Back to list
GPTomics

bio-genome-assembly-hifi-assembly

by GPTomics

a set of SKILLS.md for doing bioinformatics with agents like claude code

65🍴 17📅 Jan 24, 2026

SKILL.md


name: bio-genome-assembly-hifi-assembly description: High-quality genome assembly from PacBio HiFi reads using hifiasm with phasing support. Use when building reference-quality diploid assemblies from HiFi data, especially with trio or Hi-C phasing for fully resolved haplotypes. tool_type: cli primary_tool: hifiasm

HiFi Assembly

Basic Assembly

# Primary assembly (single haplotype consensus)
hifiasm -o output_prefix -t 32 reads.hifi.fastq.gz

# Output files:
# output_prefix.bp.p_ctg.gfa  - Primary contigs
# output_prefix.bp.a_ctg.gfa  - Alternate contigs
# output_prefix.bp.hap1.p_ctg.gfa - Haplotype 1 (if phased)
# output_prefix.bp.hap2.p_ctg.gfa - Haplotype 2 (if phased)

# Convert GFA to FASTA
awk '/^S/{print ">"$2;print $3}' output_prefix.bp.p_ctg.gfa > assembly.fasta

Trio-Binned Phasing

# With parental short reads for trio binning
hifiasm -o trio_asm -t 32 \
    -1 paternal.yak \
    -2 maternal.yak \
    child.hifi.fastq.gz

# Create yak databases from parental Illumina reads first
yak count -b37 -t16 -o paternal.yak paternal_R1.fq.gz paternal_R2.fq.gz
yak count -b37 -t16 -o maternal.yak maternal_R1.fq.gz maternal_R2.fq.gz

Hi-C Phasing

# Use Hi-C reads for phasing (no parents needed)
hifiasm -o hic_asm -t 32 \
    --h1 hic_R1.fastq.gz \
    --h2 hic_R2.fastq.gz \
    reads.hifi.fastq.gz

# Produces fully phased hap1 and hap2 assemblies

Key Parameters

ParameterDefaultDescription
-t1Threads
-l0Purge level (0=none, 1=light, 2=aggressive)
-s0.55Similarity threshold for duplicate detection
--primary-Output primary contigs only (no alternates)
--n-hap2Expected number of haplotypes
-D5.0Drop reads with depth > D*average
-N100Consider up to N overlaps for each read

Purge Duplicates

# Aggressive purging for high heterozygosity
hifiasm -o asm -t 32 -l 2 reads.hifi.fastq.gz

# Minimal purging for inbred samples
hifiasm -o asm -t 32 -l 0 reads.hifi.fastq.gz

Ultra-Long ONT Integration

# Combine HiFi accuracy with ONT length
hifiasm -o hybrid_asm -t 32 \
    --ul ont_ultralong.fastq.gz \
    hifi_reads.fastq.gz

# UL reads help span complex repeats

Assembly Stats

# Quick stats with seqkit
seqkit stats assembly.fasta

# Detailed with assembly-stats
assembly-stats assembly.fasta

# QUAST assessment
quast.py -o quast_output assembly.fasta

# BUSCO completeness
busco -i assembly.fasta -l mammalia_odb10 -o busco_out -m genome

Memory and Runtime

Genome SizeHiFi CoverageRAMTime (32 cores)
3 Gb30x~200 GB12-24 hours
3 Gb60x~400 GB24-48 hours
500 Mb40x~64 GB2-4 hours

Python Wrapper

import subprocess
from pathlib import Path

def run_hifiasm(hifi_reads, output_prefix, threads=32, purge_level=0,
                hic_r1=None, hic_r2=None, ul_reads=None):
    cmd = ['hifiasm', '-o', output_prefix, '-t', str(threads), '-l', str(purge_level)]

    if hic_r1 and hic_r2:
        cmd.extend(['--h1', hic_r1, '--h2', hic_r2])

    if ul_reads:
        cmd.extend(['--ul', ul_reads])

    cmd.append(hifi_reads)
    subprocess.run(cmd, check=True)

    gfa = Path(f'{output_prefix}.bp.p_ctg.gfa')
    fasta = Path(f'{output_prefix}.fasta')

    with open(fasta, 'w') as out:
        with open(gfa) as f:
            for line in f:
                if line.startswith('S'):
                    parts = line.strip().split('\t')
                    out.write(f'>{parts[1]}\n{parts[2]}\n')

    return fasta

# Example
assembly = run_hifiasm('sample.hifi.fq.gz', 'sample_asm', threads=48, hic_r1='hic_R1.fq.gz', hic_r2='hic_R2.fq.gz')

Troubleshooting

IssueSolution
High duplicationIncrease purge level (-l 2)
Missing haplotypesAdd Hi-C or trio data for phasing
Memory errorsReduce -D parameter or downsample reads
Fragmented assemblyCheck read quality; consider UL ONT addition
  • genome-assembly/quality-assessment - QUAST and BUSCO
  • genome-assembly/scaffolding - YaHS Hi-C scaffolding
  • genome-assembly/contamination-detection - CheckM2 decontamination
  • long-read-sequencing/read-qc - HiFi quality control

Score

Total Score

65/100

Based on repository quality metrics

SKILL.md

SKILL.mdファイルが含まれている

+20
LICENSE

ライセンスが設定されている

+10
説明文

100文字以上の説明がある

0/10
人気

GitHub Stars 100以上

0/15
最近の活動

3ヶ月以内に更新がある

0/10
フォーク

10回以上フォークされている

+5
Issue管理

オープンIssueが50未満

+5
言語

プログラミング言語が設定されている

+5
タグ

1つ以上のタグが設定されている

0/5

Reviews

💬

Reviews coming soon