Telomere-to-telomere (T2T) genome has been emerging as a new hotspot in the field of genomics. Typically, we obtain a T2T genome with datasets including both high-accuracy PacBio HiFi long reads and Oxford Nanopore Technologies (ONT) ultra-long reads. Although genomes obtained using HiFi long reads have considerably higher qualities, however, they still contain a handful of assembly errors in regions where HiFi long reads stumble as well, such as homopolymer or low-complexity microsatellite regions. Additionally, a typical gap-filling step is accomplished using ONT ultra long reads which contain a certain amount of errors. Hence, the current T2T genomes assembled still require further improvement in terms of consensus accuracy. NextPolish2 can be used to fix these errors (SNV/Indel) in a high quality assembly. Through the built-in phasing module, it can only correct the error bases while maintaining the original haplotype consistency. Therefore, even in the regions with complex repeat elements, NextPolish2 will still not produce overcorrections. In fact, in some cases it can reduce switching errors in the heterozygous region. NextPolish2 is not an upgraded version of NextPolish, but an additional supplement for the pursuit of extremely-high-quality genome assemblies.
If you are concerned about the overcorrection problem, please refer to the HG005 dataset benchmarking and the NextPolish2 article for more information.
- Installation
- General usage
- Getting help
- Citation
- License
- Limitations
- Benchmarking
- Overcorrection
- FAQ
conda install nextpolish2NextPolish2 is written in rust, try below commands (no root required) or refer here to install Rust first.
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | shgit clone --recursive git@github.com:Nextomics/NextPolish2.git
cd NextPolish2 && cargo build --releasecd test && bash hh.shNextPolish2 takes a genome assembly file, a HiFi mapping file and one or more k-mer dataset files from short reads as input and generates the polished genome.
#prefer using winnowmap
meryl count k=15 output merylDB asm.fa.gz
meryl print greater-than distinct=0.9998 merylDB > repetitive_k15.txt
winnowmap -t 5 -W repetitive_k15.txt -ax map-pb asm.fa.gz hifi.fasta.gz|samtools sort -o hifi.map.sort.bam -
# or mapping using minimap2
# minimap2 -ax map-hifi -t 5 asm.fa.gz hifi.fasta.gz|samtools sort -o hifi.map.sort.bam -
# indexing
samtools index hifi.map.sort.bam- Prepare k-mer dataset files (yak). Here we only produce 21-mer and 31-mer datasets, you can produce more k-mer datasets with different k-mer size.
# produce a 21-mer dataset, remove -b 37 if you want to count singletons
./yak/yak count -o k21.yak -k 21 -b 37 <(zcat sr.R*.fastq.gz) <(zcat sr.R*.fastq.gz)
# produce a 31-mer dataset, remove -b 37 if you want to count singletons
./yak/yak count -o k31.yak -k 31 -b 37 <(zcat sr.R*.fastq.gz) <(zcat sr.R*.fastq.gz) - Run NextPolish2.
./target/release/nextPolish2 -t 5 hifi.map.sort.bam asm.fa.gz k21.yak k31.yak > asm.np2.fa
# or try with -r
# ./target/release/nextPolish2 -r -t 5 hifi.map.sort.bam asm.fa.gz k21.yak k31.yak > asm.np2.faOptional: If your genome is assembled via trio binning. You can discard reads that have different haplotype with the reference before the mapping procedure, see here for an example.
Use ./target/release/nextPolish2 -h to see options.
Feel free to raise an issue at the issue page.
Note: Please ask questions on the issue page first. They are also helpful to other users.
For additional help, please send an email to huj_at_grandomics_dot_com.
Jiang Hu, Zhuo Wang, Fan Liang, Shan-Lin Liu, Kai Ye, De-Peng Wang, NextPolish2: A Repeat-aware Polishing Tool for Genomes Assembled Using HiFi Long Reads, Genomics, Proteomics & Bioinformatics, 2024, qzad009, https://doi.org/10.1093/gpbjnl/qzad009
NextPolish2 is only freely available for academic use and other non-commercial use.
- NextPolish2 can only correct the regions that are mapped by HiFi reads. For regions without HiFi reads mapping (usually cause by high error rate), you can try to adjust mapping parameters.
- The performance of NextPolish2 relies heavily on the quality of short reads. Please use high-quality short reads to avoid overcorrection errors, which can falsely improve estimated QV but reduce actual accuracy.
- NextPolish2 can only fix some structural misassemblies.
| Source | Software | QV | Switch error rate (‱) |
|---|---|---|---|
| A. thaliana | Hifiasm (primary) | 47.67 | 1.99 |
| ^(simulated data, primary contigs)^ | NextPolish2 | 65.42 | 0.35 |
| A. thaliana | Hifiasm (primary) | 58.03 | |
| ^(Col-XJTU, primary contigs)^ | NextPolish2 | 64.26 | |
| H. sapiens | Hifiasm (primary) | 60.25 | 0.15 |
| ^(HG002, primary contigs)^ | NextPolish2 | 62.87 | 0.14 |
| H. sapiens | Hifiasm (trio) | 59.77 | 0.21 |
| ^(HG002, paternal contigs)^ | NextPolish2 | 63.49 | 0.20 |
| H. sapiens | Hifiasm (trio) | 59.78 | 0.33 |
| ^(HG002, maternal contigs)^ | NextPolish2 | 63.29 | 0.30 |
In addition to evaluating the overcorrection problem discussed in the NextPolish2 article, we used the HG005 data to further assess this issue. First, we assembled the HG005 genome with approximately 30x HiFi data using HiFiasm, followed by polishing the assembled genome with NextPolish2. To minimize the impact of evaluation method limitations, we employed three approaches to evaluate the genome's accuracy before and after polishing:
- Merqury: To assess the quality value (QV).
- DeepVariant: To count homozygous high-quality variants as potential errors.
- Paftools.js: To count variants between GRCh37 and HG005 that are not in the high-confidence benchmarking variants (GIAB) as potential errors.
The results demonstrated that NextPolish2 improved the QV of the assembled genome from 53.8544 to 55.6257 and reduced the number of homozygous high-quality variants (potential errors) from 14,155 to 6,955. Additionally, NextPolish2 increased the number of high-quality variants called from 2,520,470 to 2,522,817 and reduced the error rate from 0.029927% to 0.029575%.
Overall, these results indicate that NextPolish2 effectively reduces the error rate of the assembled HG005 genome. Detailed step-by-step instructions are available here.
You can track updates by tab the Star button on the upper-right corner at the github page.