IEDB Analysis Resource - Labs Logo

Lymphocyte Receptor Automated Modelling (LYRA) - Tutorial

The LYRA server predicts structures for either T-Cell Receptors (TCR) or B-Cell Receptors (BCR) using homology modelling. Framework templates are selected based on BLOSUM score, and complementary determining regions (CDR) are then selected if needed based on a canonical structure model and grafted onto the framework templates. The final output is a complete three-dimensional model of the target antibody that can be downloaded or displayed on-line.
How to obtain predictions
1. Specify sequences
First, input a pair of TCR or BCR protein sequences in any order. The sequences should either be entered directly into the text area field labeled "Enter protein sequence(s)”, or can be taken from a file that has to be uploaded using the "Browse" button. The sequences can be supplied in two different formats: The input sequence will be interpreted as FASTA if an opening ">" character is found, or as a continuous sequence otherwise.
All sequences have to be amino acids specified in single letter code
(ACDEFGHIKLMNPQRSTVWY)
• Format for the upload sequence file:
The file should contain heavy and light chains (or alpha and beta chains) in fasta format.
Example:
>light chain header
sequence for light chain here...

>heavy chain header
sequence for heavy chain here...
2. Advanced options:
Advanced options consist of options to select different side chain modeling methods. It also provides option to blacklisting any template(s) for modeling.
• Side Chain Modeling Method:
Different strategies can be selected to model the side chain of the output model.

HMMER:
HMMER is used for searching sequence databases for sequence homologs, and for making sequence alignments.

HMMER + SCWRL 4.0:
HMMER is used for searching sequence databases for sequence homologs, and for making sequence alignments. Non conserved residues side chains are modeled using SCWRL 4.0. SCWRL uses conserved residues side chains as constraints when modeling remaining side chains. This is the default choice.

• Blacklisted PDB:
It should be a comma-separated list of PDB ids (e.g. 1DEE,8FAB,1fve) to explicitly prevent lyra to use them as framework or loops template for both chains.
3. Submit the prediction
This one is easy. Click the submit button, and a result screen similar to the one below should appear.
Interpreting prediction output

The Summary table shows template structures and calculated canonical structures for both chains.

Canonical Structures(CS)
Shown below is the definitions of the canonical structures for TCR chains. Structures are predicted using the affinity propagation algoritm, and the exemplars are shown as being the representative PDB structure.

LoopLoop LengthCanonical StructureRepresentative PDB
AnyAny0Unable to determine CS
A1812VLR
923QFJ
33TJH
43ARB
54G8E
1064OZH
74JRX
A2513QIU
24OZG
634L3E
744P23
53MV8
63REV
73SCM
884G9F
A3913SKN
1023QEQ
1134L4V
44P46
1253PL6
63QFJ
1373TF7
1483HUJ
93MV8
102BNQ
15114P4K
121MI5
LoopLoop LengthCanonical StructureRepresentative PDB
AnyAny0Unable to determine CS
B1913QEU
24MAY
1033O6F
44JFF
B216None
1714JFH
1822IAL
33MV8
1944MAY
53QIU
61KB5
2073O6F
B3Any0B3 loops are scored/grafted solely on BLOSUM62 score.

BCR Canonical structures are predicted using sequence-based rules. The rules are described in the following table:

LoopLoop LengthCanonical StructureConstraints
AnyAny0Unable to determine CS
H164
71
82
93
H23171:AVL
271:RK
4371:AVL
471:RK
6571:AVL
671:RK
H331
4394:ACDEFGHILMNPQSTVYW
494:RK
5694:ACDEFGHILMNPQSTVYW
794:RK
6994:ACDEFGHILMNPQSTVYW
1094:RK
71294:ACDEFGHILMNPQSTVYW
1394:RK
81594:ACDEFGHILMNPQSTVYW
1694:RK
91894:ACDEFGHILMNPQSTVYW
1994:RK
102194:ACDEFGHILMNPQSTVYW
2294:RK
112494:ACDEFGHILMNPQSTVYW
2594:RK
122794:ACDEFGHILMNPQSTVYW
2894:RK
133094:ACDEFGHILMNPQSTVYW
3194:RK
143394:ACDEFGHILMNPQSTVYW
3494:RK
153694:ACDEFGHILMNPQSTVYW
3794:RK
163994:ACDEFGHILMNPQSTVYW
4094:RK
174294:ACDEFGHILMNPQSTVYW
4394:RK
184594:ACDEFGHILMNPQSTVYW
4694:RK
194894:ACDEFGHILMNPQSTVYW
4994:RK
205194:ACDEFGHILMNPQSTVYW
5294:RK
LoopLoop LengthCanonical StructureConstraints
AnyAny0Unable to determine CS
K16129:VIL
7229:VIL
8629:VIL
11529:VIL
12429:VIL
13329:VIL
K231
K34490:Q
5396:P 90:Q
690:Q 94:L
6195:P 90:HNQ
290:Q 94:P
7595A:P 90:Q
LoopLoop LengthCanonical StructureConstraints
AnyAny0Unable to determine CS
L17728:VIL 66:N
88
9125:G
525:R 28:G
10225:G 66:K 90:S 31:FHY
366:L 90:L
625:G 66:K 90:S 31:ND
L231
L361
7392:D 95:ST
495:CVPILDNHFYWMEQKRAG
82

Alignment: The sequences are aligned with the input sequence, the template sequence and the variable loops.


Modeled Structure is displayed using JSmol Viewer (JavaScript-Based Molecular Viewer From Jmol).
It allows the user to display or hide both chains and to highlight each loop.


Click on Download button to download the model generated by LYRA.

The downloaded model file can be visualized locally using Jmol viewer. Jmol is a free, open source molecule viewer that is cross-platform. Instructions on how to download, install and run the viewer can be found here.

The original developer of LYRA tool is Paolo Marcatili.