Structural Bioinformatics and Structural Biology - Science topic

try this one:

all the best

fred

Hi, how did you get the model? did you input the sequence or you downloaded the sequence from their website. It is unlikely that AFold should give homoserine rather than serine since its training material contains natural sequences.

To preserve all the information of a ChimeraX session, you have to save it as a session file (extension .cxs). This enables you to continue the session from exactly the point where you saved it, colouring, orientation and all.

Saving as .pdb would only save the coordinates, no ChimeraX specific information.

You use the .pdb format (or .cif) to transfer the coordinates to other programs which cannot interpret the chimeraX specific information.

See https://www.cgl.ucsf.edu/chimerax/docs/user/commands/save.html to see a listing of all supported formats, and which part of the session they cover - the session file is the only one that covers all the information in your ChimeraX session - but it can only be opened by ChimeraX! All other formats only export the part of the information that can be interpreted by a particular class of other programs, e.g. images, that are 2D views that you can print or insert into a word processing file or a powerpoint presentation.

It depends on how you want to identify domains.

AlphaFold2 uses HMMER(http://hmmer.org/) to search seq DB(Pfam)s, and make alignment files, then based on these alignment files calculate the contact maps for folding. So...

1) aa seq-based domain info:

You can grab it directly from AlphaFold2 after HMMER calculation. However, simply you can put your aa seq into the Pfam website will give a similar result. This kinda method will work only when a similar seq is on the DB. AlphaFold2 does some stupid things when it is given multi-domain protein seq which is not on the Pfam DB.

2) Without having aa seq info, use only the PDB file to predict the domain:

well, this could be a harder question, a) either you write custom python code to calculate the center of mass and define the cutoff value to identify two domains,s or b) use at least two PDB files of the same protein and then align them then calculate the relative deviation of the other. We have a couple of unpublished multi-domain protein structures and compared them with AlphaFold2's performance. it predicted well on domains but not the relative position of each domain.

I think you need to provide more info about why and what is next step or what is the purpose of identifying the domain from AlphaFold2 prediction, if not Pfam domain info?

Unfortunately, most of the databases contain 3D structures only for canonical isoforms. But, you would try Google Colab, a phyton-based online notebook running AlphaFold2, which can predict the structure of any custom sequence or noncanonical isoform. Check this out https://youtu.be/le7NatFo8vI

Thai Leong Lai

Several studies have found that AlphaFold predicts better structures than other approaches. The EBI alphafold database (alphafold.ebi.ac.uk/faq) displays the per-residue confidence score (pLDDT). It provides information about the model's accuracy. The majority of plant proteins in the EBI AlphaFold DB (alphafold.ebi.ac.uk/search/text/Oryza%20sativa%20subsp.%20japonica) have a pLDDT score of >50. Even proteins with less identity appear to be accurately predicted by the AF. For longer proteins and disordered proteins, AlphaFold has some limitations.

Yes, it is mandatory to determine the proper molecular arrangement in space since the atomic coordinates of the protein structures are not energetically favorable. . The aim of energy Minimization is to find a set of coordinates representing the minimum energy conformation for the given coordinates b/w the amino acids. Lower the energy more stable they are.

Also check https://www.mdpi.com/1422-0067/21/15/5183/pdf

A good decoy would be randomizing the sequence of your query peptide, an construct some of them. This would give you a set of same length peptides with the same aa composition. This is similar to the procedure in BLAST for estimating E-values for alignments

Dear Prem Prakash ,

No personal experience but it seems to me that ClusPro as described in their paper https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5540229/ is one of the best performing programs available (and cited >1000 times). A general rule is to never rely on one (bioinformatics) method. So, I would:

-Use at least one other docking protein and realise that no method is 100% accurate (usually round 90% is the best you can get) and every method has their pros and cons

-Rely on the biochemical evidence (more). However, do realise that the method used might not be flawless or is unable to show subtle differences that might be relevant in your case

-Use another (biochemical/biophysical) experimental method to confirm your earlier experimental findings

Best regards.

It is okay to do that. The results are supposed to be machine independent, so that peers can also reproduce the results. Make sure that the software version is the same in all three devices.

You need to have Prime license to run the protein minimization.

I checked your PDB file with both VMD and UCSF chimera. I could not find overlaps or missing .....

Then i checked it with gromacs 5.1.1 and generate a topology file for it ( OPLAS FF)...

Could you please discuss about your exact problem, becuase i think you can minimize your this PDB file with out any problem in Gromacs or VMD or ....

I also checked your PDB file in text format and i could not find any overlap residues,

Please read SMOG help or tutorial for specific instruction of pdb file, hope this help you ....

Please check the attached files

As pointed out, you cannot know all the relationships (the functional form V({ϕi,ψi})) between the different amino acids (sterical clashes, interactions based on charges or hydrophobicity) to predict the energetically favorable combinations of phi and psi and hence need to sample them. The state distribution is affected by the amino acid sequence, may differ with the force fields and simulations methods, does depend on the solvent (ions etc.) and temperature.

Have a look here to see how complex the conformational space of small peptides (13-15) can already be:

Bests

To find such low-identity templates, you use threading algorithms: these do not simply compare two sequences, but basically aligns the query sequence to the template structure: https://en.wikipedia.org/wiki/Threading_(protein_sequence) and evaluates how "happy" your query sequence is in the template fold by appropriate energy functions.

e..g. I-Tasser https://zhanglab.ccmb.med.umich.edu/I-TASSER/

"I-TASSER (Iterative Threading ASSEmbly Refinement) is a hierarchical approach to protein structure and function prediction. It first identifies structural templates from the PDB by multiple threading approach LOMETS, with full-length atomic models constructed by iterative template-based fragment assembly simulations. Function insights of the target are then derived by re-threading the 3D models through protein function database BioLiP. I-TASSER (as 'Zhang-Server') was ranked as the No 1 server for protein structure prediction in recent community-wide CASP7, CASP8, CASP9, CASP10, CASP11, CASP12, and CASP13 experiments. It was also ranked as the best for function prediction in CASP9. The server is in active development with the goal to provide the most accurate structural and function predictions using state-of-the-art algorithms.

(Reference: A. Roy et al. 2010. Nature Protocols 5: 725-738)"

Try Ligandscout or Maestro, Schrodinger.

These days journals demand for 100 ns and Online tools are not reliable for MDS. The best solution for MDS is always gromacs or schrodinger.

Download the 2D and convert into 3D using Marvin sketch or chemsketch or chemdraw.

Hi can you share the notebook ?

Also check https://www.ncbi.nlm.nih.gov/books/NBK26820/

Even in a well-refined structure some residual difference density is always expected. In this case the small negative peak could have to do with imperfect bulk solvent modelling/scaling. I wouldn't worry about it too much, it's very minor in this case. The negative difference density at the carboxylic acid group of the glutamate however could be indicative of radiation damage; in that case refining with lower occupancy can be a solution. Other tell-tale signals for radiation damage are e.g. cleaved Cys-Cys disulfid bonds. HTH, Jonathan.

PROCHECK already accepts .pdb files. As far as I understand, Brookhaven might be an alternative way to call the PDB format (an old one, maybe?). I found some references to it around the internet, e.g. "Brookhaven Protein Databank format".

Anyway, just a remark: check to see if protein chains are specified in the MODELLER output file (sometimes MODELLER doesn't set chain IDs; this can cause problems with PROCHECK). If in doubt, don't use the chain parameter in the PROCHECK command, e.g. `procheck model.B99990004.pdb 2.0`. The resolution parameter is necessary, but any number is accepted (I've used 2.0 in the example above).

@Richard Mariadasse can you suggest me how do you solve the problem?

I'm having the same issue

Hello,

If you want to add a post translational modification in an existing pdb file You can try Vienna 2.0

Best,

Christophe

Dear Pranav Khade

Are you trying to play chess in 3D?

Your idea is a good one, but your assumptions are not.

You need to give a clear definition of paths, so I suggest for you to start in one 3D box, it includes 8 points. I prefer to give each point the following notation

P(i, j, k), so the locations of the 8 points are at

(0,0,0) (1,0,0), (0,1,0), (0,0,1), (1,1,0), (1,0,1),(0,1,1) and (1,1,1).

Study this cube carefully, define each Delanoy edge (axioms of the path), and then add another box, which means 12 points, etc.

If you find the closed formula that allows you to calculate all possible paths from the starting point at the origin to the farthest point at the upper corner of the rectangular box, then you are on the right track.

I wish you good luck.

I think for drawing of protein 3-D structure, softwares such as SPDBV and PyMOL are proper.

This may be too late to be relevant to your work but here's a recommendation anyway.

For a Simulation environment AND cloud service combo that's least expensive (IMO) I'd recommend going with NAMD on AWS.

Here's a helpful link: https://www.ks.uiuc.edu/Research/cloud/

Relatively inexpensive and pretty easy to set up (I think they even have an image that you can just purchase without needing to set much up).

Best of luck.

ModRefiner at (http://zhanglab.ccmb.med.umich.edu/ModRefiner/)

and Galaxyrefiner at (http://galaxy.seoklab.org/)

Fabrizio Fierro great approach. I am doing something similar. I agree that enrichment factor relative to decoy is a key metric for model evaluation. What are the typical values of enrichment did you end up getting?

Yuhong Mao I am not using Mac

@Martiniano Bello , Thank you very much. I have download the golp.tgz and unzip it to the force field folder in gromacs or the input file folder . But I didn't find the relevant force field after runing the 'gmx pdb2gmx' command. Could you explain how to use the Golp for a greenhand？ Thank you.

PyMOl is the batter solution for this for the countdown of the volume of the protein active site as I know. Furthermore, the total volume of the drug-protein complex or only protein volume with the unit is accurately counted by the YASARA dynamics suit.

Thanks

Hi Saman,

"None" refers to no secondary structure predicted, sometimes referred to as random coil.

Arg Nε-Hε and at low pH Arg Nη-Hη and Lys Nζ-Hζ are visible in HSQC but these are aliased/folded peaks. Go for an open Sweep Width (in 15N dimension) HSQC where these peaks will appear at their exact 15N ppm values and you can identify them easily. If you go to the R/K side chains' 15N plane in CBCANH, you will observe CD and CG of R or CE and CD of K. The Trp side-chain usually shifted downfield and appear near the bottom left corner.

Start with the assignment, you will understand which one is the backbone NH because in CBCANH and CACB(CO)NH they will show only the CA and CB (these you have to identify based on their C ppm values). But the side chains will be different.

Best wishes.

Dear Fatemeh,

besides Lepidoptera the other group with well studied sperm dimorphism are the prosobranch gastropods. My studies of Serpulorbis strongly suggest, that the "atypical" spermatozoa supply (or are) nutrition to the typical spermatozoa (which are the regular, fertile spermatozoa). There might be some similarity in your objects of research.

Try IonCom (Zhang Lab) for any metal binding prediction.

Pymol software

You can open your pdb in Chimera, go to Tools - Sequence - Sequence and it will give you the answer

Hi Bala,

I should see your question earlier :) Our CASTp Server would solve your problem:

Thanks Jeff! Now we've upgraded CASTp into new 3.0 version, this link is correct and didn't change: http://sts.bioe.uic.edu/castp/

Jeffrey R Brender Ruhshan Ahmed Abir

Because proteins are amphiphilic and can act as surfactants in a similiar way detergents do. They can adsorb at air-water interface and lower the surface tension of the solution which allows formation of the foam.

Dawid

Such short peptides usually do not assume a single defined structure in solution. The structure of the peptide in the complex is not necessarily the dominant conformation in the ensemble of structures in solution, but induced by the interaction with the binding partner. As a consequence, rigid docking of peptides usually is not possible.

Go through this article:

Pharmacophore based virtual screening for identification of marine bioactive compounds as inhibitors against Mip protein of Chlamydia trachomatis. February 2016, RSC Advances 6(23):18946-18957

DOI: 10.1039/C5RA24999F

Since you got both regions from PDB and the hinge region could be found at the files shown by Honegger, you may use any homology modeling tool (such as modeller) to create a full protein based on these tridimensional structures and the amino-acid sequence of your antibody.

There's a free online version of ChemDraw, which is kinda cool.

Few notes...

1. RMSD below one Angstrom is not big.

2. If you increase the y-axis upper limit from one to two Angstroms, the RMSD line will appear less noisy, and your conclusion about the (in)stability of the protein may change.

3. If you decrease the number of y-axis ticks from ten to five, the RMSD line will appear less noisy, and your conclusion about the (in)stability of the protein may change.

4. By plotting moving average, you can smooth out most of the noise, and your conclusion about the (in)stability of the protein may change.

5. Calculate RMSD after aligning different parts of the protein to identify which parts contribute most to the total RMSD value.

6. Some twenty years ago people were able to achieve excellent quantitative agreement with commensurable experimental data using structures and energies obtained from 500 ps (yes, picosecond) MD trajectories.

7. MD simulation is most useful if used for sampling of the configurational space near the initial structure---and for this purpose, in most cases, i would say, the shorter the trajectory, the better; rather than running one long MD simulation, it is generally better to run many short MD simulations, starting from different initial conditions---this is also very useful if one wants to assess the uncertainty of the quantitative results.

Hi Mubasher, may I suggest you taking a look at the web application Kdeep, available free of charge at the PlayMolecule suite (www.playmolecule.org). Convolutional neural networks are used to predict binding affinity (Kd and binding free energy) for a set of docked protein-ligand complexes. Therefore, since you seem to have the binding pose of the ligand, you can input the ligand in its docked pose in SDF format, plus the protein in PDB format and Kdeep will return you a predicted affinity. I hope it is useful for you :)

Dear BreAnne,

actually, it's highly difficult to predict an interaction between two proteins because, as said Steingrimur, it might be strongly related to the 3D structure.

However, there is case for which it becomes a bit less difficult: when the protein binds to a linear motif. A motif is a short amino acid sequence (of about 3 to 6 residues). In this case, the interaction is mostly driven by the sequence since the motif is unfolded most of the time.

You may find interesting details at the ELM website (http://elm.eu.org/), which stands for Eukaryotic Linear Motif. It's a manually curated database based on literature. You enter a protein sequence and you'll see all the putative motifs bound to it. Then it's also possible to search for a special motif if you already know it.

Best regards

Yves

Hi, Kirill! you can try fingerprints based clustering, e.g. RDkit for python can do this (http://www.rdkit.org/docs/Cookbook.html#clustering-molecules)

Few tricks be needed to carryout successfully completion.

1) Update the AD4_parameter.dat according to your necessary metal

2) Update the AD4.1_bound.dat also according to your necessary metal

3) then copy that two files into your installation folder

4) then during grid generation and docking dpf file generation steps, there at set map types option you should mention the required metal atom types and also there is a option named as other option; there u should mention the newly generated parameter file

5) finally the newly generated 2 files should be kept on your docking directory

Finally docking should successfully run.

take a look at this video

best,

Thank you all for your kindly help!

You should first check the literature to see if this selenocysteine is actually functional. Sometimes the heavy atom derivative of a side chain (e.g. selenomethionine, selenocysteine) is used to simply get better diffraction of crystals, and the actual enzyme has the normal sulfur-containing amino acids. If that's the case, just replace Se by S and proceed as normal.

If it really is a selenocysteine, you will have to parametrize it. To do this, you must follow the prescribed parametrization procedure of the parent force field. This can be quite laborious to do.

You should be careful with such a big system because the degrees of freedom is too high. Optimization process in Gaussian does not do a conformational search. It only does a single point optimization. Therefore, you should be very sure about the initial guess structure (input structure). Before going for QM, you may try conformational search or simulated annealing to generate a good starting structure.

Hi Stellamaris, what I read from your question is that you are describing receptor antagonists, or competitors.

Matching the ligand binding site of a receptor relies on designing an antagonist that looks like a ligand (based on topology and computer docking programs), but  does not act as a natural ligand and thus inhibits the receptor.    

@Woojong Lee  I already tried Ultrafiltration using amicon filters according to the guideline. Thank you, I will try it again with more concern. 

HI Rohit,

There are other servers like Zdock , ,Hex, Cluspro, Gramm-X , as and alternative there is Megadock4.0 , (it does not have a server you would have to install it ). However, I think they still will have the same problem due to the large size of your dimer.

One of the options you may try is to extract the interface of the protein and use that as the protein receptor in the docking. Your next option is to establish a collaboration with one of the groups so they run their programs locally in their clusters, with the whole protein. 

From the Dock 6 documentation:

"SASA_Score:

#percentage of ligand exposed times 100"

So your SASA_Descriptor Score of 14.15 is telling you that just over 14% of the molecule is exposed to ligand (or that just under 86% is buried in the receptor).

Higher scores mean that more of your molecule is sticking out into water.  Lower scores mean that more of the molecule is buried in the protein.  For drug applications, you'll probably want to go with the lowest score when choosing an ideal molecule to pursue.  Natural ligands lying on surface grooves might have very different priorities when deciding on an optimum pose.  So which score is 'best' will depend upon what question you're asking.

This is embarrassing. I just realized that I never answered your follow-up question. Unfortunately, even if I gave the best possible answer now, it would most likely prove absolutely useless to you. I apologize.

Dear collegues,

I have received an answer from Proffessor Sanner (Michel Sanner), who works at the Scripps Research Institute. 

I hope it can  also be useful for you all:

"I have been  working in a version of AutoDock that supports making receptor side  chain flexible and as part of these software developments I wrote the  AGFR software program.

This command line program allows the compute grids (using AutoGrid4) and supports a range of options for automatically positioning of docking

boxes, including using: a known ligand, or a set of residues in the

receptor, a binding pocket identified by our pocket finding method

AutoSite, etc...

The software is not 100% ready for release yet but if you want to give

it a shot you can download the code from

please look at the Use cases on that web page for examples on how to use  the software. (Be aware though that currently the graphical user  interface AGFRgui is broken) but the command like version should be working.

The .trg file contains is a zip file that contains all the maps generated by AutoGrid4 and that can be used for AutoDock4.

So you have to unzip the .trg file and give AutoDock4 the the .map files located in the folder created by unzip.

> unzip myfile.trg

Archive: myfile.trg

inflating myfile/rigidReceptor.C.map

inflating .. "

Thank you all for your help! I really appreciate it!

Huge hugs 

Stellamaris

Hello,

I have experienced this as well.  You indicated that your solution was fully clear before autoclaving, I usually heat it to boiling to be extra careful because if it is not completely dissolved it will present with results like what you are seeing.  I also experienced an issue where my solution was being autoclaved for 30 minute liquid cycle instead of 15 minute liquid cycle, this caused a similar situation as well. I believe it ended up having to do with the glycerol content of the media being very sensitive to extended heat exposure.  Once I corrected these two problems I had not more precipitate at the bottom of my solution after autoclaving.  If the precipitate is of a brown color, which I am assuming based on your 'sandy' description, I would assume something is actually burning or denaturing in the autoclave and double check the things I listed above.  I purchase my 7H9 from Hardy Diagnostics and I have on a couple occasions received sub-par supplies from them so, it also may not be anything you are doing.

Best

I would suggest you to target, cell wall synthesizing protein. 

There several ubiquitin-related databases out there, and the most comprehensive one I know is https:/www.omicstools.com.  It has at least nine tools for use in querying about ubiquitin:

UbiNet:

Offers users an effective platform to efficiently study protein ubiquitylation networks among large-scale ubiquitylation data. The current version of UbiNet was designed specifically for humans to…

E3Net:

Provides a comprehensive collection of available E3-substrate specificities and a systematic framework for the analysis of E3-mediated regulatory networks of diverse cellular functions. Currently,…

mUbiSiDa:

mammalian Ubiquitination Site Database; Provides a scientific community with a comprehensive, freely and high-quality accessible resource of mammalian protein ubiquitination sites. The mUbiSiDa was designed to be a widely used tool for…

PlantsUBQ:

Describes the network of Arabidopsis proteins responsible for the covalent attachment of ubiquitin (Ub).  PlantsUBQ provides comprehensive information on the role of this post-translational…

hUbiquitome:

A public resource for the retrieval of experimentally verified human ubiquitination enzymes and substrates. hUbiquitome is the first comprehensive database of human ubiquitination cascades.…

plantsUPS:

A database of higher plants' ubiquitin 26S/proteasome system (UPS). Both automated search and manual curation were performed in identifying candidate genes. Extensive annotations referring to…

UUCD:

Ubiquitin and Ubiquitin-like Conjugation Database; A family-based database for ubiquitin and ubiquitin-like conjugation, which is one of the most important post-translational modifications responsible for regulating a variety of cellular processes,…

UbiProt:

A knowledge base of ubiquitylated proteins. UbiProt contains retrievable information about overall characteristics of a particular protein, ubiquitylation features, related ubiquitylation and…

SCUD:

Saccharomyces Cerevisiae Ubiquitination Database; A web-based database for the ubiquitination system in Saccharomyces cerevisiae (Baker's yeast). SCUD aims to represent a comprehensive yeast ubiquitination system, and is easily expandable with…

What is "possible" is limited only by the quality of the hardware and software (and how much patience you have for waiting on the results). Modern simulations often go well into the microsecond range pretty easily.

The better question is what is necessary to answer the question(s) at hand, and sampling is a general concern. A single, microsecond-length simulation may be less valuable than, say, 10 x 100-ns simulations, but of course that again depends on the purpose(s) of the simulation. Long simulations have exposed the fact that conformations often get trapped in local minima for long periods of time. This can be circumvented with enhanced sampling techniques or by simply running many more simulations from different starting conformations and/or initial (random) velocities.

PDB codes 1DR1, 1DR3, 1DR4, 1DR6 in th eprotein data bank (rcsb.org) is what you are looking for. Relevant publication is:

Crystal structure of chicken liver dihydrofolate reductase complexed with NADP+ and biopterin.

McTigue, M.A., Davies 2nd., J.F., Kaufman, B.T., Kraut, J.

(1992) Biochemistry 31: 7264-7273

The Protein-Ligand-Interaction Profiler https://projects.biotec.tu-dresden.de/plip-web/plip/ adds some details that are not covered by HBplus. Besides the Eeb interface, a command-line driven version is available for download.

I noticed that you have 19 Atoms in topology file (psf) for the first residue, but in the PDB file you have 18 atoms. There should be the same number, are they both created in CHARMM?

When you try to chose which software to use for running MD simulations, there are a few criteria that you may prioritize, such as:

1) possibility of modeling different biomolecular systems (e.g., protein, membranes, DNA, RNA and their complexes);

2) possibility on modeling the solvent environment explicitly and/or implicitly (water, ions, etc,);

3) flexibility on using different force fields for parametrization of the systems and efficient treatment of long-range electrostatic interactions;

4) computation efficiency (in terms of speed) for running long time dynamics (e.g., is the code parallelized, can you use GPU, etc,);

5) possibility on running the simulations in different statistical ensembles (e.g., NVT, NPT) and possibility of using different periodic boundary conditions based on the geometry of the simulation box;

6) implementation of stable numerical integrator methods for solving equations of motion;

7) being easy on building up equilibration protocols (e.g., through documentation and tutorials);

8) providing tools and/or modules for analyzing the trajectories;

9) being able to create outputs in a format which can be easy read from other data visualization and/or analyzing software;

10) being able to read and manipulate the inputs structures (e.g., PDB format files);

11) in some cases, being able to compute complex thermodynamic properties, such as free energies, using advanced molecular dynamics methods;

12) implementation of efficient conformation sampling algorithms;

13) and maybe others (depending on the problem).

Some molecular dynamics engines that satisfy the above mentioned criteria include CHARMM, GROMACS, NAMD, AMBER, DESMOND, TINKER, DL_POLY, and maybe others.

I hope that this helps this discussion!

thank you madhusudan

Measure the effect of temperature on the UV absorbance of the DNA at different pHs. As the DNA denatures with heating, the UV absorbance increases. This effect is used to measure the melting temperature of the DNA. The equipment required for this experiment is a spectrophotometer with a thermostatted cuvette chamber.

Another way to do the melting experiment is to use a differential scanning calorimeter, which measures the heat absorbed as the DNA melts during heating.

Please see

Decreasing intensity and broadening of both the positive and the negative CD bands leads to a shift in the x-axis intersection point towards higher wavelengths, and both bands also exhibit a slight red shift. These changes can be attributed to absorption flattening and differential scattering phenomena. Absorption flattening is a consequence of the non-random distribution of chromophores within the sample, i.e. to differential absorption from different portions of the sample. In peptide-membrane samples it occurs when the peptide chromophores are sequestered in discrete regions of the sample with high local density. The extent of absorption flattening is proportional to the concentration of absorbers in each particle [43]. Thus, with increasing binding, more and more peptides seem to form extended β-pleated aggregates on the bilayer and thereby enhance the absorption flattening effects in the CD spectrum. Differential scattering of the incident radiation becomes more and more important with increasing aggregate size [44]. Thus, the more extended the β-sheets, the more severe is the loss in spectral intensity.