contigN50 too small

[shiyi-pan]

Hi, I've used wtdbg2 to assemble a soybean genome with 62-fold Hifi data. The resulting fasta file has 4604 contigs, with contigN50 of 595316 bp, which is too small.
Here is my script,
wtdbg2 -x ccs -g 1144.6m -t 8 -i Glycine_max.ccs.bam.fasta.gz -fo wtdbg2
wtpoa-cns -t 8 -i wtdbg2.ctg.lay.gz -fo wtdbg2.raw.fa
Could you give me some advise to improve contigN50 ? Thank you very much.

[ruanjue]

The results look so bad. Have you tried hifiasm?

[shiyi-pan]

Yes, the hifiasm obtains an assembly which has 1389 contigs with contigN50 of 26905840 bp using the same data.
Here is the hifiasm scripts:
hifiasm -o NN1138_hifiasm -t 8 Glycine_max.ccs.bam.fasta.gz -l0 -f0

Could you give me some advises to get an better assembly with wtdbg2.

[ruanjue]

ccs: -p 21 -k 0 -AS 4 -K 0.05 -s 0.5
Try -p 0 -k 19/21, try -s 0.6/0.7 ....

[shiyi-pan]

Hi, thank you for your reply. Do you means I run the following commands one by one and choose the best one?
wtdbg2 -x ccs -g 1144.6m -t 8 -i Glycine_max.ccs.bam.fasta.gz -fo wtdbg2 -p 0 -k 19 -s 0.6
wtdbg2 -x ccs -g 1144.6m -t 8 -i Glycine_max.ccs.bam.fasta.gz -fo wtdbg2 -p 0 -k 19 -s 0.7
wtdbg2 -x ccs -g 1144.6m -t 8 -i Glycine_max.ccs.bam.fasta.gz -fo wtdbg2 -p 0 -k 21 -s 0.6
wtdbg2 -x ccs -g 1144.6m -t 8 -i Glycine_max.ccs.bam.fasta.gz -fo wtdbg2 -p 0 -k 21 -s 0.7
Thank you again.

[shiyi-pan]

Hi, I tried the following script and the contig number is 5601 and N50 is 424654 bp.

wtdbg2 -x ccs -g 1144.6m -t 8 -i Glycine_max.ccs.bam.fasta.gz -fo NN1138_wtdbg2_test1 -p 0 -k 19 -s 0.6
wtpoa-cns -t 8 -i NN1138_wtdbg2_test2.ctg.lay.gz -fo NN1138_wtdbg2_test2.raw.fa

[ruanjue]

Please decrease -s to 0.4/0.3 maybe error rate of those CCS is a bit higher.

[ruanjue]

Please decrease -s to 0.4/0.3 maybe error rate of those CCS is a bit higher.

[shiyi-pan]

I really appreciate your reply and will try -s 0.4/0.3,ruanjue.

[shiyi-pan]

Hi, I have tried the '-p 0 -k 19 -s 0.4' parameter but not performed worse with contig N50 of 141547 bp.
Here is my logs. Could you give me some other advise, thank you a lot.
184279.err.txt

[ruanjue]

From the log file, it looks a highly repetitive genome. It is better to increase -k to 21/23, also try to increase the -s to 0.5/0.6/0.7. Any progress, please send the full log file.

[shiyi-pan]

Thank you for your help, Dr. ruanjue. This time I tried the -p 0 -k 23 -s 0.7 and got a best assembly with longset total length (905541315) and contig N50(724860). But still shorter than other Assemblers' results.
The specie I am studying is soybean. It has 50% TEs in genomes.
Here is my logs. I will try -p 0 -k 23 -s 0.6.
199135.err.txt
199135.out.txt
thank you again.

[ruanjue]

Also try change --node-max to a larger value. You can --load-alignments from previous results, and set different --node-max. Don't invoke wtpoa-cns before you get fine result, because it is the most time-cost.

[shiyi-pan]

Thank you for your help, Dr. ruanjue, sorry for relpy late. I didn't know the --load-alignments parameter, so I deleted the alignment file for saving hard drive space.
I tried the -p 0 -k 23 -s 0.6 but it's worse than before. The final assembly has 889781517 bp with contigN50 of 527048 bp. Here is my log file.
251452.err.txt
268921.err.txt