## Warning: package 'knitr' was built under R version 3.5.2

Learning Objectives

  • Upload a sequence from UniProt to Chimera
  • Perform a Blast search against the PDB
  • Select a template
  • Build a model
  • Edit, evaluate and refine the model

Introduction

The MODELLER software1 is used for homology or comparative modeling of protein three-dimensional structures (Webb and Sali 2016, Marti-Renom et al. (2000)) based on an alignment of a sequence to be modeled with known related structure(s). MODELLER automatically calculates one or multiple models containing all non-hydrogen atoms based on a comparative protein structure modeling algorithm by satisfaction of spatial restraints (Sali and Blundell 1993, Fiser, Do, and Sali (2000)).

MODELLER is a set of python scripts and tables that can perform many additional tasks, including de novo modeling of loops in protein structures, optimization of various models of protein structure […].

Therefore the default method to MODELLER is via python scripts and line command.

There are various commercial and non commercial packages that exist as “wrappers” around MODELLER to make use of a Graphical User Interface (GUI) and this workshop is dedicated to the GUI offered by the molecular graphics software UCSF Chimera2 (Pettersen et al. 2004).

MODELLER process

Regardless of the method used, the steps from sequence to model are the same. However, it should be noted that the quality of the alignment between the target sequence and the 3D template(s) will determine the quality of the final model.

The main steps of the procedure can be visually summarized:

MODELLER process flow.

Figure 1.

In more details, homology modeling is a multi-step process:

  1. Template recognition and initial alignment
  2. Alignment correction
  3. Backbone generation
  4. Loop modeling
  5. Side chain modeling
  6. Model optimisation
  7. Model validation
  8. Iteration to correct mistakes of flaws (if any)

Prerequisites

Attendees should have some basic knowledge of using Chimera and MODELLER from personal experience or from attending previous workshops on this subject.

Register on MODELLER web site:

Modeller is 9.18 is intalled on all the iMacs.

However, each user should register with the web site to obtain the install keyword at https://salilab.org/modeller/registration.html

Set-up

Target and Template

In comparative (homology) modeling, theoretical models of a protein are built using at least one known related structure and a sequence alignment of the known and unknown structures. The protein to be modeled is the target, and a related known structure used for modeling is a template.

The target in this tutorial is the human GADD45 beta. This protein is involved in the regulation of growth and apoptosis. Its gene is a member of a group of genes whose transcript levels are increased following stressful growth arrest conditions and treatment with DNA-damaging agents. The genes in this group respond to environmental stresses by mediating activation of the p38/JNK pathway.

At the time of creating this tutorial (May 2016), no structure was available for GADD45 beta.

Its recommended name is Growth arrest and DNA damage-inducible protein GADD45 beta (Uniprot ID: O75293 (GA45B_HUMAN), gene name: GADD45B.)

The FASTA format for this sequence can be displayed from this web site and looks like this:

>sp|O75293|GA45B_HUMAN Growth arrest and DNA damage-inducible protein GADD45 beta OS=Homo sapiens GN=GADD45B PE=1 SV=1
MTLEELVACDNAAQKMQTVTAAVEELLVAAQRQDRLTVGVYESAKLMNVDPDSVVLCLLA
IDEEEEDDIALQIHFTLIQSFCCDNDINIVRVSGMQRLAQLLGEPAETQGTTEARDLHCL
LVTNPHTDAWKSHGLVEVASYCEESRGNNQWVPYISLQER

Chimera

For this workshop we will use the iMacs onto which have been installed both Chimera and MODELLER. However, the process, method and graphical interface would be exactly the same on a Windows or Linux computer.

This button will invite you to act: Open the Chimera software.

  • On a Macintosh Chimera is installed in /Applications.
  • Double click on the Chimera icon.
  • You can also find Chimera by using Spotlight Search: the “magnifying glass” icon on the top right of the Mac.

Note: On a Windows computer type Chimera within the botton left Start or Cortna space. On Linux either find the Chimera icon or type chimera at the shell prompt. Once Chimera is started the rest of the method is identical on all platforms.

Tutorial caveats:

  • As database contents and web services are updated, the results of calculations are likely to differ from what is described here. However, this tutorial is meant to illustrate the general process rather than any specific result.
  • This tutorial is not meant to indicate the optimal parameter settings for comparative modeling, as these will vary depending on the system of interest and the information available at the time.
  • This tutorial describes only one of several equally valid approaches. For example, the sequence of the target could be read from a FASTA file instead of fetched from the UniProt database.

A few of the commands will be shown as line commands. Therefore the line command option needs to be engaged, we can do it now for later:

Show Chimera line command with the following menu cascade:

  • Favorites > Command Line

This will open a Command: strip at the bottom of the graphical window where line commands can be typed or pasted in later.

Homology modeling of GADD45 beta.

Target sequence to model

Fetch the protein sequence from Uniprot with Chimera.

Use the following menu cascade to fetch the protein sequence to model (the target.)

  • File > Fetch by ID...
  • Click on Uniprot at the bottom of the new window and enter the sequence of the target, the human GADD45 beta UniProt ID: GA45B_HUMAN

  • Click Fetch at the bottom

  • Note: clicking on Web Page would open the Uniprot page on a your default Internet Browser window (usually Safari on a Mac.) This is a useful way to verify that the sequence is the one you really wanted.

Two new window will open:

  • The sequence in one window with the protein name
  • A “Region Browser” window with properties of the sequence. The S or Show column checkboxes in the Region Browser can be used to show feature annotations as colored boxes within the sequence window. This window can be closed with the Close button. It can be accessed at any time from the sequence window Info menu.
Window with target sequence.


“Window with ‘target’ sequence.”

Template selection

Select template..

In the Blast results dialog:

  • Click the Columns button to reveal several checkboxes for controlling which columns of information are shown.
  • Hide (uncheck) Description
  • Show Resolution
  • Show Chain names.
Change columns shown on BLAST results.


“Change columns shown on BLAST results.”

Some decisions:

  • MODELLER can compute models based on multiple templates but we will use only one for simplicity here.
  • NMR structures contain multiple models which can make things too complicated.
  • Therefore we will just use 3FFM_A (PDB entry 3FFM, chain A).

Align and show template structure

Input template as sequence and structure..

  • In the BLAST results dialog, click to highlight the row for 3FFM_A
  • At the bottom of the window: click Show in MAV (bottom left) to display the query-hit sequence alignment from BLAST in another Multalign Viewer (MAV) window that will open.
MAV alignment window.


“MAV alignment window.”

  • In the BLAST:query results window: highlight 3FFM_A and click Load Structure (bottom left). This will automatically fetch the 3FFM PDB structure and display it within the Chimera graphics window as a ribbon.
  • Click Quit (bottom right) to close the BLAST results window.

To more easily find the N- and C-terminii we can color the ribbon with a rainbow gradient of colors ranging from bue (Nt) to red (Ct) with either of the following methods:

  • Menu: Presets > Interactive 1 (ribbons)
  • Line command: rainbow

Letting the mouse “hover” over the graphical display the name of the amino acid, number and chain will be displayed, for example for the existing Nt: Serine 15 from chain A as shown.

Template structure 3FFM chain A.


“Template structure 3FFM chain A.”

Alignment verification

Comparative modeling requires a template structure and a target-template sequence alignment. The sequence alignment is crucial; it determines which residues in the template are used to model which residues in the target, and any inaccuracies in the alignment will result in the use of incorrect constraints during 3D modeling. Regardless of how the sequence alignment was obtained, it should be examined and adjusted as needed before initiating the more computationally intensive 3D modeling calculations.

An alignment was generated above as a by-product of the BLAST similarity search. However, BLAST is meant to identify local similarities quickly rather than to give accurate full-length alignments.

The alignment from BLAST is not necessarily adequate for modeling as it may miss segments. In that case the target-template(s) sequence alignment can be generated in some other ways e.g. with external software on the web or on the command line (not shown here.)

Alignment correction/optimization

We will create a new alignment based on the Needleman-Wunsch global alignment algorithm used to to align protein or nucleotide sequences (Needleman and Wunsch 1970).

This button will invite you to act on Create new target-template alignment.

  • Click Quit to close the sequence alignment from BLAST
  • Locate the original GA45B_HUMAN (target) sequence window.

If you had closed the sequence window: the target sequence can be fetched again as before or using in the following Chimera command line:

Command: open uniprot:GA45B_HUMAN
  • From the sequence window menu choose: Edit... > Add Sequence.

The resulting dialog contains tabs for different ways of obtaining the sequence.

  • Click on the From Structure tab.

  • Verify that the default sequence name is 3FFM (at the top) and at the bottom a pull-down menu should show the association of the sequence and the structure as 3FFM (#0) chain A.

  • Click on OK.

Within the sequence window, the association of sequence and structure is indicated with a colored box around a sequence name derived from a structure, here 3FFM.

Note: PDB files headers contain the complete sequence of a solved protein (SEQRES lines) even if not all amino acids are present in the 3D Structure. Red outline boxes enclose residues that are in the sequence but not in the associated structure. 14 residues are missing at the N-terminus in the structure.

Sequence for structure 3FFM chain A is added. AA not part of the structure are boxed in red (Nt).


“Sequence for structure 3FFM chain A is added. AA not part of the structure are boxed in red (Nt).”

Note: The default color scheme of the sequence alignment is that of ClustalX and can be changed with the menu found on the sequence alignment window: Preferences > Appearance, then at the bottom left in the box named Residue letter coloring - Color scheme: the options are : black, ClustalX, ribbon, Kyte-Doolilttle hydrophobicity. There also more options to change the appearance of the sequence: font size, line length etc.

Coloring the sequence to match the structure ribbon is only one of several approaches for sequence-structure mapping. For example:

  1. highlighting residues in the sequence with the mouse selects them in the structure
  2. selecting residues in the structure highlights them in the sequence (green boxes)
  3. structure helix and strand assignments can be shown on the sequence with menu Structure... > Secondary Structure... > show actual - this will place a yellowish box behind the sequences that are in alpha helix and a greenish box behind those in beta sheet.

Model building

We are now ready to ask MODELLER to build a model for our target sequence based on the structure template.

There are 2 ways to run MODELLER:

  1. natively on your computer if it is installed (it is on the classroom iMacs)
  2. via a web service, which could be slower.

Note: Registered users (free for academics) obtain a KEY that is used upon installation of MODELLER on your computer, or provided to Chimera if the web service is requested.

This button will invite you to act on Run Modeller.

From the sequence window menu:

  • Choose Structure... > Modeller (homology)... to open the Chimera interface to comparative modeling with MODELLER.
  • The target should be set to GA45B_HUMAN (top right)
  • Click 3FFM in the dialog to choose it as the template.
  • Click the Advanced Options button to reveal additional settings. This is where the number of alternate models can be requested. We will keep the default of 5.

We will run MODELLER locally and it is necessary to specify where the software is located: /usr/local/bin/mod9.18. Note that if MODELLER is updated the numbers following mod would be updated as well.

  • Click Run Modeller locally and enter /usr/local/bin/mod9.18
  • Click OK to launch the model building.
  • You can see the progress of calculation at the bottom of the main graphics window reported as percent completion.
Comparative MODELLER interface.


“Comparative MODELLER interface.”

When all models have been generated they will be opened within the Chimera graphics and a Modeller Results window will open with the evaluation scores.

Modeller Results window displays evaluation scores..


“Modeller Results window displays evaluation scores..”

By default the last model is selected and shown within the graphics display and highlighted within the result window. The models can be viewed individually or collectively by choosing rows in the dialog with the mouse (use SHIFT to select contiguous or CTRL to select non contiguous list names.)

The different scores from Modeller use different criteria and will not necessarily agree on which models are best:

  • GA341 - model score derived from statistical potentials; a value > 0.7 generally indicates a reliable model, >95% probability of having the correct fold
  • zDOPE - normalized Discrete Optimized Protein Energy (DOPE), an atomic distance-dependent statistical score; negative values indicate better models.

Remove untemplated terminii

This button will invite you to act on Remove untemplated N- and C-termii.

Note: the GADD45 beta sequence was longer than the template structure at the Nterminal (see above and red-lined boxes). Therefore the model termini extend beyond the template structure and consequently cannot be modeled reliably. Displaying all the models at once shows little conformational variability except in the N terminus (GADD sequence residues 1 to 11). This conclusion is reinforced by the RMSD histogram in the sequence window, where bar heights indicate root-mean-square distances among the atoms of the residues associated with a column.

Although there is also a Chimera interface to MODELLER for untemplated building and refinement, in this tutorial we will simply remove the termini and rescore the models.

Note: The NMR model 2KG4 that we did not use as a template is an NMR structure with 20 NMR model solutions, where it could be noted that the Nt are similarly moving in all directions and not not exhibit specific secondary structure. however they have a more consistent direction. Therefore it is for now best to remove the terminii.

Click into the sequence window, then move the cursor over the residues to see the corresponding structure residue numbers near the bottom of the window. In the comparative models (#1.1-5), Lys-15 is aligned with the first residue in the template structure (Arg-18) and Glu-159 is aligned with the last residue in the template structure.

Delete the termini in the comparative models that extend beyond the template. This will delete 14 aa at Nt and one aa at Ct:

Command: del #1:start-14,160-end

All models are contained within model #1. You can see this by opening the model panel with the main menu: Favorites > Model Panel.

Rescore models

This button will invite you to act on Rescore models.

To rescore the models find the Modeller Results:

  • Choose Fetch Scores > zDOPE and Estimated RMSD/Overlap
  • Rescoring uses a web service provided by the Sali lab at UCSF.
  • NOTE: THIS IS A CASE WHERE YOU NEED TO ENTER THE KEY LICENSE
Modeller Results window displays evaluation scores..


“Modeller Results window displays evaluation scores..”

After a minute or few, more favorable zDOPE values are obtained, along with the additional scores:

  • Estimated RMSD - TSVMod-predicted C-alpha root-mean-square deviation (RMSD) of the model from the native structure
  • Estimated Overlap (3.5 Å) - TSVMod-predicted native overlap (3.5 Å), fraction of C-alpha atoms in the model within 3.5 Å of the corresponding atoms in the native structure after rigid-body superposition

Model analysis in Chimera

The comparative models are calculated at the atomic level and can be further analysed and displayed within Chimera as any structure downloaded from the PDB.

Here is an example of showing the hydrophobicity of the model. A similar display with molecular surface could be had via the main menu Presets > Interactive 3 (hydrophobicity surface) but the following method uses the line command:

Command: ~modeldisp #0 
Command: disp 
Command: ~ribbon 
Command: rangecol kdHydrophobicity min dodger blue mid white max orange red 
Command: preset apply pub 1 
Command: repr sphere

Commands explanations: ~ creates the opposite effect of any command e.g. ~disp UN-display

  • ~modeldisp #0: UN-display model 0
  • disp: display stick models for all amino acids
  • ~ribbon: UN-display the ribbon for all models.
  • rangecol...: command to create a color range (blue white orange) based on hydrophobicity property
  • preset apply pub 1: equivallent to mouse menu Presets > Publication 1 (silhouette, rounded ribbon)
  • repr sphere: show atoms as sphere. Equivallent to mouse menu Actions > Atoms/Bonds > sphere

The coloring and representation will apply equally to models 1-5 as part of model 1 (#1) which can be selected to view from the Modeller Results panel.

Homology modeling of naja toxin

Now that we know the steps to create a model and IF there is additional time during the workshop we could explore more options based on a tutorial by Jean-Didier Maréchal3 “Tutorial 1.3 - Homology modeling.”

The gists of the tutorial is very similar to the exercise we have just done and is reproduced almost verbatim with minimal editing. No TASK button is added, just follow the flow of instructions.

Learning objectives:

The objective of this tutorial is to generate a 3D model of a snake toxin (naja snake) using the homology modeling program Modeller.

  1. Generate 3D models using a comparative modeling software 2.Appreciate the quality of the models produced in function of the sequence similarity between the target and the template
  2. See how missing elements in the sequence alignment influence the resulting models
  3. Run tools for the validation of the structural quality of the results

Target sequence

Pir file of the sequence of naja toxin

Provided on the web site:

>P1;3176829
3176829
LKCNKLVPLFYKTCPAGKNLCYKMYMVAMPKVPVKRGCIDVCPKSSLLVKYVCCNTDRCN*

Alternate FASTA format contains 2 more aa at the Nt:

>AAC61316.1 cardiotoxin 4B precursor, partial [Naja sputatrix]
YTLKCNKLVPLFYKTCPAGKNLCYKMYMVAMPKVPVKRGCIDVCPKSSLLVKYVCCNTDRCN

Generate alignment

After template(s) have been determined we first have to generate the alignment between the target and the template(s). The alignment could be done with many programs, but here we’ll use the UCSF Chimera alignment interface

  • Select Select 2CRT_A on the BLAST results
  • Click Show in MAV button, at the left bottom of the window.
  • The MultiAlgin Viewer window now appears.
  • Look at the alignment.
  • Discuss its quality.

Note: The alignment will show a consensus and just a few differences can be found

query       LKCNKLVPLFYKTCPAGKNLCYKMYMVAMPKVPVKRGCIDVCPKSSLLVK 50
2CRT_A      LKCNKLVPLFYKTCPAGKNLCYKMFMVATPKVPVKRGCIDVCPKSSLLVK 50
            ************************:*** *********************

query       YVCCNTDRCN 60
2CRT_A      YVCCNTDRCN 60
            **********

Note: Due to the high identity of sequences and lack of gap(s), it seems not necessary to perform a better alignment as we did on the previous exercise.

Generate model

Now, we can perform the modeling.

  • On the MAV window, select Structure >Modeller(Homology)`.
  • This will open the Comparative Modeling with Modeller window.
  • On the top, Choose the target should be already set to query. Since we pasted the sequence earlier it is simply called query.
  • Choose at least one template: Click on 2CRT_A within the list.

Note that the identity level is extremely high at 96.7%

  • Choose Run Modeller locally and enter the location of the program at /usr/local/bin/mod9.18 (likely already there from the previous exercise.)
  • Click OK
  • Calculation progress appear as % progress at the bottom of the graphical window and should be very quick.
  • When MODELLER is finished a results window with scores appear with 5 models.

Advanced modes

The tutorial continues with more advanced options that we may not have time to review in class. They are below for complete records.

Alternate templates

Repeat the same process and the same discussions with pdb 3PLC and 1JGK.

Notes: - 3PLC contains three identical chains. - 1JGK is an NMR structure with 19 versions of the structure.

Suggestion: rather than use one template at a time try to opt for creating a model with multiple templates, perhaps using some others than those suggested here.

Advanced options

In the window from **Structure >Modeller(Homology)** click onAdvanced Options` at the bottom.

  • Ask a calculation performing Build models with hydrogens.
  • Appreciate the time required for this option regarding the former one.
  • Analyse the differences betweens models generated with and without hydrogen atoms in the force field.

Suggestion: if you have time you can try the option Use thorough optimization: (recommended with MDA)

MDA: Multidomain Assembler (Hertig et al. 2015)

Applying restraints

MODELLER works by a procedure called of satisfaction of spacial restraints. Sometime the user can benefit from several experimental or common knowledge to guide certain restrictions. To do so, MODELLER offers a while panel of user driven restraint.

In Chimera you can provide your own constraints in a file that you can import into the interface. The file should have the format like:

23  89  30  0.1
435-677 901-989 50  0.1
55  443-502 78  0.1

corresponding to:

res1  res2  dist  stdev

where res1 and res2 are residue numbers or ranges of residue numbers in the target sequence, dist is the distance in Å, and stdev is the standard deviation.

If a single residue is specified, its C-alpha will be used to anchor the restraint. If a residue range (e.g. 233-275) is specified, the center of mass of the range will be used to anchor the restraint.

With the best model you obtained, try a constraint between atoms in two successive loops.

Analysis

You can pursue with analytical tools as provided in the slides on the quality of the model.

Loop Modelling

The final part of our modelling, will be dedicated to optimize the structural quality of the loops. We will start with a X-ray structure which has been partially derived.

  • Open the 3PLC structure.
  • From command line: open 3PLC
  • Select two of the units and remove them.
  • From command line (after the selection is done) : del sel
  • Then focus on the final one.
  • From command line: focus

You will see that there is a missing part of the molecule (from 32 to 36). We will try to generate a model which completes this part of the molecule.

Loop modeling interface can be reached in the MAV window or from the main UCSF Chimera menu. In the laster case, go to Tools > Structure Editing > Model/Loop refinment`.

New windows pop up: one with the sequence, the other with the loop modelling interface.

The sequence window already shows you the missing part of the pdb file (in red blocks). This will be one of the priority of the work.

The Loop modelling window gives you all the options for performing the simulation and works quite similarly to the pure MODELLER application.

Discuss with the teacher the different options for the modeling and test the quality of the results.

Model validations (general info)

The quality of the model will depend on the quality of the alignment and that of the template(s) used amongst many other variables.

The validity and plausibility of the model can be assessed with various software shown below.

Quoting from: https://www.researchgate.net/post/How_to_validate_our_homology_model [accessed Apr 11, 2017].

You can use:

  • PROSA, ANOLEA, VERIFY3D, RAMPAGE, ERRAT, SWISSPROT Tools.

If you want to check/know whether the model predicted is secondary structure you can use GOR V, JPRED, SOPMA.

Note: UCLA has a functional web site with many tools cited above http://services.mbi.ucla.edu/SAVES/ with additional: WHAT_CHECK

Additional suggestions:

PROSA2 https://prosa.services.came.sbg.ac.at/prosa.php : compare the global energy profile of model to energy profiles of a non redundant set of good quality models.

PBDsum will generate a full set of PDBsum structural analyses http://www.ebi.ac.uk/thornton-srv/databases/pdbsum/Generate.html

Further resources

Course

Alternate software

Brief table summary:

Protein structure prediction software
Software Web link Notes
CPHmodel http://www.cbs.dtu.dk/services/CPHmodels/ use of single template
MODELLER http://www.salilab.org/modeller/ modeling by satisfaction of spatial restraints
SWISS-MODEL http://swissmodel.expasy.org/ fully automated protein structure homology-modeling server
Phyre2 http://www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index Protein Homology/AnalogY Recognition Engine - pronounced as ‘fire’ - web-based services
HHpred http://toolkit.tuebingen.mpg.de/hhpred more sensitive in finding remote homologs
LOMATES http://zhanglab.ccmb.med.umich.edu/LOMETS/ Local Meta-Threading-Server - web service collecting high-scoring target-to-template alignments from 8 locally-installed threading programs (FUGUE, HHsearch, MUSTER, PPA, PROSPECT2, SAM-T02, SPARKS, SP3).
MODBASE http://salilab.org/modbase database of annotated comparative protein structure models.
Robetta http://www.robetta.org/ Robetta provides both ab initio and comparative models of protein domains.
chunk-TASSER http://cssb.biology.gatech.edu/skolnick/webservice/chunk-TASSER/index.html combines threading templates from SP3 and ab initio folded chunk structures (three consecutive segments of regular secondary structures). It is better for extreme hard targets
PSiFR http://psifr.cssb.biology.gatech.edu/ Protein Structure and Function predicton Resource: integrated tools for protein tertiary structure prediction and structure and sequence-based function annotation. Various methods used : TASSER (Threading/ASSembly/Refinement), TASSER-Lite, PROSPECTOR_3, METATASSER, SPARKS2
ESyPred3D http://www.fundp.ac.be/sciences/biologie/urbm/bioinfo/esypred/ increased alignment performances - 3D model made by MODELLER
PMP http://www.proteinmodelportal.org/PMP Protein Model Portal access to different partner sites for model building, and quality assessment.
ProModel http://www.vlifesciences.com/products/VLifeMDS/Protein_Modeller.php complete package for modeling proteins for Linux and Windows
SCWRL4 http://dunbrack.fccc.edu/scwrl4/index.php use of rotamer library.
VADAR http://vadar.wishartlab.com/ Volume, Area, Dihedral Angle Reporter is a compilation of more than 15 different algorithms and programs
IntFOLD http://www.reading.ac.uk/bioinf/IntFOLD/ 3D modeling, quality assessment, […] Prediction of protein-ligand binding residues.
PEPstr http://www.imtech.res.in/raghava/pepstr/ server predicts the tertiary structure of small peptides with sequence length varying between 7 to 25 residues.
BSR http://cssb.biology.gatech.edu/BSR Binding Site Refinement: template-based method for the local refinement of ligand-binding regions in protein models
KeyRecep http://www.immd.co.jp/en/product_2.html rational molecular design when the 3D structure of the target protein is unknown. […] estimate the characteristics of the binding site
PROTEUS2 http://wks16338.biology.ualberta.ca/proteus2/ web server, average prediction takes ~2 minutes per query sequence. able to achieve among the highest reported levels of predictive accuracy for signal peptides (Q2=94%), membrane spanning helices (Q2=87%) and secondary structure (Q3 score of 81.3% ).
PSIPRED http://bioinf.cs.ucl.ac.uk/psipred/ simple and accurate secondary structure prediction method
I-TASSER http://zhanglab.ccmb.med.umich.edu/I-TASSER/ Web server - ranked as the No 1 server for protein structure prediction in recent CASP7, CASP8 and CASP9 experiments.
JPred http://www.compbio.dundee.ac.uk/www-jpred/ Protein Secondary Structure Prediction server

Online tutorials:

Acknowledgments

This tutorial is based on the following online resources:

  1. homology modelling with Chimera https://github.com/allegravia/homology_modelling by Allegra Via and Domenico Raimondo.

  2. Tutorial 1.3 - Homology modeling http://gent.uab.cat/jdidier/content/tutorial-13 by Jean-Didier Maréchal

  3. Chimera tutorial: Comparative Modeling Tutorial https://www.cgl.ucsf.edu/chimera/docs/UsersGuide/tutorials/dor.html

  4. Video demonstration: Homology modeling with Modeller https://www.cgl.ucsf.edu/chimera/videodoc/Modeller/

Note: Larger chunks of mostly undedited text from the original tutorial will be printed in italics

REFERENCES

Fiser, A., R. K. Do, and A. Sali. 2000. “Modeling of loops in protein structures.” Protein Sci. 9 (9): 1753–73.

Hertig, S., T. D. Goddard, G. T. Johnson, and T. E. Ferrin. 2015. “Multidomain Assembler (MDA) Generates Models of Large Multidomain Proteins.” Biophys. J. 108 (9): 2097–2102. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4423039/.

Marti-Renom, M. A., A. C. Stuart, A. Fiser, R. Sanchez, F. Melo, and A. Sali. 2000. “Comparative protein structure modeling of genes and genomes.” Annu Rev Biophys Biomol Struct 29: 291–325.

Needleman, S. B., and C. D. Wunsch. 1970. “A general method applicable to the search for similarities in the amino acid sequence of two proteins.” J. Mol. Biol. 48 (3): 443–53. https://www.ncbi.nlm.nih.gov/pubmed/5420325.

Pettersen, E. F., T. D. Goddard, C. C. Huang, G. S. Couch, D. M. Greenblatt, E. C. Meng, and T. E. Ferrin. 2004. “UCSF Chimera–a visualization system for exploratory research and analysis.” J Comput Chem 25 (13): 1605–12. http://www.cgl.ucsf.edu/chimera.

Sali, A., and T. L. Blundell. 1993. “Comparative protein modelling by satisfaction of spatial restraints.” J. Mol. Biol. 234 (3): 779–815.

Webb, B., and A. Sali. 2016. “Comparative Protein Structure Modeling Using MODELLER.” Curr Protoc Bioinformatics 54 (June): 1–5.

  1. https://salilab.org/modeller/↩

  2. http://www.cgl.ucsf.edu/chimera↩

  3. http://gent.uab.cat/jdidier/content/tutorial-13↩