Database Summary & Statistics

This page provides a comprehensive overview of the Iggytop database created by the create_anndata.py pipeline. It reports details on the distribution of data sources, species, MHC information, and receptor quality.

This report summarizes the processed and deduplicated AIRR data used for downsteam analysis and knowledge graph construction. If you want to reporodice this, please run create_anndata.py to generate the data.

Executive Summary

A snapshot of the current database size and content after cross-database deduplication.

The database version used here is: data-2026.09.01.021958.

Total Records

700,671

Deduplicated Records

370,772

Reduction by 47.1%

Unique Epitopes

11,606

Publications

4,659

TCR Records

362,458

BCR Records

8,314

Note: If you find this report to be outdated or encounter any issues with the data, please open an issue on GitHub.

Source and Organism Distributions

We track where the data originates and the species diversity. Many entries are found in multiple databases simultaneously.

MHC Context Availability

Availability of MHC class and gene information varies significantly by database origin. We visualize the proportion of records containing MHC Class I or Class II information.

Missing MHC class (unknown I/II): 9,781. Missing MHC gene (MHC_gene_1): Class I = 0, Class II = 0.

Source counts for 'HLA class I' entries:

source
IEDB          87016
CEDAR|IEDB       35
Name: count, dtype: int64

TCR Chain Configuration (QC)

Using the scirpy.tl.chain_qc tool, we categorize the structural quality of the receptors. A high proportion of “Productive” pairs (Alpha+Beta) indicates better biological quality for structural modeling.

Sequence Distributions and Coverage

CDR3 (junction_aa) and Epitope lengths for biological consistency validation. Consistent distributions across sources suggest cross-database compatibility.

Junction Sequence Logos by Receptor Type

Because junction_aa sequences vary in length, logos are computed on fixed windows of 8 amino acids from the left (start) and from the right (end). For each receptor class (TCR, BCR), we show four logos side by side in a 2x2 layout: VJ-left, VJ-right, VDJ-left, and VDJ-right.

/home/docs/checkouts/readthedocs.org/user_builds/iggytop/envs/latest/lib/python3.13/site-packages/logomaker/src/error_handling.py:58: UserWarning:  Warning: Character '*' is not in color_dict. Using black.
  warnings.warn(str(Error))

/home/docs/checkouts/readthedocs.org/user_builds/iggytop/envs/latest/lib/python3.13/site-packages/logomaker/src/error_handling.py:58: UserWarning:  Warning: Character 'X' is not in color_dict. Using black.
  warnings.warn(str(Error))

/home/docs/checkouts/readthedocs.org/user_builds/iggytop/envs/latest/lib/python3.13/site-packages/logomaker/src/error_handling.py:58: UserWarning:  Warning: Character '*' is not in color_dict. Using black.
  warnings.warn(str(Error))

V/J Gene Presence

	Chain	Gene	Unique genes	Records
0	VDJ_1	J	157	304216
1	VDJ_1	V	850	313737
2	VJ_1	J	215	148840
3	VJ_1	V	779	158221

Records with complete annotations for both chains
(V gene + J gene + CDR3 amino-acid sequence for VJ_1 and VDJ_1):

97,588 / 370,772 (26.32%)