Tuning Parameters for Plant Assembly

[FedericoFontana94]

Hi,

We are looking to optimize the polishing step for reconstructing a plant genome using PacBio HiFi reads (Q30). Specifically, we tested both NextPolish and NextPolish2, followed by a variant calling step on both the raw assembly and the polished genome in order to evaluate the polishing impact.

With NextPolish, we observed a significant reduction in homozygous INDELs (from 3,420 to 945), as expected, but also an increase in homozygous SNPs (from 853 to 2,678). In contrast, using NextPolish2 led to a slight increase in both homozygous INDELs (from 3,420 to 3,741) and SNPs (from 853 to 862).

For NextPolish2, we used the default Yak k-mer sizes (21 and 31) and the default minimap2 alignment settings.

My question is: Are there optimal parameters recommended for polishing a plant genome using NextPolish2?

Thank you in advance for your help.

Best regards,
Federico

[moold]

Can you provide detailed commands and input data statistics, also NextPolish2 version.

[FedericoFontana94]

Thank you for the quick response. Here are the commands we used:

$minimap2
--eqx
-ax map-hifi
-t 20
../../1_Assemblaggio/HIFIASM/q30/raw_assembly.fasta
../../0_Reads/PacBio/q30_hifi_reads.fastq.gz
| samtools sort -@ 20 -o hifi.map.sort.bam -;

samtools index hifi.map.sort.bam;

yak count
-o k21.yak
-k 21
-b 37
<(zcat ../../0_Reads/Illumina_WGS/trimmed_NextPolish2/trimmed_NP2_R1.fastq.gz)
<(zcat ../../0_Reads/Illumina_WGS/trimmed_NextPolish2/trimmed_NP2_R2.fastq.gz);

yak count
-o k31.yak
-k 31
-b 37
<(zcat ../../0_Reads/Illumina_WGS/trimmed_NextPolish2/trimmed_NP2_R1.fastq.gz)
<(zcat ../../0_Reads/Illumina_WGS/trimmed_NextPolish2/trimmed_NP2_R2.fastq.gz);

nextPolish2 -t 20 hifi.map.sort.bam ../../1_Assemblaggio/HIFIASM/q30/raw_assembly.fasta k21.yak k31.yak -o asm.np2.fa

And the input statistics:

• PacBio HiFi reads (=>Q30): Coverage 36.0 X;
• Illumina reads: Coverage 31.7 X;
• Raw Assembly: BUSCO (C:95.6%[S:76.1%,D:19.5%],F:0.3%,M:4.1%​), ContigN50 (6,718,969), Size (694,411,396);
• NextPolish2: 0.2.1 (bioconda env).

best,
Federico

[moold]

1. To maximize correction accuracy, quality filtering steps (fastp) such as adapter removal, global or quality trimming, and read filtering are essential for short reads.
2. We recommend using >=60X short reads. if you only have 30x reads, may be you need to optimize the value of -k.
3. Try to mapping hifi reads using winnowmap, it usually gives better results in our tests.
4. try -r