SPiP: Benchmark Dataset for Deep Learning Splice Prediction Algorithms

A benchmark dataset for evaluating deep learning splice prediction algorithms, converted and shared by Nathan Fortier. The dataset excludes 15 specific genetic variants due to formatting issues or complex variant types that could not be unambiguously resolved. It was last updated on May 13, 2026.

Use Cases

• Benchmarking deep learning models for splice prediction based on variant data.
• Training models to predict splicing outcomes from genetic variants.
• Evaluating algorithm performance on a standardized set of splicing-relevant variants.
• Studying the impact of specific excluded variant types on prediction tasks.

Strengths

• Dataset is explicitly designed as a benchmark for algorithm comparison.
• Specifics on data cleaning are provided, noting 15 excluded variants due to formatting or complexity.

Limitations

• The dataset is 4.0 MB, which suggests a relatively small scale.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Converted from a GitHub repository associated with the author.

Collection Method

Likely compiled from public genomic data sources for benchmarking purposes.

Freshness

Last updated 2026-05-13 17:42:55