quick_reference - ConesaLab/SQANTI-browser GitHub Wiki

Quick Reference

One-page cheat sheet for common SQANTI-browser tasks.

How to Run

  • python -m sqanti_browser — Recommended. Uses your active Python (e.g. conda).
  • sqanti_browser — After pip install -e . with the correct Python.
  • python sqanti_browser.py — From project root; set PYTHONPATH=. if needed.

⚡ Most Common Commands

# Basic conversion
python -m sqanti_browser \
    --gtf corrected.gtf \
    --classification classification.txt \
    --output my_hub \
    --genome hg38

# With all validation tracks
python -m sqanti_browser \
    --gtf corrected.gtf \
    --classification classification.txt \
    --output my_hub \
    --genome hg38 \
    --star-sj SJ.out.tab \
    --CAGE-peak CAGE_peaks.bed \
    --polyA-peak polyA_peaks.bed \
    --refGTF reference.gtf \
    --tables

# Custom genome (.2bit)
python -m sqanti_browser \
    --gtf corrected.gtf \
    --classification classification.txt \
    --output my_hub \
    --genome my_species_v1 \
    --twobit genome.2bit

📤 Upload to UCSC (3 Steps)

Step 1: Upload to GitHub

# In your output directory
git init
git add .
git commit -m "Add SQANTI track hub"
git remote add origin https://github.com/username/repo.git
git push -u origin main

Step 2: Get Raw URL

https://raw.githubusercontent.com/username/repo/main/hub.txt

Step 3: Load in UCSC

  1. Go to genome.ucsc.edu
  2. My Data → Track Hubs → My Hubs
  3. Paste your raw URL
  4. Click Add Hub

🔧 Common Options

Flag What it does When to use
--tables Generate HTML reports Always recommended
--sort-by FL Sort by expression Show highest expressed first
--no-category-tracks Skip category tracks Faster, simpler hub
--category-tracks FSM,ISM,NIC Only these categories Custom subset of tracks
--no-highlight Disable highlighted isoforms Uniform category colors
--star-sj FILE Add junction track You have STAR output
--CAGE-peak FILE Validate TSS Assess 5' accuracy
--polyA-peak FILE Validate TTS Assess 3' accuracy
--refGTF FILE Show reference Compare with annotation
--twobit FILE Custom genome Non-model organism

🚨 Quick Troubleshooting

Problem Quick Fix
Tools not found bash install_ucsc_tools.sh
Hub not loading Use raw GitHub URL (raw.githubusercontent.com)
Tracks not updating Add udcTimeout=5 to URL after hgTracks?
Filters not working Right-click track → Configure
Search returns nothing Use prefixed terms: structural_category_novel_in_catalog

📊 Performance Expectations

Dataset Size Processing Time Memory Needed
~1,000 transcripts 1-2 minutes 500MB
~10,000 transcripts 5-10 minutes 1GB
~100,000 transcripts 30-60 minutes 4GB

🔎 Search Cheat Sheet

# Find specific gene
BRCA1

# Find FSM isoforms
structural_category_full-splice_match

# Find coding NIC on plus strand
structural_category_novel_in_catalog strand_plus coding_coding

# Find validated novel isoforms
structural_category_novel_in_catalog within_CAGE_peak_True

# Find potential NMD targets
predicted_NMD_True

🎯 Validation Before Upload

# Test your setup
python tests/test_sqanti_browser.py

# Validate inputs only
python -m sqanti_browser ... --validate-only

# Validate hub structure (run from output directory)
hubCheck hub.txt

# Dry run (test without creating hub)
python -m sqanti_browser ... --dry-run

# Run example workflow
python example/example_usage.py

📁 Expected Output Structure

my_hub/
├── hub.txt                    ← Main entry point
├── genomes.txt
└── hg38/
    ├── hg38_sqanti3.bb        ← Main track (all transcripts)
    ├── hg38_sqanti3_*.bb      ← Category tracks
    ├── trackDb.txt
    └── trix.ix/ixx            ← Search indexes

💡 Pro Tips

  1. Always use --tables for interactive HTML reports
  2. Add --refGTF to compare with reference annotation
  3. Use hubCheck before uploading to catch errors early
  4. Sort by FL to see highest expressed isoforms first
  5. Keep your hub small - use --no-category-tracks if you have >50K transcripts

📚 Need More Details?