Restoration of missing helical fragments of the structure
Dear Chimera-X users ! Suppose I have a pdb structure with a missed C-terminal fragment corresponding to the transmembrane helical part that should anchor the protein in the membrane. Is it possible to use some built-in utilities (modeller?) to predict this missing fragment based on the full length sequence of the entire construct? Many thanks in advance Enrico
Hello, (A) yes you can use the ChimeraX tool Model Loops to model missing parts with Modeller as long as these parts are included in your sequence. See tool help for description and instructions: <https://rbvi.ucsf.edu/chimerax/docs/user/tools/modelloops.html> However, that may not produce a helical conformation. Alternatively there are other things you could try: (B) See if that protein is already in the AlphaFold Database or ESM Metagenomic Atlas of already-predicted structures. Often the prediction will be almost the same as the known part and additionally have the unknown part. See ChimeraX tools AlphaFold and ESMFold, and/or commands "alphafold match" "alphafold search" "esmfold match" "esmfold search" <https://rbvi.ucsf.edu/chimerax/docs/user/tools/alphafold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/tools/esmfold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/alphafold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/esmfold.html> (C) If not in already-predicted structure databases, you could try running a new prediction with AlphaFold (if the sequence is not too long) or ESMFold. See links above for the tools, and/or commands "alphafold predict" "esmfold predict" These other approaches (B,C) might not produce a helix either, although my guess is that they are more likely than (A) to do so. If you absolutely want a helix there is yet another approach: (D) use Build Structure, Start Structure and build a helical peptide. Then just stick the peptide model onto your original protein model with the Join Models section of Build Structure. See tool help for description and instructions: <https://rbvi.ucsf.edu/chimerax/docs/user/tools/buildstructure.html> I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On Aug 20, 2024, at 2:52 AM, Enrico Martinez via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Dear Chimera-X users !
Suppose I have a pdb structure with a missed C-terminal fragment corresponding to the transmembrane helical part that should anchor the protein in the membrane. Is it possible to use some built-in utilities (modeller?) to predict this missing fragment based on the full length sequence of the entire construct?
Many thanks in advance
Enrico
Hello Elaine, Thank you very much for these suggestions ! I've already tried the model loop plugin using my pdb with the full-length sequence but indeed it predicted the missed fragment as a disordered fragment. I have a question regarding alpha fold plugin that I tried with the same initial data: open my_pdb_with_missed.pdb open ./sequence.fasta sequence associate /A alphafold match #1/A This took only in account the sequence of the pdb (with missed fragment) but not the whole sequence that was associated to this pdb. I also tried alphafold match sequence.fasta but it did not work. So how could I use the alphafold searching using the both the conformation with the missed parts and the whole sequence (in the same manner like we do it with modeller)? Many thanks in advance Enrico Il giorno mar 20 ago 2024 alle ore 17:20 Elaine Meng <meng@cgl.ucsf.edu> ha scritto:
Hello, (A) yes you can use the ChimeraX tool Model Loops to model missing parts with Modeller as long as these parts are included in your sequence. See tool help for description and instructions: <https://rbvi.ucsf.edu/chimerax/docs/user/tools/modelloops.html>
However, that may not produce a helical conformation.
Alternatively there are other things you could try:
(B) See if that protein is already in the AlphaFold Database or ESM Metagenomic Atlas of already-predicted structures. Often the prediction will be almost the same as the known part and additionally have the unknown part.
See ChimeraX tools AlphaFold and ESMFold, and/or commands "alphafold match" "alphafold search" "esmfold match" "esmfold search" <https://rbvi.ucsf.edu/chimerax/docs/user/tools/alphafold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/tools/esmfold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/alphafold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/esmfold.html>
(C) If not in already-predicted structure databases, you could try running a new prediction with AlphaFold (if the sequence is not too long) or ESMFold. See links above for the tools, and/or commands "alphafold predict" "esmfold predict"
These other approaches (B,C) might not produce a helix either, although my guess is that they are more likely than (A) to do so. If you absolutely want a helix there is yet another approach:
(D) use Build Structure, Start Structure and build a helical peptide. Then just stick the peptide model onto your original protein model with the Join Models section of Build Structure. See tool help for description and instructions: <https://rbvi.ucsf.edu/chimerax/docs/user/tools/buildstructure.html>
I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On Aug 20, 2024, at 2:52 AM, Enrico Martinez via ChimeraX-users < chimerax-users@cgl.ucsf.edu> wrote:
Dear Chimera-X users !
Suppose I have a pdb structure with a missed C-terminal fragment corresponding to the transmembrane helical part that should anchor the protein in the membrane. Is it possible to use some built-in utilities (modeller?) to predict this missing fragment based on the full length sequence of the entire construct?
Many thanks in advance
Enrico
You can use "alphafold match" with the sequence that you opened instead of the pdb structure. Association does not matter, and in fact you don't even need to have the structure open. The problem is that you did not specify the sequence correctly. The help page says how to specify the sequence with "alphafold match": <https://rbvi.ucsf.edu/chimerax/docs/user/commands/alphafold.html#fetch> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/seqspec.html> example commands: open ~/Desktop/stash/insulin.fa alphafold match insulin.fa:1 ... meaning to use sequence 1 in alignment (which is really a single sequence) with ID "insulin.fa" -- After you open the sequence, its ID is shown in the title bar of the sequence window. I hope this helps, Elaine
On Aug 21, 2024, at 2:52 AM, Enrico Martinez <jmsstarlight@gmail.com> wrote:
Hello Elaine,
Thank you very much for these suggestions !
I've already tried the model loop plugin using my pdb with the full-length sequence but indeed it predicted the missed fragment as a disordered fragment. I have a question regarding alpha fold plugin that I tried with the same initial data:
open my_pdb_with_missed.pdb open ./sequence.fasta sequence associate /A
alphafold match #1/A
This took only in account the sequence of the pdb (with missed fragment) but not the whole sequence that was associated to this pdb. I also tried alphafold match sequence.fasta but it did not work.
So how could I use the alphafold searching using the both the conformation with the missed parts and the whole sequence (in the same manner like we do it with modeller)?
Many thanks in advance
Enrico
Il giorno mar 20 ago 2024 alle ore 17:20 Elaine Meng <meng@cgl.ucsf.edu> ha scritto: Hello, (A) yes you can use the ChimeraX tool Model Loops to model missing parts with Modeller as long as these parts are included in your sequence. See tool help for description and instructions: <https://rbvi.ucsf.edu/chimerax/docs/user/tools/modelloops.html>
However, that may not produce a helical conformation.
Alternatively there are other things you could try:
(B) See if that protein is already in the AlphaFold Database or ESM Metagenomic Atlas of already-predicted structures. Often the prediction will be almost the same as the known part and additionally have the unknown part.
See ChimeraX tools AlphaFold and ESMFold, and/or commands "alphafold match" "alphafold search" "esmfold match" "esmfold search" <https://rbvi.ucsf.edu/chimerax/docs/user/tools/alphafold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/tools/esmfold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/alphafold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/esmfold.html>
(C) If not in already-predicted structure databases, you could try running a new prediction with AlphaFold (if the sequence is not too long) or ESMFold. See links above for the tools, and/or commands "alphafold predict" "esmfold predict"
These other approaches (B,C) might not produce a helix either, although my guess is that they are more likely than (A) to do so. If you absolutely want a helix there is yet another approach:
(D) use Build Structure, Start Structure and build a helical peptide. Then just stick the peptide model onto your original protein model with the Join Models section of Build Structure. See tool help for description and instructions: <https://rbvi.ucsf.edu/chimerax/docs/user/tools/buildstructure.html>
I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On Aug 20, 2024, at 2:52 AM, Enrico Martinez via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Dear Chimera-X users !
Suppose I have a pdb structure with a missed C-terminal fragment corresponding to the transmembrane helical part that should anchor the protein in the membrane. Is it possible to use some built-in utilities (modeller?) to predict this missing fragment based on the full length sequence of the entire construct?
Many thanks in advance
Enrico
Thank you very much Elaine ! Actually I found a similar possibility via providing a uniptot id .. My question was related to combining the both sequence and spefic conformation together to build the missing fragments directly in this structure (like we do it with modeller) using IA predictions based on the whole sequence (instead of homology modeling). This feature would be very cool since it can allows to leverage IA predictions directly with the experimental data. Anyway, suppose I have an AF prediction and an open x-ray structure in another model that I have superimposed via Match Macker. Would it be possible to further alter the positions of the fragments in the AF prediction to make it look like the X-ray (e.g. by morphing?)? Il giorno mer 21 ago 2024 alle ore 17:15 Elaine Meng <meng@cgl.ucsf.edu> ha scritto:
You can use "alphafold match" with the sequence that you opened instead of the pdb structure. Association does not matter, and in fact you don't even need to have the structure open. The problem is that you did not specify the sequence correctly.
The help page says how to specify the sequence with "alphafold match": <https://rbvi.ucsf.edu/chimerax/docs/user/commands/alphafold.html#fetch> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/seqspec.html>
example commands:
open ~/Desktop/stash/insulin.fa alphafold match insulin.fa:1
... meaning to use sequence 1 in alignment (which is really a single sequence) with ID "insulin.fa" -- After you open the sequence, its ID is shown in the title bar of the sequence window.
I hope this helps, Elaine
On Aug 21, 2024, at 2:52 AM, Enrico Martinez <jmsstarlight@gmail.com> wrote:
Hello Elaine,
Thank you very much for these suggestions !
I've already tried the model loop plugin using my pdb with the full-length sequence but indeed it predicted the missed fragment as a disordered fragment. I have a question regarding alpha fold plugin that I tried with the same initial data:
open my_pdb_with_missed.pdb open ./sequence.fasta sequence associate /A
alphafold match #1/A
This took only in account the sequence of the pdb (with missed fragment) but not the whole sequence that was associated to this pdb. I also tried alphafold match sequence.fasta but it did not work.
So how could I use the alphafold searching using the both the conformation with the missed parts and the whole sequence (in the same manner like we do it with modeller)?
Many thanks in advance
Enrico
Il giorno mar 20 ago 2024 alle ore 17:20 Elaine Meng <meng@cgl.ucsf.edu> ha scritto: Hello, (A) yes you can use the ChimeraX tool Model Loops to model missing parts with Modeller as long as these parts are included in your sequence. See tool help for description and instructions: <https://rbvi.ucsf.edu/chimerax/docs/user/tools/modelloops.html>
However, that may not produce a helical conformation.
Alternatively there are other things you could try:
(B) See if that protein is already in the AlphaFold Database or ESM Metagenomic Atlas of already-predicted structures. Often the prediction will be almost the same as the known part and additionally have the unknown part.
See ChimeraX tools AlphaFold and ESMFold, and/or commands "alphafold match" "alphafold search" "esmfold match" "esmfold search" <https://rbvi.ucsf.edu/chimerax/docs/user/tools/alphafold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/tools/esmfold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/alphafold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/esmfold.html>
(C) If not in already-predicted structure databases, you could try running a new prediction with AlphaFold (if the sequence is not too long) or ESMFold. See links above for the tools, and/or commands "alphafold predict" "esmfold predict"
These other approaches (B,C) might not produce a helix either, although my guess is that they are more likely than (A) to do so. If you absolutely want a helix there is yet another approach:
(D) use Build Structure, Start Structure and build a helical peptide. Then just stick the peptide model onto your original protein model with the Join Models section of Build Structure. See tool help for description and instructions: <https://rbvi.ucsf.edu/chimerax/docs/user/tools/buildstructure.html>
I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On Aug 20, 2024, at 2:52 AM, Enrico Martinez via ChimeraX-users < chimerax-users@cgl.ucsf.edu> wrote:
Dear Chimera-X users !
Suppose I have a pdb structure with a missed C-terminal fragment corresponding to the transmembrane helical part that should anchor the protein in the membrane. Is it possible to use some built-in utilities (modeller?) to predict this missing fragment based on the full length sequence of the entire construct?
Many thanks in advance
Enrico
I don't know of any way to combine the modeling techniques into one. Instead you could try to manually combine your different models somehow. As far as I know, there is no scientifically "established" way to combine the structures and no simple recipe, but of course you could morph, or you could just join some residues from one model to the other model with the Join Models part of the Build Structure tool. Of course if you publish this model, you should probably describe whatever procedure you used to make it. <https://rbvi.ucsf.edu/chimerax/docs/user/commands/morph.html> <https://rbvi.ucsf.edu/chimerax/docs/user/tools/buildstructure.html> Elaine
On Aug 22, 2024, at 1:16 AM, Enrico Martinez <jmsstarlight@gmail.com> wrote:
Thank you very much Elaine ! Actually I found a similar possibility via providing a uniptot id .. My question was related to combining the both sequence and spefic conformation together to build the missing fragments directly in this structure (like we do it with modeller) using IA predictions based on the whole sequence (instead of homology modeling). This feature would be very cool since it can allows to leverage IA predictions directly with the experimental data.
Anyway, suppose I have an AF prediction and an open x-ray structure in another model that I have superimposed via Match Macker. Would it be possible to further alter the positions of the fragments in the AF prediction to make it look like the X-ray (e.g. by morphing?)?
Il giorno mer 21 ago 2024 alle ore 17:15 Elaine Meng <meng@cgl.ucsf.edu> ha scritto: You can use "alphafold match" with the sequence that you opened instead of the pdb structure. Association does not matter, and in fact you don't even need to have the structure open. The problem is that you did not specify the sequence correctly.
The help page says how to specify the sequence with "alphafold match": <https://rbvi.ucsf.edu/chimerax/docs/user/commands/alphafold.html#fetch> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/seqspec.html>
example commands:
open ~/Desktop/stash/insulin.fa alphafold match insulin.fa:1
... meaning to use sequence 1 in alignment (which is really a single sequence) with ID "insulin.fa" -- After you open the sequence, its ID is shown in the title bar of the sequence window.
I hope this helps, Elaine
On Aug 21, 2024, at 2:52 AM, Enrico Martinez <jmsstarlight@gmail.com> wrote:
Hello Elaine,
Thank you very much for these suggestions !
I've already tried the model loop plugin using my pdb with the full-length sequence but indeed it predicted the missed fragment as a disordered fragment. I have a question regarding alpha fold plugin that I tried with the same initial data:
open my_pdb_with_missed.pdb open ./sequence.fasta sequence associate /A
alphafold match #1/A
This took only in account the sequence of the pdb (with missed fragment) but not the whole sequence that was associated to this pdb. I also tried alphafold match sequence.fasta but it did not work.
So how could I use the alphafold searching using the both the conformation with the missed parts and the whole sequence (in the same manner like we do it with modeller)?
Many thanks in advance
Enrico
Il giorno mar 20 ago 2024 alle ore 17:20 Elaine Meng <meng@cgl.ucsf.edu> ha scritto: Hello, (A) yes you can use the ChimeraX tool Model Loops to model missing parts with Modeller as long as these parts are included in your sequence. See tool help for description and instructions: <https://rbvi.ucsf.edu/chimerax/docs/user/tools/modelloops.html>
However, that may not produce a helical conformation.
Alternatively there are other things you could try:
(B) See if that protein is already in the AlphaFold Database or ESM Metagenomic Atlas of already-predicted structures. Often the prediction will be almost the same as the known part and additionally have the unknown part.
See ChimeraX tools AlphaFold and ESMFold, and/or commands "alphafold match" "alphafold search" "esmfold match" "esmfold search" <https://rbvi.ucsf.edu/chimerax/docs/user/tools/alphafold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/tools/esmfold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/alphafold.html> <https://rbvi.ucsf.edu/chimerax/docs/user/commands/esmfold.html>
(C) If not in already-predicted structure databases, you could try running a new prediction with AlphaFold (if the sequence is not too long) or ESMFold. See links above for the tools, and/or commands "alphafold predict" "esmfold predict"
These other approaches (B,C) might not produce a helix either, although my guess is that they are more likely than (A) to do so. If you absolutely want a helix there is yet another approach:
(D) use Build Structure, Start Structure and build a helical peptide. Then just stick the peptide model onto your original protein model with the Join Models section of Build Structure. See tool help for description and instructions: <https://rbvi.ucsf.edu/chimerax/docs/user/tools/buildstructure.html>
I hope this helps, Elaine ----- Elaine C. Meng, Ph.D. UCSF Chimera(X) team Resource for Biocomputing, Visualization, and Informatics Department of Pharmaceutical Chemistry University of California, San Francisco
On Aug 20, 2024, at 2:52 AM, Enrico Martinez via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:
Dear Chimera-X users !
Suppose I have a pdb structure with a missed C-terminal fragment corresponding to the transmembrane helical part that should anchor the protein in the membrane. Is it possible to use some built-in utilities (modeller?) to predict this missing fragment based on the full length sequence of the entire construct?
Many thanks in advance
Enrico
participants (2)
-
Elaine Meng -
Enrico Martinez