3D-HiFi: A fast and accurate pipeline for chromosome-scale scaffolding from HiFi-C data

Introduction

  As an emerging high-resolution long-read chromosome conformation capture technique, HiFi-C currently lacks dedicated software tools. Researchers have to adapt pipelines designed for Pore-C, which may underestimate the true potential of HiFi-C data and negatively impact downstream analyses such as scaffolding. To address this issue, we have developed 3D-HiFi, a tool specifically designed for efficient scaffolding using HiFi-C data. 3D-HiFi also supports Pore-C data.

Overview

  Given the substantial length of HiFi-C reads, we employed an analysis strategy involving in silico fragmentation at restriction enzyme cleavage sites prior to alignment.

Table of contents

• Dependencies
• Installation
• Quick start
• Comparison
• Output File
• Get help
• Citating

Dependencies

Software:

• python >=3.7
• pigz
• minimap2
• seqkit
• 3d-dna
• parallel
• juicebox
• mummer

Installation

3D-HiFi has been tested and validated on servers running Linux.

# (1) Download 3D-HiFi from GitHub
$ git clone https://github.com/FrasergenBioinformatics/3D-HiFi.git
# (2) Resolve dependencies
# We strongly recommend using conda to install dependencies. 
$ conda env create -f environment.yml
# Activate the HapHiC conda environment
$ conda activate 3D-HiFi # or: source /path/to/conda/bin/activate 3D-HiFi
# (3) Install 3d-dna
$ git clone https://github.com/aidenlab/3d-dna.git
#Give executable permissions
$ cd 3d-dna
$ chmod +x  *.sh */*

Quick start

usage_example: 3D-HiFi -r PATH/contig.fa -i PATH/HiFi-C.fq.gz -p '-x map-hifi' -t 30 -c 10000 -e GATC -d PATH/3d-dna -o your_species

options:
  -h, --help           show this help message and exit
  -r, --ref            contig genome
  -i, --fq_in          <fastq file>  HiFi-C/Pore-C data
  -p, --map_params     <minimap2 align parameter> (if your data is Pore-C,set "-x map-ont" )
  -t, --threads        number of threads
  -c, --chunk_size     Number of records per processing chunk, If the dataset is large, you can increase the `chunk_size` parameter.
  -e, --enzyme_site    Enzyme recognition site: GATC (MboI/DpnII), AAGCTT (HindIII), CATG(NlaIII)
  -d, --_3ddna_path    3ddna software path
  -o, --output_prefix  output prefix
  -a, --polyploid      Enable polyploid mode can rescue collapsed contigs  (default: disabled)

Comparison

The following table summarizes a performance comparison of 3D-HiFi against other tools (wf-pore-c and Cphasing) across various biological datasets, detailing metrics such as valid reads, processing time, and memory usage.

Dataset	Software	Valid reads	Pairs number	Contacts/Reads	Wall time	RAM
Ceratitis_capitata (27X)	3D-HiFi	1,185,394	11,508,130	8.75	32min	42G
	wf-pore-c	1,163,280	3,264,519	2.48	2.5h	39G
	Cphasing	1,015,477	5,888,352	4.48	48min	42G
Anopheles_coluzzii (56X)	3D-HiFi	2,282,982	64,712,163	27.35	1.2h	70G
	wf-pore-c	2,159,056	9,216,303	3.90	3.5h	65G
	Cphasing	2,258,817	50,310,732	21.27	1.2h	67G
Homo_sapien (26X)	3D-HiFi	9,198,589	720,830,448	78.28	1d5h	147G
	wf-pore-c	8,743,704	46,912,773	5.09	1d7h	61G
	Cphasing	9,144,776	364,842,740	39.62	17.9h	128G
Plecia_longiforceps (49X)	3D-HiFi	15,223,238	143,783,824	6.03	4.7h	96G
	wf-pore-c	13,474,187	35,116,293	1.47	11.2h	86G
	Cphasing	12,793,521	69,399,906	2.91	3.8h	88G
Rosa_hybrida (23X)	3D-HiFi	13,338,018	539,607,578	37.38	16.2h	82G
	wf-pore-c	10,956,145	37,981,725	2.63	1d2h	53G
	Cphasing	7,846,147	53,071,524	3.68	5.6h	71G

Output Files

Primary Output Files and Their Specifications

.
├── 01.split_minimap
│   ├── your_species.paf        # minimap2 result
│   ├── your_species.len        # contig size if you set --polyploid parameter
│   ├── contig.depth            # depth average contig if you set --polyploid parameter
│   ├── collapsed.contig.list   # collapsed contig list if you set --polyploid parameter
│   └── contig.dup.fasta        # rescued contig genome if you set --polyploid parameter
├── 02.paf2mnd
│   ├── your_species.mnd.txt
│   ├── your_species.mnd.sort.txt
│   ├── dups.txt
│   ├── merged_nodups.txt        # nodups mnd file
│   └── tmp
├── 02.02.paf2mnd_dup            # if you set --polyploid parameter
│   ├── your_species.mnd.txt
│   ├── your_species.mnd.dup.txt # new mnd file if you set --polyploid parameter
│   ├── your_species.mnd.sort.txt
│   ├── dups.txt
│   ├── merged_nodups.txt        # nodups mnd file
│   └── tmp
├── 03.3ddna
│   ├── contig.0.asm
│   ├── contig.0_asm.scaffold_track.txt
│   ├── contig.0_asm.superscaf_track.txt
│   ├── contig.0.cprops
│   ├── contig.0.assembly  # Input of Juicebox to manually correct
│   ├── contig.0.hic       # Input of Juicebox to manually correct
│   ├── contig.cprops
│   └── contig.mnd.txt
├── 03.3ddna_dup          # use this directory's result to manually correct if you set --polyploid parameter
│   ├── temp.contig.dup.0.asm_mnd.txt
│   ├── contig.dup.0_asm.scaffold_track.txt
│   ├── contig.dup.0_asm.superscaf_track.txt
│   ├── contig.dup.0.assembly  # Input of Juicebox to manually correct
│   └── contig.dup.0.hic       # Input of Juicebox to manually correct
└── read.summary          # reads mapping stat

Get help

Help

For detailed instructions regarding chromosome ordering, orientation, and visualization, please see automated_orient_visualization_pipeline.

Feel free to raise an issue at the isssue page

Note: Please ask questions on the issue page first. They are also helpful to other users.

Contact

For addtional help, please send an email to wanghuan@frasergen.com or zhengshang@frasergen.com.

Citating

If you use 3D-HiFi in your work,please cite: