anisotropy. > > > > I would guess this can be found in the literature, but how prevalent is > > anisotropy in the PDB? > > > > Jacob Keller > > > > + > > Jacob Pearson Keller > > Research Scientist / Looger Lab > > HHMI Janelia Research Campus > > 19700 Helix Dr, Ashburn, VA 20147 > > Desk: (571)209-4000 x3159 > > Cell: (301)592-7004 > > + > > > > The content of this email is confidential and intended for the recipient > > specified in message only. It is strictly forbidden to share any part of > > this message with any third party, without a written consent of the sender. > > If you received this message by mistake, please reply to this message and > > follow with its deletion, so that we can ensure such a mistake does not > > occur in the future. > > > > -Original Message- > >
Re: [ccp4bb] A new capability on the STARANISO server: "PDBpeep"
Hi Gerard, Very nice server indeed. I reviewed the 3D plots of some recent high res structures and am struck by the granularity which is not necessarily globally directional. For instance, I ran 4QRN and 5EHI and see "ripples" in the I/sigI plot between the various "layers" (higher res yellow to orange "transition" features are most easily visible but the purple to blue low res are very interesting as well). Can you comment on the significance of these intensity-weighted reciprocal lattice features as pertains to real space? It is tempting to see a manifestation of the solvent mask in the color coding :-) best, Jeff Jeffrey B. Bonanno, Ph.D. Department of Biochemistry Albert Einstein College of Medicine 1300 Morris Park Avenue Bronx, NY 10461 off. 718-430-2452 fax. 718-430-8565 email jeffrey.bona...@einstein.yu.edu -Original Message- From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Gerard Bricogne Sent: Wednesday, March 28, 2018 2:35 PM To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] A new capability on the STARANISO server: "PDBpeep" Dear Jacob, Thank you for the appreciative comment. I am not sure that there is any such thing as an up-to-date estimate of the prevalence of anisotropy in the PDB - but now you can get a feel for it yourself by looking at any entries you want. However please do not submit the whole PDB to the server - yet ;-) . From looking at anisotropy as visible through the overall scaling Debye-Waller factor, I would say that whenever anisotropy is allowed by the Laue group, it will be present, even if mild. Symmetry lower than cubic means that intermolecular contacts along directions that are not symmetry-equivalent will be different, and there is no reason why different contacts should create identical degrees of long-range order. Have fun! Gerard. -- On Wed, Mar 28, 2018 at 06:13:29PM +, Keller, Jacob wrote: > Wow, this is really cool--just tried a quick look at a recent membrane > protein (5eqi) and you can see right away that there is anisotropy. > > I would guess this can be found in the literature, but how prevalent is > anisotropy in the PDB? > > Jacob Keller > > + > Jacob Pearson Keller > Research Scientist / Looger Lab > HHMI Janelia Research Campus > 19700 Helix Dr, Ashburn, VA 20147 > Desk: (571)209-4000 x3159 > Cell: (301)592-7004 > + > > The content of this email is confidential and intended for the recipient > specified in message only. It is strictly forbidden to share any part of this > message with any third party, without a written consent of the sender. If you > received this message by mistake, please reply to this message and follow > with its deletion, so that we can ensure such a mistake does not occur in the > future. > > -Original Message- > From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Gerard > Bricogne > Sent: Wednesday, March 28, 2018 11:56 AM > To: CCP4BB@JISCMAIL.AC.UK > Subject: [ccp4bb] A new capability on the STARANISO server: "PDBpeep" > > Dear all, > > Ever since the WebGL viewer became available on the STARANISO server, we > have found ourselves using it with increasing frequency to take a quick look > (a "peep") at the diffraction data deposited with various PDB entries - for > example, to try and identify a root cause for some sub-optimal refinement > results, or, quite often, just out of sheer curiosity! > > This involved a totally straightforward procedure whereby the > diffraction data file associated with a given PDB entry was downloaded from > the PDB and subsequently uploaded to the STARANISO server. > > Gradually, however, this operation became so popular among some of us > that we thought it would be useful to implement this simple procedure as an > autonomous capability - and thus was born "PDBpeep" ! > > You can access this new feature by connecting to > >http://staraniso.globalphasing.org/cgi-bin/PDBpeep.cgi > > and enter a PDB code into the box. As indicated on that page, this provides > only a cursory look at the overall quality of each dataset, and any further > analysis or output can only be obtained by submitting the datafile to the > STARANISO server. Better results would clearly be obtainable if the raw > images for these datasets had been deposited and could be reprocessed, with > the untruncated output of that processing then being submitted to the > STARANISO server (reprocessing the images with autoPROC would combine those > two steps into a single one). > > Most of the deposited datasets
Re: [ccp4bb] A new capability on the STARANISO server: "PDBpeep"
>> I am not sure that there is any such thing as an up-to-date estimate of >> the prevalence of anisotropy in the PDB - but now you can get a feel for it >> yourself by looking at any entries you want. However please do not submit >> the whole PDB to the server - yet ;-) . It would be great to have a figure showing distributions of anisotropy across the pdb, perhaps filtering for various parameters like resolution, lattice size, membrane proteins, etc. And this should be helpful for assaying the importance of dealing with anisotropy using STARANISO and others. >> From looking at anisotropy as visible through the overall scaling >> Debye-Waller factor, I would say that whenever anisotropy is allowed by the >> Laue group, it will be present, even if mild. Symmetry lower than cubic >> means that intermolecular contacts along directions that are not >> symmetry-equivalent will be different, and there is no reason why different >> contacts should create identical degrees of long-range order. Yes, that makes all the sense in the world. I wonder how one would evaluate in a given case whether the usual isotropic approximation is good enough? Or perhaps there's no reason to use the isotropic assumption at all, now that there are ways of dealing with anisotropy? JPK
Re: [ccp4bb] A new capability on the STARANISO server: "PDBpeep"
Dear Jacob, Thank you for the appreciative comment. I am not sure that there is any such thing as an up-to-date estimate of the prevalence of anisotropy in the PDB - but now you can get a feel for it yourself by looking at any entries you want. However please do not submit the whole PDB to the server - yet ;-) . From looking at anisotropy as visible through the overall scaling Debye-Waller factor, I would say that whenever anisotropy is allowed by the Laue group, it will be present, even if mild. Symmetry lower than cubic means that intermolecular contacts along directions that are not symmetry-equivalent will be different, and there is no reason why different contacts should create identical degrees of long-range order. Have fun! Gerard. -- On Wed, Mar 28, 2018 at 06:13:29PM +, Keller, Jacob wrote: > Wow, this is really cool--just tried a quick look at a recent membrane > protein (5eqi) and you can see right away that there is anisotropy. > > I would guess this can be found in the literature, but how prevalent is > anisotropy in the PDB? > > Jacob Keller > > + > Jacob Pearson Keller > Research Scientist / Looger Lab > HHMI Janelia Research Campus > 19700 Helix Dr, Ashburn, VA 20147 > Desk: (571)209-4000 x3159 > Cell: (301)592-7004 > + > > The content of this email is confidential and intended for the recipient > specified in message only. It is strictly forbidden to share any part of this > message with any third party, without a written consent of the sender. If you > received this message by mistake, please reply to this message and follow > with its deletion, so that we can ensure such a mistake does not occur in the > future. > > -Original Message- > From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Gerard > Bricogne > Sent: Wednesday, March 28, 2018 11:56 AM > To: CCP4BB@JISCMAIL.AC.UK > Subject: [ccp4bb] A new capability on the STARANISO server: "PDBpeep" > > Dear all, > > Ever since the WebGL viewer became available on the STARANISO server, we > have found ourselves using it with increasing frequency to take a quick look > (a "peep") at the diffraction data deposited with various PDB entries - for > example, to try and identify a root cause for some sub-optimal refinement > results, or, quite often, just out of sheer curiosity! > > This involved a totally straightforward procedure whereby the > diffraction data file associated with a given PDB entry was downloaded from > the PDB and subsequently uploaded to the STARANISO server. > > Gradually, however, this operation became so popular among some of us > that we thought it would be useful to implement this simple procedure as an > autonomous capability - and thus was born "PDBpeep" ! > > You can access this new feature by connecting to > >http://staraniso.globalphasing.org/cgi-bin/PDBpeep.cgi > > and enter a PDB code into the box. As indicated on that page, this provides > only a cursory look at the overall quality of each dataset, and any further > analysis or output can only be obtained by submitting the datafile to the > STARANISO server. Better results would clearly be obtainable if the raw > images for these datasets had been deposited and could be reprocessed, with > the untruncated output of that processing then being submitted to the > STARANISO server (reprocessing the images with autoPROC would combine those > two steps into a single one). > > Most of the deposited datasets have been isotropically truncated, and > their 3D view in WebGL often suggests that this truncation was too drastic. A > number of entries will show infelicities - such as cusps and/or missing > angular ranges, or even stripes caused by gaps between the modules of pixel > detectors if the beam centre is at a position symmetric relative to those > gaps - all marked up in blue. > > > Our purpose in sharing this capability with the community is to bring a > further contribution to the process of making everyone more "data quality > aware" and keen to scrutinise more closely the protocols by which they > collect diffraction data, or have such data collected on their behalf. > > > We will be grateful to receive feedback about PDBpeep, just as we have > been about the STARANISO server itself. > > > With best wishes, > > The STARANISO developers. -- === * * * Gerard Bricogn
Re: [ccp4bb] A new capability on the STARANISO server: "PDBpeep"
Wow, this is really cool--just tried a quick look at a recent membrane protein (5eqi) and you can see right away that there is anisotropy. I would guess this can be found in the literature, but how prevalent is anisotropy in the PDB? Jacob Keller + Jacob Pearson Keller Research Scientist / Looger Lab HHMI Janelia Research Campus 19700 Helix Dr, Ashburn, VA 20147 Desk: (571)209-4000 x3159 Cell: (301)592-7004 + The content of this email is confidential and intended for the recipient specified in message only. It is strictly forbidden to share any part of this message with any third party, without a written consent of the sender. If you received this message by mistake, please reply to this message and follow with its deletion, so that we can ensure such a mistake does not occur in the future. -Original Message- From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Gerard Bricogne Sent: Wednesday, March 28, 2018 11:56 AM To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] A new capability on the STARANISO server: "PDBpeep" Dear all, Ever since the WebGL viewer became available on the STARANISO server, we have found ourselves using it with increasing frequency to take a quick look (a "peep") at the diffraction data deposited with various PDB entries - for example, to try and identify a root cause for some sub-optimal refinement results, or, quite often, just out of sheer curiosity! This involved a totally straightforward procedure whereby the diffraction data file associated with a given PDB entry was downloaded from the PDB and subsequently uploaded to the STARANISO server. Gradually, however, this operation became so popular among some of us that we thought it would be useful to implement this simple procedure as an autonomous capability - and thus was born "PDBpeep" ! You can access this new feature by connecting to http://staraniso.globalphasing.org/cgi-bin/PDBpeep.cgi and enter a PDB code into the box. As indicated on that page, this provides only a cursory look at the overall quality of each dataset, and any further analysis or output can only be obtained by submitting the datafile to the STARANISO server. Better results would clearly be obtainable if the raw images for these datasets had been deposited and could be reprocessed, with the untruncated output of that processing then being submitted to the STARANISO server (reprocessing the images with autoPROC would combine those two steps into a single one). Most of the deposited datasets have been isotropically truncated, and their 3D view in WebGL often suggests that this truncation was too drastic. A number of entries will show infelicities - such as cusps and/or missing angular ranges, or even stripes caused by gaps between the modules of pixel detectors if the beam centre is at a position symmetric relative to those gaps - all marked up in blue. Our purpose in sharing this capability with the community is to bring a further contribution to the process of making everyone more "data quality aware" and keen to scrutinise more closely the protocols by which they collect diffraction data, or have such data collected on their behalf. We will be grateful to receive feedback about PDBpeep, just as we have been about the STARANISO server itself. With best wishes, The STARANISO developers.
[ccp4bb] A new capability on the STARANISO server: "PDBpeep"
Dear all, Ever since the WebGL viewer became available on the STARANISO server, we have found ourselves using it with increasing frequency to take a quick look (a "peep") at the diffraction data deposited with various PDB entries - for example, to try and identify a root cause for some sub-optimal refinement results, or, quite often, just out of sheer curiosity! This involved a totally straightforward procedure whereby the diffraction data file associated with a given PDB entry was downloaded from the PDB and subsequently uploaded to the STARANISO server. Gradually, however, this operation became so popular among some of us that we thought it would be useful to implement this simple procedure as an autonomous capability - and thus was born "PDBpeep" ! You can access this new feature by connecting to http://staraniso.globalphasing.org/cgi-bin/PDBpeep.cgi and enter a PDB code into the box. As indicated on that page, this provides only a cursory look at the overall quality of each dataset, and any further analysis or output can only be obtained by submitting the datafile to the STARANISO server. Better results would clearly be obtainable if the raw images for these datasets had been deposited and could be reprocessed, with the untruncated output of that processing then being submitted to the STARANISO server (reprocessing the images with autoPROC would combine those two steps into a single one). Most of the deposited datasets have been isotropically truncated, and their 3D view in WebGL often suggests that this truncation was too drastic. A number of entries will show infelicities - such as cusps and/or missing angular ranges, or even stripes caused by gaps between the modules of pixel detectors if the beam centre is at a position symmetric relative to those gaps - all marked up in blue. Our purpose in sharing this capability with the community is to bring a further contribution to the process of making everyone more "data quality aware" and keen to scrutinise more closely the protocols by which they collect diffraction data, or have such data collected on their behalf. We will be grateful to receive feedback about PDBpeep, just as we have been about the STARANISO server itself. With best wishes, The STARANISO developers.
