Tutorials

BRIDGE provides notebook-based, task-oriented workflows that map directly to the main use cases in the repository. Each tutorial is designed to be runnable end-to-end, with clearly specified inputs, key parameters, and expected outputs.

---

Train BRIDGE models for the RBPs in different cell lines

Train BRIDGE models for the RBPs in different cell lines

Benchmark across six cell lines

Evaluate trained models on held-out data

Validate the performance of models using BRIDGE

Benchmark across six cell lines

Run dynamic cross-cell-type prediction of RBP–RNA interactions

Dynamic cross-cell-type prediction of RBP–RNA interactions

Benchmark across six cell lines

RBP motif extraction and visualization

Using BRIDGE to discover motifs

Motif discovery

Catalog variants tutorial (variant_aware.py — ClinVar/TCGA/1000G batches)

ClinVar / TCGA / 1000 Genomes pipeline tutorial

Variant-aware prediction

Ribosnitch tutorial (variant_aware.py — ribosnitch mode)

Ribosnitch pipeline tutorial

Variant-aware prediction

GWAS tutorial (variant_aware.py — GWAS mode)

GWAS pipeline tutorial

Variant-aware prediction

Contents

Benchmarking

Benchmarking refers to training and evaluating BRIDGE models under standardized settings (e.g., across multiple RBPs and cell lines), and optionally testing transfer/generalization via dynamic cross-cell-type prediction.

Motif discovery

Motif discovery refers to extracting interpretable binding patterns from BRIDGE outputs and comparing/dissecting these patterns with downstream motif tools and visualizations.

Variant-aware prediction

Variant-aware prediction refers to scoring genetic variants by comparing model predictions between reference and alternative alleles (or perturbed sequences), producing variant impact scores that can be aggregated for clinical datasets or genome-wide scans.