lh3 / miniasm

Ultrafast de novo assembly for long noisy reads (though having no consensus step)
MIT License
303 stars 69 forks source link

Parameter tweaking for assembling error-corrected reads #16

Closed suryasaha closed 8 years ago

suryasaha commented 8 years ago

I am assembling error corrected pacbio reads from canu for a heterogeneous sample (pool of sexually reproducing individuals) using parameters in the README. Raw coverage was 80X that reduces to 26X after correction. I don't get any better assemblies with the error corrected reads.

Would you recommend any alternative parameter choices when handling error corrected pacbio reads? Thanks

lh3 commented 8 years ago

I don't get any better assemblies with the error corrected reads.

That was also my experience on bacterial assembly: N50 and misassembly rate are about the same with or without PBcR/falcon_sense error correction. I tried different settings that time, but the final assembly was not obviously improved given error corrected reads.

suryasaha commented 8 years ago

Same experience here but this is an insect genome. Is there any setting to tweak minimap to handle a heterozygous sample?

lh3 commented 8 years ago

Probably not. The error rate of error corrected reads and minimap's resolution of sequence differences are higher than heterozygosity.

suryasaha commented 8 years ago

Thank you for the quick responses @lh3 !