[Chimera-users] Cannot dock molecule to targeted region?
Tom Goddard
goddard at sonic.net
Thu Aug 20 19:02:26 PDT 2020
Hi Yeping,
Chimera does not need high map resolution for fitting -- in fact high resolution (e.g. 4 Angstroms) makes it harder to find the correct orientation of the molecule because there can be many locally optimal fits and Chimera only finds locally optimal fits and has limited search capabilities. So for a high resolution map it is often helpful to first Gaussian smooth it (Tools / Volume Data / Volume Filter or equivalent "vop gaussian" command) to get a lower resolution blob, fit the atomic model in that, then switch to fitting into the high resolution map once the approximately correct orientation is found at lower resolution.
The main difficult fitting problem you are likely to encounter is that Chimera is only fitting the molecule as a rigid body. So if the shape of your atomic model does not match the conformation seen in the map then you won't find any good fit.
Tom
> On Aug 20, 2020, at 6:49 PM, sunyeping <sunyeping at aliyun.com> wrote:
>
> Hi Greg,
>
> Thank you for your help.
> The atomic model of the Ab was built with Rosetta. Maybe one reason it cannot be automatically fitted into the map is that it does not has a correct H-L orientation.
> But is high local resolution of the map a prerequisite for successful automatical fitting with chimera?
>
> Best,
> Yeping
> ------------------------------------------------------------------
> From:Tom Goddard <goddard at sonic.net>
> Sent At:2020 Aug. 21 (Fri.) 00:50
> To:孙业平 <sunyeping at aliyun.com>
> Cc:chimera-users <chimera-users at cgl.ucsf.edu>
> Subject:Re: [Chimera-users] Cannot dock molecule to targeted region?
>
> Hi Yeping,
>
> You have a difficult problem and Chimera is not going to be able to automatically produce the right fit, and neither am I. The preprint on the map you are fitting EMDB 22158 is here
>
> https://www.biorxiv.org/content/10.1101/2020.06.17.153486v2.full <https://www.biorxiv.org/content/10.1101/2020.06.17.153486v2.full>
>
> They do not provide a PDB for the 22158 spike + antibody, although they do provide PDB 6xey for the related EMDB 22156 which is at higher resolution. Why is that? Probably because they are not confident of their antibody fit to 22158 or could not even make a trustworthy model of that antibody given the low resolution. What is the resolution? If you look at figure 13e where they show the local resolution, the antibody region has map resolution in the range 9.5 to 11.5 A. That is definitely not good enough to build an antibody model, nor even good enough to see the beta sheets of the antibody. By comparison look at the higher resolution EMDB 22156 and PDB 6xey they determined shown in figure 12 -- map resolution of around 4-6 A near the antibody, and they show their antibody model and density in figure 12e and you can see it is pretty reasonable.
>
> In summary, I don't know if your antibody atomic model is in the right conformation and you are trying to fit it into low ~10 Angstrom density. Chimera is not able to exhaustively sample configurations and tell you the best and even if it did it could very well be wrong with these uncertainties. Chimera is primarily used to interactively do fitting allowing you to understand whether a reliable fit is even possible.
>
> Good luck!
>
> Tom
>
>
> Fig 13 from paper.
> <F13.large.jpg>
> Fig 12 from paper
>
> <F12.large.jpg>
> On Aug 19, 2020, at 8:59 PM, sunyeping <sunyeping at aliyun.com <mailto:sunyeping at aliyun.com>> wrote:
>
> Hi Tom,
>
> Thank you for your suggestions and instructions. I have tried your method. Yes, the "Optimize correlation" option in the "Fit in Map" tool can avoid large shit of the molecule. However, the problem with this method is that the final fitted result is determined by the initial position of the antibody molecule. I have no idea how the antibody binds the antigen, and my expectation is to find the correct binding positon of the antibody to the antigen by using the fitting tools of chimera. The only thing I know is that the CDRs face of the antibody should be placed towards the antigen. I have put the coordinations for the antigen (Ag, S.pdb) and the antibody (Ab, 4-8_H3-0175.pdb) at the following link:
> https://drive.google.com/drive/folders/16ELIDsUxiNwRGp9ely2SRn-zjAdHCudH?usp=sharing <https://drive.google.com/drive/folders/16ELIDsUxiNwRGp9ely2SRn-zjAdHCudH?usp=sharing>
>
> Firstly, I fitted the Ag to the map for the ag-ab complex (emd_22158.map), then I put the antibody to target area of the map, and then used Fit in Map tool to fit it and get the complex coordinate fit-in-map_1.pdb. After that, I had another try. I put the Ab at the same initial position in the target area but rotated it by 180 degree around its long axis, then fit it, and get complex coordinate fit-in-map_1.pdb. Both of the two complex coordinate can also be found at the above link. In these two complex structures, the heavy chain (H) and light chain (L) of the Ab are totally opposite (a rotation of 180 degree).
> Maybe you could try to do the fitting and find the most reasonable binding mode of the Ab? One clue is that the CDR3 of the H chain (residue 93-107) should usually be involved in contact with the antigen.
>
> Best regards,
> Yeping
> ------------------------------------------------------------------
> From:Tom Goddard <goddard at sonic.net <mailto:goddard at sonic.net>>
> Sent At:2020 Aug. 20 (Thu.) 01:15
> To:孙业平 <sunyeping at aliyun.com <mailto:sunyeping at aliyun.com>>
> Cc:chimera-users <chimera-users at cgl.ucsf.edu <mailto:chimera-users at cgl.ucsf.edu>>; Greg Pintilie <gregdp at gmail.com <mailto:gregdp at gmail.com>>
> Subject:Re: [Chimera-users] Cannot dock molecule to targeted region?
>
> Hi Yeping,
>
> As Greg says Chimera is trying to move your antibody into higher density. Besides the masking he suggests, another thing you can do is just use the Fit in Map tool (not Fit to Segments), place your anti-body by hand in approximately the correct position using the "Active" checkbuttons in model panel and the mouse, and under options in the Fit in Map tool choose "Optimize correlation" instead of "Optimize overlap". The default optimize overlap means to move the molecule into the highest density, but optimizing correlation instead moves it to get the best agreement between the shape of the density. That option will be grayed out until you check "Use map simulated from atoms" since correlation needs a simulated map, and you should enter the approximate resolution of your map. I don't think Fit to Segments offers that option to optimize correlation. When you use Fit in Map it just does a local optimization so your initial placement and orientation of the antibody has to be approximately right.
>
> Tom
>
> <fit_in_map.png>
>
> On Aug 19, 2020, at 5:13 AM, Greg Pintilie <gregdp at gmail.com <mailto:gregdp at gmail.com>> wrote:
>
>
> Hi Yeping,
>
> It seems that the part of the map you are fitting to has lower values, and the Fit-To-Segments procedure is moving the model (your antibody) towards higher values. To prevent this, there is an option in the Fit To Segments dialog ‘Mask map with region(s) to prevent large drifts’. If you enable this option you should not see the model move outside the region. You may have to press the Options button at the bottom of the dialog first to reveal this option if you don’t see it. Also make sure to try ‘Rotational search’ if the default/faster ‘Align principal axes’ doesn’t seem to find the right fit.
>
> There is also the ‘fitmap’ command in Chimera which does a more global search, perhaps the Chimera team may have some advice on it in this scenario, but I think you have the right idea with segmenting first and then doing the fit - it’s a bit more work but you can target the search better with it.
> https://www.cgl.ucsf.edu/chimera/current/docs/UsersGuide/midas/fitmap.html <https://www.cgl.ucsf.edu/chimera/current/docs/UsersGuide/midas/fitmap.html>
>
> Greg
>
>
> On Aug 19, 2020, at 1:53 AM, 孙业平 <sunyeping at aliyun.com <mailto:sunyeping at aliyun.com>> wrote:
>
> Dear all,
>
> I am trying to fit an antibody molecule to the cryo-EM density map for an antigen-antibody complex. The related files can be found at the following link
> https://drive.google.com/drive/folders/16ELIDsUxiNwRGp9ely2SRn-zjAdHCudH?usp=sharing <https://drive.google.com/drive/folders/16ELIDsUxiNwRGp9ely2SRn-zjAdHCudH?usp=sharing>.
> I first segmented the map ("emd_22158.map") and get the segment map ("emd_22158.seg") in chimera. Then I selected the regions corresponding to the antibody and use the "Fit to segments" tool of chimera to fit the antibody molecule. The "image1.png" shows the inital antibody postion and the expected docking postion in the segmented complex density (indicated by the red arrow). However, the antibody molecule cannot be fitted into the selected target region after this operation, but moved to the position shown in "image2.png (only the target regions are shown).
> I have tried to ungoup and regroup the target region to make its overall shape more simial to the antibody, and place the antibody into the region and ajust its orienation, but whatever I tried, the results are the same: the antibody will move to the position shown in "image2.png" after performing the "Fit to segments" action.
> I wonder whether it is impossible to dock the antibody to the expected position with chimera or there is anything wrong with my operation. Could you help me with this issue? I will appreciate any help from you.
>
> Best regards,
> Yeping Sun
> _______________________________________________
> Chimera-users mailing list: Chimera-users at cgl.ucsf.edu <mailto:Chimera-users at cgl.ucsf.edu>
> Manage subscription: https://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users <https://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users>
>
> _______________________________________________
> Chimera-users mailing list: Chimera-users at cgl.ucsf.edu <mailto:Chimera-users at cgl.ucsf.edu>
> Manage subscription: https://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users <https://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users>
>
>
> _______________________________________________
> Chimera-users mailing list: Chimera-users at cgl.ucsf.edu <mailto:Chimera-users at cgl.ucsf.edu>
> Manage subscription: https://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users <https://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users>
>
>
> _______________________________________________
> Chimera-users mailing list: Chimera-users at cgl.ucsf.edu
> Manage subscription: https://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://plato.cgl.ucsf.edu/pipermail/chimera-users/attachments/20200820/c140c99f/attachment-0001.html>
More information about the Chimera-users
mailing list