Releases · CCU-Bioinformatics-Lab/longphase-to

Overview

LongPhase-TO is a tumor-only variant phasing and analysis pipeline extending LongPhase for ONT long-read sequencing data.
It focuses on somatic variant refinement, LOH-aware phasing, and downstream analyses in cancer genomics.

This release provides a complete end-to-end workflow from aligned ONT reads and somatic VCFs to phased variants, LOH calls, purity estimation, and haplotagged BAMs.

Key Features

Tumor-only somatic phasing
- Joint phasing of germline and somatic variants from tumor-only ONT data
- Supports SNP-only phasing and SNP+indel co-phasing (--indels)
Somatic variant recalibration
- Haplotype-aware re-calling that refines the VCF FILTER field based on somatic analysis
- Adds LP_nonSomatic and LP_PON FILTER tags
- Optional --disable-refine-somatic to keep original caller filters
LOH (Loss of Heterozygosity) support
- Arm-level LOH detection (--loh)
- LOH-aware representation of genotypes using extended GT, GT2, and GT3 fields
- Explicit modeling of germline, somatic, and LOH states in the phased VCF
- CNV/BFB detection (--sge)
Tumor purity estimation
- Estimates tumor purity from phased somatic and LOH-aware information
- Writes the result into the VCF header using the ##tumor_purity=<value> tag
Haplotagging
- Assigns each read in BAM to a phased haplotype (HP and PS tags)
- Supports optional region restriction and output in BAM or CRAM
Panels of Normals (PoNs) integration
- Uses PoN VCFs to distinguish somatic vs. non-somatic variants
- --pon-file (position-based) and --strict-pon-file (position+ALT-based) options
- Compatible with PoN resources from ClairS-TO and related pipelines
Caller compatibility
- Designed for ONT long-read data
- Supports tumor-only somatic VCFs from:
  - clairs_to_ss
  - clairs_to_ssrs
  - deepsomatic_to