SQANTI_examples

Repository with the examples and data needed to run SQANTI3

This repository contains datasets and comprehensive examples for running SQANTI3, a tool for the quality control and classification of long-read transcripts.

Repository Structure

SQANTI_examples/
├── data/                           # Input data for examples
│   ├── UHR_chr22.gtf              # Input isoform annotations
│   ├── UHR_abundance.tsv          # Abundance data
│   ├── UHR_RQ_abundance.tsv       # RQ abundance data
│   ├── UHR_chr22_short_reads.fofn # Short reads file list
│   ├── reference/                 # Reference genome and annotations
│   │   ├── GRCh38.p13_chr22.fasta
│   │   ├── gencode.v38.basic_chr22.gtf
│   │   └── ...
│   ├── polyA_motifs/              # PolyA motif annotations
│   ├── ref_TSS_annotation/        # TSS reference annotations
│   └── short_reads/               # Short read data
│
└── example/                        # Example workflows and results
    ├── run_all_examples.sh         # Master script to run all examples
    ├── run_SQANTI3_QC.sh           # Quality control example
    ├── run_SQANTI3_MLfilter.sh     # Machine learning filter example
    ├── run_SQANTI3_rules_filter.sh # Rules-based filter example
    │
    ├── config_files/                # Configuration files for each workflow
    │   ├── qc_config.yaml           # QC configuration
    │   ├── qc_config_reference.yaml # QC of the reference with same orthogonal data as the isoforms
    │   ├── filter_ml.yaml           # ML filter configuration
    │   ├── filter_rules.yaml        # Rules filter configuration
    │   ├── rescue_ml.yaml           # ML rescue configuration
    │   ├── rescue_rules.yaml        # Rules rescue configuration
    │   └── rescue_automatic.yaml    # Automatic rescue configuration
    │
    ├── QC_isoforms/               # QC results for isoforms
    ├── QC_reference/              # QC results using reference mode
    ├── filter_ml/                 # ML filtering results
    ├── filter_rules/              # Rules-based filtering results
    ├── rescue_ml/                 # ML rescue results
    ├── rescue_rules/              # Rules-based rescue results
    ├── rescue_automatic/          # Automatic rescue results
    ├── sqanti_reads_test/         # SQANTI-reads test results
    └── Figures_paper/             # Code to generate publication figures
        ├── main_figures_code.R
        └── ExtendedData_figures_code.R

Examples Overview

The example/ directory contains complete workflows demonstrating different SQANTI3 functionalities:

1. Quality Control (QC)

• Script: run_SQANTI3_QC.sh
• Config: config_files/qc_config.yaml, config_files/qc_config_reference.yaml
• Output: QC_isoforms/, QC_reference/
• Description: Performs quality control on long-read isoforms, generating classification files, corrected annotations, and comprehensive QC reports.

2. Machine Learning Filtering

• Script: run_SQANTI3_MLfilter.sh
• Config: config_files/filter_ml.yaml
• Output: filter_ml/
• Description: Uses random forest models to filter isoforms based on quality metrics. Generates classification results, confusion matrices, and filtered output files.

3. Rules-Based Filtering

• Script: run_SQANTI3_rules_filter.sh
• Config: config_files/filter_rules.yaml
• Output: filter_rules/
• Description: Applies user-defined rules to filter isoforms. Produces filtering reason reports and filtered annotations.

4. Rescue Workflows

• Configs: rescue_ml.yaml, rescue_rules.yaml, rescue_automatic.yaml
• Output: rescue_ml/, rescue_rules/, rescue_automatic/
• Description: Demonstrates how to rescue filtered isoforms using different strategies (ML-based, rules-based, or automatic).

5. SQANTI-reads

• Output: sqanti_reads_test/
• Description: Example workflow for SQANTI-reads, which performs read-level quality control.

6. Publication Figures

• Directory: Figures_paper/
• Description: R scripts to reproduce figures from the SQANTI3 publication.

Running the Examples

To run all examples at once:

cd example/
./run_all_examples.sh

To run individual examples:

# Quality Control
./run_SQANTI3_QC.sh

# ML Filtering
./run_SQANTI3_MLfilter.sh

# Rules Filtering
./run_SQANTI3_rules_filter.sh

Data Description

The data/ directory contains:

• Input isoforms: UHR (Universal Human Reference) chr22 GTF files
• Reference genome: GRCh38 chromosome 22 FASTA and annotations
• Abundance data: Transcript expression levels
• Short reads: For junction validation
• Annotation resources: PolyA motifs and TSS (Transcription Start Site) references

Requirements

• SQANTI3 installed and configured
• Python 3.7+
• R 4.0+ (for filtering and rescue functions)
• Required R packages: randomForest, ggplot2, gridExtra, etc.

For detailed installation instructions and requirements, please refer to the SQANTI3 GitHub repository.