Antibody Sequences for Antigen Classification with Machine Learning Features

A collection of curated antibody heavy chain sequences from NCBI databases, used to develop computational models for antigen specificity classification. The dataset includes extracted features such as physicochemical properties, structural composition, sequence order, and evolutionary information. It was created by Jia Lin and last updated in May 2026.

Use Cases

• Training classifiers to predict antibody specificity for antigens like dengue virus or SARS-CoV-2 based on sequence features.
• Analyzing the importance of specific amino acids like cysteine or isoleucine for antibody function using feature importance frameworks.
• Benchmarking deep learning architectures like Feature-Based Transformers against tree-based models for sequence classification tasks.

Strengths

• Dataset is associated with a published study where a CatBoost model achieved an accuracy of 0.7713 for classification.
• Features were extracted from sequences using a multifaceted approach, including physicochemical and evolutionary information.
• The dataset is small (5.5 KB), facilitating quick download and initial exploration.

Limitations

• Row count is unknown, which may limit suitability assessment for large-scale modeling.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is limited to five predefined antigen classes (e.g., anti-dengue, anti-SARS-CoV-2).

Provenance

Source

Curated antibody sequences from NCBI databases.

Collection Method

Sequences were collected and multifaceted features were computationally extracted.

Freshness

Last updated 2026-05-21 07:32:41; freshness should be verified.