AbNGS DB

Large dataset of unpaired heavy and light chains.

• Training language models.

• Data mining for biologically relevant chains.

• Very large dataset.

• Rich metadata annotations.

No target information.

PairedAbNGS DB

Large database of paired heavy and light chains from human and mouse.

• Training paired language models.

• Heavy/light pairing preferences prediction.

• Very large dataset.

• Rich metadata annotations.

No target information.

AbDesign DB

Datasets of 7 antigens with 2 antibodies each, with multiple heterogeneous CDR-H3 mutations.

• Benchmarking antibody redesign methods in structural context.

• Heterogeneous mutational data.

• Uniform experimental methods.

• ELISA measured Ab concentration.

Very precise.

NAStructural DB

Structures from the PDB, sorted into single chains complexes, antibodies, nanobodies, non-antibody proteins processed for analytics & machine learning applications.

• Training structural models.

• Searching for antibodies with a given target.

• Atoms, Residues and contacts based statistics computations.

• Antibody, Nanobody®/VHH, protein structures with their complexes.

• Clean and well-defined data format.

• Deduplicated datasets.

Very precise.

Patents

Antibodies from USPTO, WIPO, DDBJ and PSIPS, associated with their patent families and metadata from EPO.

• Quick way for initial for freedom of operation checks.

• Perform per-family target identification.

• Very good coverage.

• Target information, when available.

To within a protein class, however with buyer beware on unreliability of patented sequence data.

Therapeutics

Database of antibody therapeutics with their associated clinical metadata.

• Develop therapeutic scores such as TAP, LAP.

• Data-driven reference for clinical information on approved biologics therapices.

• Full antibody sequences (including constant regions).

• Post translational modifications.

• Disease/indication information.

• Antigen characterization.

• Clinical information.

Precise.

Literature

Database of antibodies and targets manually curated from publications.

• Searching for antibodies with a given target.

• Manual annotation.

To within a protein class (e.g. human vs mouse EGFR).

GenBank

Antibodies from GenBank.

• Find antibodies from publications/GenBank and their targets.

• Target characterization for analysis and retrievals.

• Publication metadata when available.

To within a protein class (e.g. human vs mouse EGFR).

Nanobodies

Database of nanobodies from public sources.

• Train nanobody-specific models.

• Train nanobody-specific models.

• Large dataset across several databases.

• Better target annotations.

To within a protein class (e.g. human vs mouse EGFR).