lh3 / miniasm

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

Decreasing minimum required read coverage for assembly #84

Open chocotwig opened 4 years ago

chocotwig commented 4 years ago

Hi,

I observe that while I can manually draw out a single path between, say 5 reads, miniasm only generates a contig that has >2x read coverage at all locations along its length (not including the hanging ends of the terminal reads either). Eg. my 5 reads correspond to a genomic region with the left to right sequence of this region consisting of three segments as such: LEFT-MIDDLE-RIGHT. LEFT and RIGHT have 2x read coverage and MIDDLE, which connects both segments with large overlap, has only 1x read coverage. So miniasm only reports contigs corresponding to LEFT and RIGHT, ignoring the 1x covered MIDDLE. How can I make miniasm also include the 1x covered MIDDLE, and report a contig of the form LEFT-MIDDLE-RIGHT?

Thanks!