FuFiHLA: Full Field HLA allele typing for Long Reads

FuFiHLA is a pipeline for full field HLA allele typing and consensus sequence construction from long-read sequencing data. It currently supports PacBio HiFi data on six clinically important transplant genes: HLA-A, -B, -C, -DQA1, -DQB1, -DRB1.

Highlights

• Reference-free: does not depend on a specific version of reference genome such GRCh38 or CHM13
• Improved consensus accuracy compared to StarPhase

Citation: TBD

---

Installation

Install from Bioconda (recommended):

conda install -c bioconda -c conda-forge fufihla

Quick Test

with 'test.fa.gz' under the folder "test", run:

fufihla --fa test.fa.gz --out test_dir

The output includes:

• test_dir/ → pipeline logs
• test_dir.out → result output
• test_dir.err → stderr log

Usage

To use the latest reference allele sequences from IMGT, type:

fufihla-ref-prep

This will create a directory called ref_data, which would contain the reference allele sequence ref.gene.fa.gz.

run the pipeline:

# with default reference allele sequences, version IPD-IMGT/HLA-V3.61.0
fufihla --fa <input_reads.fa.gz> --out <output_dir>
# or with the specific version of reference data
fufihla --fa <input_reads.fa.gz> --out <output_dir> --refdir <reference data directory> --hifi/--ont --debug

Arguments

• <input_reads.fa.gz> : raw PacBio HiFi reads (.fa/.fa.gz/.fq/.fq.gz)
• <output_dir> : directory for pipeline outputs
• --refdir <reference_data_directory>(optional): path to reference allele dataset; if omitted, uses the default bundled set
• --hifi/--ont(optional): choose HiFi long reads or Nanopore long read data as input, default is --hifi
• --debug(optional): keep all intermediate files; otherwise only consensus results are kept

Outputs

A typical run produces:

<outdir>/consensus/*_asm*.fa        → consensus allele FASTA sequences

Allele calls are printed to <output_dir>.out in PAF-like format with minimap2 tags. Example:

HLA-A*01:01:01:01  cons_HLA-A*01_01_01_01  ...  cs:Z::3503
HLA-A*26:01:01:01  cons_HLA-A*26_01_01_01  ...  cs:Z::3517

• Column 1 → the allele name called by FuFiHLA
• Column 2 → the consensus sequence build upon allele in the suffix
• Last column (cs:Z) → minimap2 cs tag encoding base-level matches/mismatches:
  • Known Alleles: cs:Z::3503 → perfect match over 3503 bp
  • Novel Alleles → cs:Z contains substitutions (*), insertions (+), or deletions (-)

Running tips

• Extract reads from exist bam files can also generate similar result as using WGS reads.

## save the six gene locations into bed format based on the gene annotation file
echo "
chr6	29942254	29945755
chr6	31268254	31272571
chr6	31353362	31357442
chr6	32578769	32589848
chr6	32636717	32643200
chr6	32660031	32667132" > sel.bed

## Extract the reads covering the target gene region
samtools view -bh ${bam} --region-file sel.bed | samtools fasta | gzip -c > out.fa.gz