What parameters best resmble blastn

[ghost]

I built a minimap2 index of the NCBI NT database.

I have a series of metagenomic contigs I want to align against said database. Using blast (megablast), around 40% of my sequences have mappings against the nt reference.

Using minimap2 with -x asm10, I only get ~20% contigs with mappings to my reference.

What parameters in minimap2 would best fit this task? I've tried reducing mismatch and gap penalties without success.

[lh3]

What is the exact command line? I would use something close to the default setting like:

minimap2 -c nt.gz contigs.fa

Note that you need to rebuild the index as asm10 uses different -k and -w.

Let me know what you get. Thanks.

[lh3]

PS: asm10 uses -k 19 -w 19. The default uses -k 15 -w 10. It will be much more sensitive to divergent hits. If sensitivity is still an issue, try -k 15 -w 5.

[ghost]

Command line was:
index building:
minimap2 -x asm10 -d nt.mmi nt.gz

mapping
minimap2 -x asm10 -t 64 -K 10G -N 25 nt.mmi contigs.fa.gz > contigs.paf

Trying with default and -k 15 -w 5

[lh3]

Just curious: how is your experiment going? Can minimap2 achieve acceptable sensitivity?

[ghost]

Hi,
Have had some parsing problems with downstream visualization and MD tags, but so far I have had excellent results with k15 w5. Much faster and memory efficient than blast, last,lastz, and malt(http://ab.inf.uni-tuebingen.de/data/software/malt/download/welcome.html).

In terms of mappability, results are consistent with short reads. Now I'm checking if taxonomic profiles are also the same. Will let you know asap

[lh3]

I am closing this for now. When you have more information to share, feel free to post new information here or create a new issue. I will be notified either way. Thanks!