Assembly result has some 0-depth point of TGS(hifi) data

[WangZhSi]

Hi:

I'm testing difference between assembly using 0.24.0 and 0.19.9 version, and I find some 0-depth assembly point in result of 0.24.0 after mapping TGS data back to p_ctg.fa. Here are the details:

• Software version: 0.24.0
• Data: SRR26555721(watermelon, hifi)
• Assembly parameters: default parameter: hifiasm -o out_prefix -t 32 SRR26555721.fq.gz
• Mapping parameters: minimap2 -x map-hifi -x asm20 --MD -a -t 24 asm0.24.0.bp.p_ctg.fa SRR26555721.fq.gz > out.sam; and then convert out.sam into sort.bam

Visualize bam using igv, here is the result:

It did not happen in version 0.19.9:

I've check this 0-depth area of 0.24.0 and found this part is repetitive unit of telomere(CCCTAAA).

Do you have any idea how this happened? Do you have any suggestion?

Thanks!

[chhylp123]

Do you think it is a misassembly?

[WangZhSi]

@chhylp123 Thank you for your response!

I conducted further checks on ctg0001, and here is the feedback:

Assembly size:

• v0.24.0: 35,325,984 bp
• v0.19.9: 35,015,304 bp

It can be observed that the v0.24.0 assembly is approximately 300 kb larger. Initially, I thought this difference might be caused by low-depth regions, but after checking the number of telomeric repeat units (CCCTAAA), I found:

• v0.24.0: 2,684 times
• v0.19.9: 1,529 times

This difference does not fully explain the 300 kb size increase. Subsequently, I used mummer4 to compare ctg0001 between the two versions, and the results are as follows:

It appears that the assembly results are largely consistent between the two versions. However, I am still uncertain how to interpret the uncovered regions(0-depth regions) and even wonder if it could be an issue with alignment.

Do you have any suggestions?

[chhylp123]

Could you check which reads hifiasm is using for this region by looking at the corresponding A-lines?

[WangZhSi]

@chhylp123 Do you means A-lines in .gfa file? Sorry I'm not very familiar with this file.

For 0.24.0 assembly, the used reads is SRR26555721.2353299

But in mapping paf file, this reads align to ctg0006

I also check this in assembly from 0.19.9, the reads used for ctg0001 started region mapped back correctly.

Thanks for your reply!