SAbDab Help





Dunbar, J., Krawczyk, K. et al (2014). Nucleic Acids Res. 42. D1140-D1146

What is:

  • an antibody? - an immunoprotein responsible for specifically recognising and binding to potentially pathogenic molecules.
  • an antigen? - the molecule that the antibody targets.
  • a heavy chain? - the longer antibody chain. This folds to form one variable domain (VH) and three or more constant domains (CH1, CH2 and CH3).
  • a light chain? - the shorter antibody chain. This folds to form one variable domain (VL) and one constant domain (CL1).
  • a variable domain (VH or VL)? - the variable part of the antibody.
  • a CDR? - a Complimentarity Determining Region. These are generally characterised by six hypervariable loops: three on VH (H1, H2 and H3) and three on VL (L1, L2 and L3). Diversity in the sequence and structure of the CDRs are the main determinants of antigen specificity and affinity.
  • the framework region? - the collective name for the residues in VH or VL that are not CDRs.
  • an Fv? - a variable fragment. The collective name for a paired VH and VL.
  • an Fab?- an antigen binding fragment. The same as an Fv but also including the first constant domains of each chain, CH1 and CL1. Typically an antibody will have two Fab arms (see above). Most structures in SAbDab contain only these fragments.
  • a numbering scheme? - a system to annotate equivalent positions in antibodies. A scheme can often be applied by examining the sequence of the antibody only.


How do I:

Inspect a particular structure?
  • Go to the Antibody Search page and click on "Get Particular pdb".
  • Enter the four digit pdb code of the antibody structure you are interested in (e.g. 12e8).
  • A results table will be returned. Click on the pdb code in order to open the summary page for the structure.
Download data for a particular structure?
  • Go to the summary page for the structure as described above. (Example)
  • Click on the "Downloads" tab. Three files will be available:
    • the structure in PDB format as deposited. (Example)
    • the structure in PDB format with the antibody chains numbered using Chothia numbering. (Example)
    • a csv (values separated by tabs) summary file containing annotations pertaining to the structure. (Example)
  • Simply click on each link to download the file. See the "Formats, Data and Downloads" section below for more information about file formats.
List all the antibodies structures in SAbDab?
  • Go to the Antibody Search page and press the "Get all" button. A table with all the current entries will be displayed.
Create a dataset?
  • Go to the Antibody Search page and click on Advanced search field.
  • Select from a number of properties including:
    • Experimental method (e.g. X-Ray, NMR etc)
    • Bound state and, if bound, antigen type (e.g. protein, hapten etc)
    • Antibody species
  • A list of structures that satisfy the conditions will be returned.
Get the dataset of antibody-antigen complexes with curated affinity data?
  • Here you go!
  • We hope this to serve as an antibody-antigen docking benchmark resource.
  • Narrow your search (e.g constrain by antigen type) using the Antibody Search.
Make a non-redundant dataset?
  • Go to the Antibody Search page and click on the Non-redundant Search tab.
  • Select the properties that you wish the structures to have as above.
  • Select the sequence-identity threshold for at which antibody sequences should be clustered at (variable domains only).
  • If you are considering bound structures, select a sequence identity threshold at which to cluster antigen sequences.
  • Note: this may a particularly useful tool for studying antibody interactions and docking protocols. (e.g. 6% of protein antigens in SAbDab are lysozyme!)
  • A non-redundant list of structures will be returned upon submission.
Download a dataset?
  • Select a dataset as described above. You will be presented with a list of structures.
  • The data for each may be downloaded individually using the links in the leftmost column of the table.
  • The whole dataset may be downloaded as a zip file by clicking on "Download All" at the bottom of the page. This will create an archive file which must be downloaded within 20 minutes of creation.
  • Alternatively, use our download script to download the data at your leisure (Help?).
Identify a template for homology modelling?
  • Go to the Template search page.
  • Paste the sequence of the antibody you wish to find structural templates for in the text boxes. This can be either a heavy chain or a light chain or both.
  • Choose the number of structures that should be returned (between 1 and 100).
  • Click "Search for templates" to return a list of structures with the highest sequence identity to your antibody (variable region only).
  • Download the structures or click on "Align" to view an annotated alignment between your sequence and each template.
  • By default sequence identity is calculated over the variable region of the antibody. Users may also choose different regions to calculate the identity over (e.g framework or CDRs). Clicking on "full coverage" ensures that there are no indels between template and query sequence.

Select a set of CDR structures?

  • Go to the CDR search page and click on "Search CDRs by attribute".
  • Choose which CDR definition to use (Chothia, Kabat or Contact)
  • Choose the type of CDR you wish to select (H1, H2, H3, L1, L2 or L3). Leave as "*" to select all.
  • Choose the length the CDRs that should be returned. Leave as "*" to select all.
  • Choose the other attributes that the structures should have.
  • Click "Search CDRs" to return a list of cdr structures.

Explore the CDR clustering?

  • Go to the CDR clustering page.
  • Choose the type of CDR you wish to view (i.e H1, H2, H3, L1, L2 or L3)
  • Choose which CDR definition to use (Chothia, Kabat or Contact)
  • Choose the length of the CDR that should be clustered (e.g between 1 to 30).
  • Choose the UPGMA cut-off. This is maximum RMSD allowed between any two structures in a cluster (e.g. the smaller the value the tighter the cluster).
  • Click "Search CDRs" to return a list of clusters. Click on each cluster to view the structures contained in it.

Explore the CDR clustering in the context of other clusterings?

  • Go to the CDR clustering page.
  • Choose the 'Mapping to other clustering methods'
  • Choose the type one of the three CDR canonical class definitions you would like to map to (Chothia, Martin & Thornton or North & Dunbrack)
  • Choose which CDR definition to use (Chothia, Kabat or Contact)
  • Choose the type of CDR you wish to view (i.e H1, H2, L1, L2 or L3)
  • Choose the UPGMA cut-off. This is maximum RMSD allowed between any two structures in a cluster (e.g. the smaller the value the tighter the cluster).
  • Click "Search CDRs" to return a list of clusters. Click on each cluster to view the structures contained in it.

Available Data

For each entry in SAbDab the following data is available for download:

  1. The structure file in PDB format as deposited to the protein data bank. Example.
  2. Chothia renumbered structure file in PDB format. Example.
    • All chain identifiers are retained.
    • Chain pairings (heavy-light-antigen) are recorded in a REMARK record in the header. e.g. for structure 1ahw: REMARK 5 PAIRED_HL HCHAIN=B LCHAIN=A AGCHAIN=C AGTYPE=PROTEIN
      REMARK 5 PAIRED_HL HCHAIN=E LCHAIN=D AGCHAIN=F AGTYPE=PROTEIN
    • Variable region of the chains are chothia numbered. e.g CA atom at chothia position H82A on chain B in structure 1ahw: ATOM 3877 CA SER B 82A -18.113 15.679 27.979 1.00 6.53 C
    • Residues outside this region are numbered sequentially.
    • "Non-antibody" chains retain their original numbering.
  3. A summary file in csv format (values separated by tabs). Example.
    • The first row of the file is a header containing the name of each field.
    • Each subsequent row corresponds to a heavy-light chain pairing and associated annotations. e.g. for structure 1ahw the first 6 fields are:
      pdb Hchain Lchain model antigen_chain antigen_type ...
      1ahw B A 0 C protein ...
      1ahw E D 0 F protein ...
    • To view in excel or open office, open the file and when prompted choose "separated by Tab".

Download methods

When any dataset (list of entries) is selected in SAbDab there is an option to download the data using two methods:

  1. Download an automatically generated zip file (preferred).
    • At the bottom of the page click the "Download all" button.
    • A zip file will be created containing the files described above for each selected structure
    • Download the .zip file and extract the archive.
    • This file will be available for 20 minutes after creation.
  2. Download using the SabDab download script.
    • Download the summary file for your selection.
    • From a unix command-line run the script, specifying the data that you wish to retrieve. e.g. Download the original structure and the imgt annotations for the structures in the summary file:
      $ python sabdab_downloader -s summary_file.csv -o path/to/output/ --original_pdb --imgt
    • This will create a folder called "sabdab_dataset" in the folder "path/to/output/". Within that will be a directory for each entry in the summary file containing the requested data.
    • Functionality has only been tested in linux. Please use the zip file approach for all other operating systems.

Summary file fields

A summary file is created when a dataset is selected in SAbDab and is available for each structure individually. Each row corresponds to a heavy-light chain pairing in a PDB structure. Each pairing is annotated with the following fields.

PDBThe PDB accession code (e.g. 12e8)
HchainThe chain identifier for the heavy chain (e.g. "H"). This is "NA" if the light chain is unpaired.
LchainThe chain identifier for the light chain (e.g. "L"). This is "NA" if the heavy chain is unpaired.
modelThe model identifier for the pairing (e.g "0","1","2"...). This is "0" for X-Ray structures )
antigen_chainThe chain identifier for the bound antigen chain (e.g. "A"). If the antigen has multiple bound antigen chains, these are separated by a "|" (e.g "X | Y"). For non-polymer antigens this refers to the chain identifier of the corresponding HETATM records (i.e. it may be the same as either the heavy or light chain identifier)
antigen_typeThe classification of the antigen. Either: protein, peptide, carbohydrate, nucleic acid or hapten. This is "NA" if the heavy-light pairing is unbound.
antigen_het_nameThe HETATM of the antigen if it is non-polynmer. This is "NA" if the antigen is a polymer or the heavy-light pairing is unbound.
antigen_nameThe name of the antigen. This is "NA" if heavy-light pairing is unbound or "?" if unknown.
short_headerThe short header of the structure. Typically a short description of the type of molecule in the structure entry.
dateThe deposition date of the structure to the PDB.
compoundThe description of the molecule in structure. Typically the title of the associated publication.
organismThe organsim(s) of the molecule(s) in the structure.
heavy_speciesThe species of the heavy antibody chain. If it is from multiple species (e.g. Chimeric or Humanized) these will be separated by a single comma.
light_speciesThe species of the light antibody chain. If it is from multiple species (e.g. Chimeric or Humanized) these will be separated by a single comma.
antigen_speciesThe species of the antigen chain. If it is from multiple species these will be separated by a single comma.
authorsThe authors of the structure.
resolutionThe resolution of the structure if determined by X-Ray diffraction or Electron Microscopy.
methodThe method with which the structure was determined.
r_freeThe Rfree value of the structure if determined by X-Ray diffraction.
r_factorThe R factor value of the structure if determined by X-Ray diffraction.
scfvWhether the structure is a single chain Fv. True or False. If true, the heavy and light chain identifiers may be the same depending on how the structure has been deposited.
engineeredWhether the structure has been engineered. True or False.
heavy_subclassThe IMGT variable subgroup of the heavy chain. Structures that are not available in the IMGT database have a subgroup assigned by SAbDab.
light_subclassThe IMGT variable subgroup of the light chain. Structures that are not available in the IMGT database have a subgroup assigned by SAbDab.
light_ctypeThe type (Kappa or Lambda) of the light chain.
affinityThe affinity of the antibody to the antigen present in the structure (KD - M).
delta_gThe ΔG of the antibody to the antigen present in the structure (kcal/mol). This has been manually calculated.
affinity_methodThe method by which the affinity data was collected (SPR, ITC or other).
temperatureThe temperature at which the affinity data was collected (°C).
pmidThe pubmed identifier that is the source of the associated affintity data.

The relative orientation between the antibody variable domains (VH and VL) influences the topology of the antigen binding site. It is also important to consider when building high-resolution models of antibodies.

The VH-VL orientation can be characterised using six measures:

HLthe torsion angle between H1 and L1
HC1the bend angle between H1 and C
HC2the bend angle between H2 and C
LC1the bend angle between L1 and C
LC2the bend angle between L2 and C
dcthe length of C

These measures have been calculated for each paired VH-VL structure in SAbDab. The distributions of a non-redundant set taken from the database can be used to visualise the VH-VL orientation space.

Selecting by angle

  • Use the distributions to select structures with particular orientations.
  • Click on a bin to select it. Multiple bins for different measures may be selected at once.
  • If no selection is made for a measure, all bins for that measure are selected by default.
  • Click "Run ABangle" to get a list of structures with the desired orientation.

Selecting by PDB

  • Enter the pdb code of an antibody structure e.g. "1ahw".
  • Click on "Add more pdbs" to compare to other structures.
  • Click on "Run ABangle" to show where each of the selected structures lays in VH-VL orientation space.
  • To select a specific Fv from a pdb, give its heavy and light chain identifiers. e.g. 12e8 contains fvs 12e8_HL and 12e8_PM.

Selecting by similar orientation

  • Enter the Fv identifier of a paired heavy and light chain e.g. "12e8_PM".
  • Select the species that the returned antibodies should come from e.g. "HOMO SAPIENS".
  • Click on "Run ABangle" to return of structures with similar orientations to the selected structure.
  • Here, similar is defined as within one bin on each of the angular measures.
  • Use this tool to find for instance Human antibodies that have a similar orientation to your murine antibody.


For a full description of the method or if you use this software, please refer to:
Dunbar et al. ABangle: characterising the VH–VL orientation in antibodies. PEDS (2013)

  • SAbDab is built and maintained by the Oxford Protein Informatics Group (OPIG)
  • Please see the FAQ and topics above for help using SAbDab
  • For any further queries please use our query form or contact us directly: opig <~at~> stats.ox.ac.uk