Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2025
|
| Online Access: | https://doi.org/10.5281/zenodo.18092744 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Table of Contents:
- <h2>A2Denovo v1.0.0</h2> <p>Initial release of A2Denovo - a hybrid <em>de novo</em> variant detection framework combining <strong>assembly-based discovery</strong> with <strong>alignment-based refinement</strong> for PacBio HiFi long-read trio sequencing.</p> <h3>Hybrid Approach</h3> <ul> <li><strong>Assembly-based discovery</strong>: Candidate DNVs identified from pangenome/assembly-based variant calls</li> <li><strong>Alignment-based refinement</strong>: ML classification using read-level features extracted from BAM alignments</li> </ul> <h3>Features</h3> <ul> <li>DNV candidate extraction from trio VCF</li> <li>Comprehensive read-level and genomic context feature extraction</li> <li>L1/L2 logistic regression classifier with Platt scaling probability calibration</li> <li>Post-prediction read-level filtering for high-confidence calls</li> <li>Pre-trained model included (trained on 7 Korean trios, PacBio HiFi Revio ~30x)</li> </ul> <h3>Requirements</h3> <ul> <li>Python 3.8+</li> <li>pysam, pandas, numpy, scikit-learn, tqdm</li> <li>Hail</li> </ul> <h3>License</h3> <p>MIT License.</p> <p>See <a href="README.md">README.md</a> for full documentation.</p>