Exercise:AntibodyStructurePrediction

Antibody Structure Modelling: From Sequence to 3D Structure

Duration: 1–2 hours

Requirements: PyMOL and a web browser.

Introduction

Antibodies recognize antigens through six complementarity-determining regions (CDRs): three in the heavy chain and three in the light chain. These loops are supported by more conserved framework (FW) regions.

Modern deep-learning tools can predict antibody structures directly from amino acid sequences. However, not all regions are predicted with equal accuracy, and a predicted structure does not necessarily reveal what an antibody binds.

In this exercise, we will predict and explore the three-dimensional structure of an antibody from a patient with chronic lymphocytic leukemia (CLL).

Learning objectives:

• Identify the framework and CDR regions of an antibody.
• Generate an antibody structure using a modern prediction tool.
• Visualize and interpret the antigen-binding site.
• Understand prediction confidence and limitations.
• Explore how amino acid properties influence antigen recognition.

Antibody sequences

We will use the variable regions of the SI135 antibody.

Heavy chain (VH):

>SI135_VH
EVQLVESGGGLVRPGGSLRLSCAASGFTFSTYSMNWVRQAPGKGLEWVSSISSSSNSIYCADSVKGRFTISRDNANNSLYMQMNSLRAEDTAVYYCARDVNTMDVWGQGILVTVLL

Light chain (VL):

>SI135_VL
SYVVTQPPSVSVAPGKTARITCGGNNIGSKSVHWYQQKPGQAPVLVIYYDNDRPSGIPERFSGSNSGNTATLTISRVEAGDEADYYCQVWDSGSDHPWVFGGGTKLTVL

These sequences correspond to the variable domains, not the complete antibody.

Part 1 — Identify the antibody regions (15 minutes)

Go to the ANARCII website.

ANARCII identifies antibody and T-cell receptor sequences and assigns standardized residue numbering.

1. Paste the light-chain protein sequence.
2. Select Antibody and the IMGT numbering scheme.
3. Submit the sequence and inspect the result.

Q1. Is the light chain kappa (κ) or lambda (λ)? Which light-chain type is more common in humans?

Part 2 — Predict the antibody structure (15–20 minutes)

Go to the ABodyBuilder2 website.

ABodyBuilder2 is a deep-learning-based tool that predicts the three-dimensional structure of an antibody from its amino acid sequence.

1. Paste the heavy-chain and light-chain sequences in the corresponding fields.
2. Select the IMGT numbering scheme.
3. Submit the prediction.
4. Explore the predicted structure and the prediction-error estimates.
5. Download the predicted structure as a PDB file.

Q2. Which regions have the highest predicted error? Are they framework regions or CDRs? Why might some regions of an antibody be harder to predict than others?

Part 3 — Explore the structure in PyMOL (25 minutes)

Download the PDB file generated by ABodyBuilder2 and open it in PyMOL.

Display the antibody as a cartoon and color the heavy and light chains differently.

You can use the graphical interface or type the following commands in PyMOL:

hide everything
show cartoon
color marine, chain H
color wheat, chain L

Note: These commands assume the chains are named H and L. Check the actual chain identifiers in your PDB file.

Identify the CDRs

Each antibody variable domain contains three CDR loops. Together, the heavy and light chains form the antigen-binding site.

Using IMGT numbering, the CDR regions correspond to:

Region	IMGT positions
CDR1	27–38
CDR2	56–65
CDR3	105–117

The exact number of residues in each loop varies between antibodies, and some IMGT positions may be absent.

Select the CDRs and color them red. If the PDB uses IMGT numbering and the chains are named H and L, you can use:

select cdrs, chain H+L and resi 27-38+56-65+105-117
color red, cdrs

If the file uses different residue numbering, use the numbering provided by the prediction tool to identify the corresponding regions.

Rotate the antibody and examine the six CDR loops.

Q3. Which CDR appears to protrude the most from the antibody surface? Is it also the longest loop?

Explore the antigen-binding site

Display the antibody as a molecular surface using PyMOL.

show surface
set transparency, 0.3

Look at the surface formed by the six CDR loops.

Q4. How would you describe the general shape of the antigen-binding site: flat, pocket, mountain, or groove? Explain briefly.

Part 4 — Explore the electrostatic properties (15–20 minutes)

The amino acid composition of the CDRs influences their interactions with antigens.

In PyMOL, generate an approximate electrostatic representation using:

Action → Generate → Vacuum Electrostatics → Protein Contact Potential

Inspect the distribution of positively and negatively charged regions around the antigen-binding site.

Remember:

• K (lysine) and R (arginine) are positively charged.
• D (aspartate) and E (glutamate) are negatively charged.
• Other amino acids can contribute to antigen recognition through hydrogen bonds, hydrophobic interactions, and shape complementarity.

Q5. Identify one or two charged residues in the CDRs that contribute to the electrostatic properties of the antigen-binding site. Where are they located?

Note: The vacuum electrostatic representation provides a qualitative indication of the surface charge distribution. It is not a rigorous calculation of binding energy.

Part 5 — How much can we trust the prediction? (10 minutes)

Return to the ABodyBuilder2 results and examine the reported prediction errors.

The framework of an antibody is usually more conserved than its CDR loops. In particular, CDR-H3 can be highly variable in both sequence and conformation.

Modern prediction tools can generate realistic antibody structures, but uncertainty remains, especially in flexible or highly variable regions.

Q6. Can we confidently determine which antigen this antibody recognizes from its predicted structure alone? Explain what we can learn from the model and what additional information would be needed.

Optional exercises

A. Compare traditional and modern antibody modelling

Go to the LYRA website.

LYRA predicts antibody structures using templates from experimentally determined structures.

1. Submit the same VH and VL sequences.
2. Examine the selected structural templates.
3. Download the predicted PDB structure.
4. Open the LYRA and ABodyBuilder2 models in PyMOL.
5. Superimpose the two structures and compare them.

Optional Q1. Which regions show the largest structural differences between the two models? Do they agree on the overall fold?

B. Modelling T-cell receptors

Antibodies are not the only immune receptors that can be predicted using modern deep-learning approaches.

Go to the TCRBuilder2+ website.

Explore an example structure prediction of a T-cell receptor (TCR).

Optional Q2. What structural similarities can you identify between antibody and TCR variable domains? How do their biological recognition functions differ?

Submission

Submit:

• One PyMOL screenshot showing the heavy and light chains with the CDRs highlighted.
• Short answers to Q1–Q6 (approximately 1–3 sentences per question).

Finished!

Antibody modelling

Requirements: Pymol

Download the version for your OS, then use the license file provided by the teachers

Modeling of CLL antibody

These are the sequences for the light and heavy chain of the antibody:

   >SI135 VH

   GAGGTGCAGCTGGTGGAGTCTGGGGGAGGCCTGGTCAGGCCTGGGGGGTCCCTGAGACTCTCCTGTGCAGCCTCTGGATTCACCTTCAGTACCTATAGCATGAACTGGGTCCGCCAGGCTCCAGGGAAGGGGCTGGAGTGGGTCTCATCCATTAGTAGTAGTAGTAATTCCATATACTGCGCAGACTCAGTGAAGGGCCGATTCACCATCTCCAGAGACAACGCCAACAACTCACTGTATATGCAAATGAATAGCCTGAGAGCCGAGGACACGGCTGTCTATTACTGTGCGAGAGATGTTAACACTATGGACGTCTGGGGCCAAGGGATCCTGGTCACCGTNCTCCTCA

   >SI135 VL

   TCCTATGTGGTGACTCAGCCACCCTCAGTGTCAGTGGCCCCAGGAAAGACGGCCAGGATTACCTGTGGGGGAAACAACATTGGAAGTAAAAGTGTGCACTGGTACCAGCAGAAGCCAGGCCAGGCCCCTGTGCTGGTCATCTATTATGATAACGACCGGCCCTCAGGGATCCCTGAGCGATTCTCTGGCTCCAACTCTGGGAACACGGCCACCCTGACCATCAGCAGGGTCGAAGCCGGGGATGAGGCCGACTATTACTGTCAGGTGTGGGATAGTGGTAGTGATCATCCTTGGGTGTTCGGCGGAGGGACCAAGCTGACCGTCCTA

Translate the nucleic acid sequences to protein sequences (use a service of your choice, e.g. expasy or virtual ribosome).

Go to the Lyra web page.

Lyra is a tool for predicting antibody and TCR structures. It is possible to use Lyra in automated or expert mode.

Insert the light and heavy chains in the proper text area and submit. After a few seconds you will arrive to the output page, where the alignment and all the templates are displayed.

Q1: what is the sequence similarity of the best H and L templates for the framework (FW)?

hint: In the output page you have a few tabs - check the PDB file, Log, and manual selection ones to find this info.

Q2: What is the light chain type? Is it the most common in humans?

Focus on the sequence alignment used to generate the model. You can find it in one of the tabs in Lyra website, as well as in the PDB file itself

Q3: where are the differences between target and template located? Which are the most drastic that could affect the modeling process (e.g. G->W)?

hint: if you don't remeber the amino acids, you can use the Blosum62 matrix to estimate the expected effect of a substitution: the lower the number, the larger the expected impact. For sake of time, you only need to give a qualitative answer.

Q4: Do the loops/CDR have the same length in target and template? What does it mean, in terms of expected accuracy?

Now, look at the templates automatically selected by Lyra to build the final model

Q5: How many templates were used? Which is the region for which the template is more dissimilar to the target sequence?

Download the pdb file and open it with pymol. There is a small "S" bottom right, it shows/hides the sequence. Show the whole protein as cartoon and color by chain. Select and color in red all the CDRs. (the selected residues appear as the "sele" item in the menu). The CDR definition can be seen in the alignment generated by Lyra.

Q6: What is the general shape of the ABS: flat, a pocket, a mountain or a groove? It can help to show the antibody as surface

Q7: Which loop is the longest and protruding the most? Is this the most common case?

Use the Domain Gap Align tool identify the somatic mutations in the V and J regions. You need to use the amino acid sequence and select human as the source organism. In PyMol, show the mutations as sticks (main chain and side chain) and color them green

Q8: Which Mutations are located inside/near the CDRs? Which are the ones that are most likely to have an impact on the binding of the antibody? A qualitative answer is sufficient

Generate the electrostatic potentials of the Ig ( Action -> generate -> vacuum electrostatic -> protein contact potentials).

Q9: Identify the residues in the CDRs that mostly contribute to the charge distribution. highlight them in the sequence and look for charged residues such as K, R, D, E

Q10: (optional) Mutate some residues in order to have a positive charge in the middle of the antigen binding site, then recalculate the electrostatic potentials. Can you see the changes?

Q11: Go to the AbodyBuilder 2 site and make a new structure prediction. Download the PDB file and compare with the model generated by LYRA. Can you see some differences?

Finished!