bio-single-cell-splicing

BRIE2 Analysis

Goal: Estimate per-cell PSI values for splicing events with uncertainty quantification.

Approach: Prepare splicing events from annotation, count reads per cell barcode, then fit a Bayesian variational inference model for probabilistic PSI estimation.

"Analyze splicing in single-cell data" -> Estimate per-cell inclusion levels for splicing events with uncertainty.

• Python: BRIE2 (probabilistic PSI, handles sparsity)
• Python/R: leafcutter2 (intron clustering, NMD detection)

import brie
import scanpy as sc
import anndata as ad

## BRIE2 uses pre-defined splicing events

brie.preprocessing.get_events(
    gtf_file='annotation.gtf',
    out_file='splicing_events.gff3'
)

## Requires cell barcodes and UMIs

brie.preprocessing.count(
    bam_file='possorted_genome_bam.bam',
    gff_file='splicing_events.gff3',
    out_dir='brie_counts/',
    cell_file='barcodes.tsv'  # Filtered cell barcodes
)