supremo_liteο
A lightweight memory-first, model-agnostic version of SuPreMo.
Key Featuresο
𧬠Personalized Genome Generation: Apply variants from VCF files to reference genomes
π― Variant-Centered Sequences: Generate sequence windows around variants
βοΈ PAM Site Analysis: Identify variants that disrupt CRISPR PAM sites
π§ͺ Saturation Mutagenesis: Systematic single-nucleotide mutations at every position for predictive modeling
π§ Memory Efficient: Chunked processing for large VCF files
πΊοΈ Chromosome Matching: Optional handling of chromosome naming differences (chr1 β 1, chrM β MT) via
auto_map_chromosomes=Trueβ‘ PyTorch Integration: Automatic tensor support when PyTorch is available
Installationο
Install from GitHub (Recommended)ο
For the latest features and bug fixes:
# Install directly latest release
pip install supremo-lite
# Or install a specific version/tag
pip install git+https://github.com/gladstone-institutes/supremo_lite.git@v0.5.0
Dependenciesο
Required dependencies will be installed automatically:
pandas- For VCF data handlingnumpy- For numerical operationspyfaidx- For FASTA file reading
Optional dependencies:
torch- For PyTorch tensor support (automatically detected)brisket - Cython powered faster 1 hot encoding for DNA sequences (automatically detected)
Quick Startο
import supremo_lite as sl
from pyfaidx import Fasta
# Load reference genome and variants
reference = Fasta('hg38.fa')
variants = sl.read_vcf('variants.vcf')
DNA Sequence Encodingο
supremo_lite uses one-hot encoding by default:
A=[1,0,0,0],C=[0,1,0,0],G=[0,0,1,0],T=[0,0,0,1]Ambiguous bases =
[0,0,0,0]Returns PyTorch tensors when available, otherwise NumPy arrays
Personalized Genome Generationο
# Apply variants to create personalized genome
personal_genome = sl.get_personal_genome(
reference_fn=reference,
variants_fn=variants,
encode=True, # One-hot encoded (or False for strings)
chunk_size=10000, # Process 10k variants at a time
verbose=True # Show progress
)
# If your VCF uses 'chr1' and reference uses '1', enable chromosome mapping
personal_genome = sl.get_personal_genome(
reference_fn=reference,
variants_fn=variants,
auto_map_chromosomes=True # Handle chromosome name differences
)
Variant-Centered Sequencesο
# Generate reference and alternate sequences around variants
# Note: get_alt_ref_sequences is a generator that yields chunks
results = list(sl.get_alt_ref_sequences(
reference_fn=reference,
variants_fn=variants,
seq_len=1000,
encode=True
))
# Unpack from the first chunk
alt_seqs, ref_seqs, metadata = results[0]
# Returns: (n_variants, seq_len, 4) shaped arrays
π Full Guide: Variant-Centered Sequences | Getting Started Notebook
Prediction Alignmentο
# Align model predictions accounting for variant coordinate changes
from supremo_lite.mock_models import TestModel
model = TestModel(n_targets=2, bin_size=8, crop_length=10)
ref_preds = model(ref_seqs)
alt_preds = model(alt_seqs)
ref_aligned, alt_aligned = sl.align_predictions_by_coordinate(
ref_pred=ref_preds[0],
alt_pred=alt_preds[0],
metadata=metadata[0],
prediction_type="1D",
bin_size=8,
crop_length=10
)
π Full Guide: Prediction Alignment | Tutorial with Visualizations
Saturation Mutagenesisο
# Mutate every position in a region
ref_seq, alt_seqs, metadata = sl.get_sm_sequences(
chrom='chr1',
start=1000,
end=1100, # 100 bp β 300 mutations (3 per position)
reference_fasta=reference
)
Documentationο
π User Guidesο
Detailed documentation for each major feature:
Personalized Genomes - Apply variants to genomes
Variant-Centered Sequences - Extract sequence windows around variants
Prediction Alignment - Align model predictions for variant effect analysis
Saturation Mutagenesis - In-silico mutagenesis workflows
Variant Classification - Flow chart showing automatic variant classification logic
π Interactive Tutorialsο
Hands-on Jupyter notebooks with visualizations:
Getting Started - Installation and basic concepts
Personalized Genomes - Genome personalization workflows
Prediction Alignment - Complete prediction workflow with visualizations β
π API Referenceο
Core Functions:
get_personal_genome()- Generate personalized genomesget_alt_ref_sequences()- Generate variant-centered sequencesalign_predictions_by_coordinate()- Align model predictionsget_sm_sequences()- Saturation mutagenesisread_vcf()- Read VCF files
For complete API documentation with all parameters, see the docs/ directory.
Issues and Supportο
We welcome feedback, bug reports, and feature requests! If you encounter any issues or have suggestions for improvements, please:
Check existing issues first to see if your problem has already been reported
File a new issue on our GitHub Issues page
Provide detailed information including:
Python version and operating system
Package version (
supremo_lite.__version__)Complete error messages and stack traces
Minimal reproducible example
Expected vs. actual behavior
Common Issues to Reportο
Performance problems with large genomes or variant files
Unexpected behavior with edge cases
Documentation gaps or unclear examples
Feature requests for new functionality
Contributingο
Interested in contributing? Check out the contributing guidelines. Please note that this project is released with a Code of Conduct. By contributing to this project, you agree to abide by its terms.
Licenseο
supremo_lite was created by Natalie Gill and Sean Whalen, based on Sequence Mutator for Predictive Models (SuPreMo) by Katie Gjoni. It is licensed under the terms of the MIT license.
Creditsο
supremo_lite was created with cookiecutter and the py-pkgs-cookiecutter template.
Quick Startο
What do you want to do? |
Go to |
|---|---|
Install and learn basics |
|
Apply variants to genome |
|
Generate variant sequences |
|
Identify PAM-disrupting variants |
|
Align model predictions |
|
Understand alignment visually |
|
Perform mutagenesis |
|
Understand variant types |