Hi-C Chromosome-Scale Assembly and 3D Chromatin Analysis — Coral Hybrid Genomes

MSc Research Project | University of Melbourne | GADI HPC (NCI Australia) Project codes: which ever you are using under your main project on server

This repository contains the full bioinformatics pipeline used for chromosome-length Hi-C assembly, comparative genomics, and 3D chromatin analysis of sperm hybrid coral genomes on the GADI HPC system (NCI Australia).

The project investigated genomic signatures, TAD organisation, A/B chromatin compartments, and large-scale chromosomal rearrangements, underpinning fitness in hybrid corals. Sample A was used as the reference genome for all comparative analyses.

---

Repository Structure

hic-coral-hybrids/
├── README.md
├── 01_index_genome/
│   └── 01_index_genome.pbs              # BWA index reference genome
├── 02_restriction_sites/
│   └── 02_generate_restriction_sites.pbs # Generate Arima restriction sites
├── 03_chrom_sizes/
│   └── 03_create_chrom_sizes.sh          # Create chromosome sizes file
├── 04_split_hic_reads/
│   └── 04_split_hic_reads.pbs            # Split Hi-C reads for parallel processing
├── 05_juicer/
│   └── 05_juicer.pbs                     # Juicer alignment and contact map generation
├── 06_yahs/
│   └── 06_yahs.pbs                       # YaHS chromosome-scale scaffolding
├── 07_juicer_pre/
│   └── 07_juicer_pre.pbs                 # Prepare JBAT input files
├── 08_juicer_tools/
│   └── 08_juicer_tools.pbs               # Generate .hic file for Juicebox
├── 09_manual_curation/
│   └── NOTES.md                          # Manual Juicebox curation (GUI — no script)
├── 10_matrix_processing/
│   └── 10_matrix_processing.pbs          # .hic → .cool → .h5, ICE normalisation
├── 11_tad_calling/
│   └── 11_tad_calling.pbs                # TAD calling with hicFindTADs
├── 12_ab_compartments/
│   └── 12_ab_compartments.pbs            # A/B compartment identification (hicPCA)
└── 13_circos/
    ├── index_karyotype.py                # Generate Circos karyotype file
    ├── colorpaf_link.py                  # Convert PAF to coloured Circos links
    └── NOTES.md                          # Circos workflow notes

---

Pipeline Overview

Raw Hi-C reads
      │
      ▼
[01] Index reference genome         BWA index
      │
      ▼
[02] Generate restriction sites     generate_site_positions.py (Arima)
      │
      ▼
[03] Create chromosome sizes        awk from .fai file
      │
      ▼
[04] Split Hi-C reads               split (20M reads per chunk)
      │
      ▼
[05] Juicer alignment               BWA + samtools → merged_nodups.bam
      │
      ▼
[06] YaHS scaffolding               yahs → chromosome-scale assembly
      │
      ▼
[07] Prepare JBAT files             juicer pre → out_JBAT.hic + .assembly
      │
      ▼
[08] Generate .hic for Juicebox     juicer_tools pre → out_JBAT.hic
      │
      ▼
[09] Manual curation (JBAT)         Juicebox GUI — correct misjoins
      │
      ▼
[10] Matrix processing              .hic → .cool → .h5, ICE normalisation
      │
      ├──► [11] TAD calling          hicFindTADs (25 kb, FDR q < 0.05)
      │
      ├──► [12] A/B compartments     hicPCA eigenvector decomposition (2.5 Mb)
      │
      └──► [13] Circos plots         index_karyotype.py + colorpaf_link.py

---

Step Details

Step 01 — Index Reference Genome

qsub 01_index_genome/01_index_genome.pbs

Creates a soft link to the reference FASTA and indexes it with BWA. Required before Juicer alignment.

Resources: 1 CPU · 50 GB · 30 min

---

Step 02 — Generate Restriction Sites

qsub 02_restriction_sites/02_generate_restriction_sites.pbs

Generates Arima restriction site positions for both haplotypes using generate_site_positions.py from the Juicer toolkit.

Resources: 1 CPU · 50 GB · 1.5 hr

---

Step 03 — Create Chromosome Sizes

bash 03_chrom_sizes/03_create_chrom_sizes.sh

Extracts chromosome names and sizes from the restriction sites file using awk. This is a short local command — no PBS needed.

awk 'BEGIN{OFS="\t"}{print $1, $NF}' \
  restriction_sites/sample_hap1_clean.fasta_Arima.txt \
  > chrom_sizes_sample_hap1

---

Step 04 — Split Hi-C Reads

qsub 04_split_hic_reads/04_split_hic_reads.pbs

Splits paired-end Hi-C reads into chunks of 20 million reads for parallel processing by Juicer.

Resources: 1 CPU · 5 GB · 1 hr

---

Step 05 — Juicer Alignment and Contact Map Generation

qsub 05_juicer/05_juicer.pbs

Runs the full Juicer pipeline: BWA alignment, duplicate removal, and generation of merged_nodups.bam. Uses Arima enzyme configuration.

Resources: 24 CPUs · 80 GB · 8 hr

---

Step 06 — YaHS Chromosome-Scale Scaffolding

qsub 06_yahs/06_yahs.pbs

Runs YaHS to scaffold contigs into chromosome-length sequences using Hi-C contact information from the merged BAM file.

Resources: 8 CPUs · 190 GB · 2 hr

---

Step 07 — Prepare Juicebox Input Files

qsub 07_juicer_pre/07_juicer_pre.pbs

Runs juicer pre to generate out_JBAT.hic and out_JBAT.liftover.assembly files for manual curation in Juicebox.

Resources: 2 CPUs · 50 GB · 30 min

---

Step 08 — Generate .hic File with Juicer Tools

qsub 08_juicer_tools/08_juicer_tools.pbs

Generates the final .hic contact map file using Juicer Tools for visualisation in Juicebox.

Resources: 4 CPUs · 64 GB · 1 hr

---

Step 09 — Manual Assembly Curation (Juicebox JBAT)

See 09_manual_curation/NOTES.md for full instructions.

This step is performed manually using the Juicebox Assembly Tools GUI. Misjoins are identified as off-diagonal signal breaks and corrected by dragging chromosome segments. The curated assembly is exported as a .assembly file.

---

Step 10 — Hi-C Matrix Processing and Normalisation

qsub 10_matrix_processing/10_matrix_processing.pbs

Converts .hic files to .cool and .h5 formats, then applies ICE normalisation using HiCExplorer and cooler. Two resolutions are generated:

• 25 kb — for TAD calling
• 2.5 Mb — for A/B compartment analysis

Resources: 8 CPUs · 64 GB · 4 hr

---

Step 11 — TAD Calling

qsub 11_tad_calling/11_tad_calling.pbs

Calls Topologically Associating Domains (TADs) using the insulation score algorithm with FDR correction (q < 0.05) at 25 kb resolution.

Outputs: _domains.bed, _boundaries.bed, _score.bedgraph

Resources: 8 CPUs · 64 GB · 4 hr

---

Step 12 — A/B Compartment Identification

qsub 12_ab_compartments/12_ab_compartments.pbs

Identifies A/B chromatin compartments via eigenvector decomposition (PC1) of observed/expected contact matrices at 2.5 Mb resolution.

• A compartment (PC1 ≥ 0): open chromatin, gene-rich, transcriptionally active
• B compartment (PC1 < 0): closed chromatin, heterochromatic, repressed

Resources: 8 CPUs · 64 GB · 4 hr

---

Step 13 — Circos Chromosomal Rearrangement Plots

See 13_circos/NOTES.md for full workflow.

Python scripts generate karyotype and link files from genome .fai index files and D-GENIES PAF alignments. Circos visualises inversions, fissions, and fusions across hybrid genomes relative to reference 2578.

python 13_circos/index_karyotype.py   # generate karyotype file
python 13_circos/colorpaf_link.py     # generate coloured link file
# then bundle, filter, and run circos — see NOTES.md

---

Key Tools

Tool	Version	Purpose
BWA	latest	Reference genome indexing and alignment
Juicer	latest	Hi-C alignment pipeline
YaHS	latest	Chromosome-scale scaffolding
Juicer Tools	1.9.9	.hic file generation
Juicebox/JBAT	latest	Manual assembly curation (GUI)
HiCExplorer	3.7	Matrix processing, TAD calling, compartments
cooler	latest	ICE normalisation
D-GENIES	1.2.0	Synteny dot plots (PAF output)
Circos	0.69	Chromosomal rearrangement plots
samtools	latest	BAM sorting and indexing
Python	3.x	Karyotype and link file generation

---

HPC Environment (GADI, NCI Australia)

Jobs are submitted using PBS Pro. To submit any step:

qsub <script>.pbs

To check job status:

qstat -u <username>

Conda environments are activated using:

source /path/to/conda-setup.sh
conda activate <env_name>

---

References

• Li et al. (2009) — samtools
• Zhou et al. (2023) — YaHS
• Durand et al. (2016) — Juicebox/JBAT
• Cabanettes & Klopp (2018) — D-GENIES
• Krzywinski et al. (2009) — Circos
• Wolff et al. (2020) — HiCExplorer
• Abdennur & Mirny (2020) — cooler
• Crane et al. (2015) — Insulation score / TAD calling
• Lieberman-Aiden et al. (2009) — A/B compartments
• Rao et al. (2014) — Hi-C analysis