search

lh3 / miniasm

Ultrafast de novo assembly for long noisy reads (though having no consensus step)
MIT License
296 stars 68 forks source link

Performing local assembly with only a few number of reads #62

Open OscarAspelin opened 5 years ago

OscarAspelin commented 5 years ago

Hi, is it possible to perform targeted assembly with a few number of reads? Say that I have 50 reads that overlap, Is it possible to assemble only these 50 reads? In that case, what parameters are suitable for this?

ekg commented 5 years ago

I'm interested in knowing if this is possible. I've made past attempts but have found this difficult to do with miniasm. I'm sure @lh3 will have suggestions.

With this low number you might consider constructing the string graph directly. You can do this with minimap2 and seqwish, and view the resulting GFA in Bandage. But this won't give you a single consensus unless someone has written a GFA to consensus generator. You might get what you want using polishing on the read set first and the applying the minimap2/seqwish pipeline. It all really depends on what you're after and the type of input reads.

OscarAspelin commented 5 years ago

Cool, I'll check that out! Do you know if there is another long read assembler that supports this kind of assembly approach?

ekg commented 5 years ago

Maybe wtdbg2 would do it?