INTRO

As part of project-lake-trout (GitHub repo), Steven asked that I assemble the genome of a lean ecotype of lake trout (GitHub Issue), Salvelinus namaycush, using PacBio HiFi reads. I previously assembled this genome using Flye, on 20230326 (notebook entry), but Steven asked me to also try hifiasm (GitHub Issue)for the assembly.

I performed the assembly on Hyak (Klone), UW’s high-performance computing cluster. I used our Apptainer (Singularity) container to run the job (srlab-R4.4-bioinformatics-container-e5bcfea.sif), which includes hifiasm version 0.25.0.

Below is the rendered markdown from 13.1-hifiasm-genome-assembly-lean.Rmd. The run was very fast; it only took a day and half, vs the ~7 days for the Flye assembly.

PRIMARY OUTPUTS

hifiasm produces three assemblies: a primary assembly, and two haplotype-resolved assemblies. Below are the links to the FastA files. The outputs are only GFA graph files. I had to convert them to FastA using gfatools.

Primary assembly:

FastA (3.5GB): https://gannet.fish.washington.edu/gitrepos/project-lake-trout/output/13.1-hifiasm-genome-assembly-lean/pb-hifiasm-lean-assembly.fa
MD5: https://gannet.fish.washington.edu/gitrepos/project-lake-trout/output/13.1-hifiasm-genome-assembly-lean/pb-hifiasm-lean-assembly.fa.md5
FastA index: https://gannet.fish.washington.edu/gitrepos/project-lake-trout/output/13.1-hifiasm-genome-assembly-lean/pb-hifiasm-lean-assembly.fa.fai

Haplotype 1:

FastA (3.2GB): https://gannet.fish.washington.edu/gitrepos/project-lake-trout/output/13.1-hifiasm-genome-assembly-lean/pb-hifiasm-lean-hap1-assembly.fa
MD5: https://gannet.fish.washington.edu/gitrepos/project-lake-trout/output/13.1-hifiasm-genome-assembly-lean/pb-hifiasm-lean-hap1-assembly.fa.md5
FastA index: https://gannet.fish.washington.edu/gitrepos/project-lake-trout/output/13.1-hifiasm-genome-assembly-lean/pb-hifiasm-lean-hap1-assembly.fa.fai

Haplotype 2:

FastA (2.7GB): https://gannet.fish.washington.edu/gitrepos/project-lake-trout/output/13.1-hifiasm-genome-assembly-lean/pb-hifiasm-lean-hap2-assembly.fa
MD5: https://gannet.fish.washington.edu/gitrepos/project-lake-trout/output/13.1-hifiasm-genome-assembly-lean/pb-hifiasm-lean-hap2-assembly.fa.md5
FastA index: https://gannet.fish.washington.edu/gitrepos/project-lake-trout/output/13.1-hifiasm-genome-assembly-lean/pb-hifiasm-lean-hap2-assembly.fa.fai

After this, I’ll compare the two assemblies (Flye vs. hifiasm) using QUAST, which will provide a more comprehensive set of assembly statistics and metrics.

1 BACKGROUND

Use hifiasm (GitHub) (Cheng et al. 2021, 2022, 2024) to assemble PacBio reads for S.namaycush lean ecotype.

Due to large file sizes, the outputs will not be on GitHub. They may be found here instead:

https://gannet.fish.washington.edu/gitrepos/project-lake-trout/output/13.1-hifiasm-genome-assembly-lean/

2 SETUP

2.1 Libraries

2.2 Set R variables

# OUTPUT DIRECTORY
data_dir <- "/mmfs1/gscratch/scrubbed/samwhite/gitrepos/RobertsLab/project-lake-trout/data/pacbio-lean"
output_dir <- "../output/13.1-hifiasm-genome-assembly-lean"

# PROGRAMS
gfatools <- c("/srlab/programs/gfatools/gfatools")
hifiasm <- c("/srlab/programs/miniforge3-24.7.1-0/envs/hifiasm-0.25.0_env/bin/hifiasm")


# SETTINGS
threads <- "40"


# Export these as environment variables for bash chunks.
Sys.setenv(
  data_dir = data_dir,
  gfatools = gfatools,
  hifiasm=hifiasm,
  output_dir = output_dir,
  threads = threads
)

# Make output directory, if it doesn't exist
mkdir --parents "${output_dir}"

cd "${output_dir}"

# Create an array of .fastq.gz files
fastqs=(${data_dir}/*.fastq.gz)

# Print the result (optional, for verification)
## newline-delimited format
echo "List of FastQs used for assembly:"

printf "%s\n" "${fastqs[@]}" \
| tee input_fastqs.txt

# Run hifiasm assembly
"${hifiasm}" \
-o pb-hifiasm-lean-assembly.asm \
-t "${threads}" \
"${fastqs[@]}" \
> pb-hifiasm-lean-assembly.log 2>&1

3 GFATOOLS

3.1 Primary contigs to FastA

This will extract a FastA file from the graph files output by hifiasm.

cd "${output_dir}"

"${gfatools}" gfa2fa \
pb-hifiasm-lean-assembly.asm.bp.p_ctg.gfa \
> pb-hifiasm-lean-assembly.fa

samtools faidx pb-hifiasm-lean-assembly.fa

3.2 Haplotype 1 contigs to FastA

This will extract a FastA file from the graph files output by hifiasm.

cd "${output_dir}"

"${gfatools}" gfa2fa \
pb-hifiasm-lean-assembly.asm.bp.hap1.p_ctg.gfa \
> pb-hifiasm-lean-hap1-assembly.fa

samtools faidx pb-hifiasm-lean-hap1-assembly.fa

3.3 Haplotype 2 contigs to FastA

This will extract a FastA file from the graph files output by hifiasm.

cd "${output_dir}"

"${gfatools}" gfa2fa \
pb-hifiasm-lean-assembly.asm.bp.hap2.p_ctg.gfa \
> pb-hifiasm-lean-hap2-assembly.fa

samtools faidx pb-hifiasm-lean-hap2-assembly.fa

3.4 Generate checksums

cd "${output_dir}"

for fasta in *.fa
do
  md5sum "${fasta}" \
  | tee "${fasta}".md5
done

## 4ba0c3cc7a7ab090d67047cb22f10bb2  pb-hifiasm-lean-assembly.fa
## 77d48fe452af73d751039245f1bd688e  pb-hifiasm-lean-hap1-assembly.fa
## 21b9cfeb83ec8ce522fcc624873b58eb  pb-hifiasm-lean-hap2-assembly.fa

4 REFERENCES

Cheng, Haoyu, Mobin Asri, Julian Lucas, Sergey Koren, and Heng Li. 2024. “Scalable Telomere-to-Telomere Assembly for Diploid and Polyploid Genomes with Double Graph.” Nature Methods 21 (6): 967–70. https://doi.org/10.1038/s41592-024-02269-8.

Cheng, Haoyu, Gregory T. Concepcion, Xiaowen Feng, Haowen Zhang, and Heng Li. 2021. “Haplotype-Resolved de Novo Assembly Using Phased Assembly Graphs with Hifiasm.” Nature Methods 18 (2): 170–75. https://doi.org/10.1038/s41592-020-01056-5.

Cheng, Haoyu, Erich D. Jarvis, Olivier Fedrigo, et al. 2022. “Haplotype-Resolved Assembly of Diploid Genomes Without Parental Data.” Nature Biotechnology 40 (9): 1332–35. https://doi.org/10.1038/s41587-022-01261-x.