Assembly of a mycete genome with HGAP4

[Antony03]

Hello,

I need to assemble a fungus genome for which no information is known. When doing the assembly with canu, I get an assembly of 4918 contigs with a total of 57 123 496 bp. However, with HGAP4, I get an assembly of 2617 contigs with a total of 27,258,314 bp. Why such a big difference? Is it because the genome is diploid and HGAP4 assembles only one haplotype while canu assembles both?

Thank you,

Antony

[rhall]

It's possible that ploidy could account for some of the difference, but it is unlikely that it is that simple. Using default parameters I wouldn't expect Canu to not collapse haplotypes and Falcon to fully collapse, although with the correct parameters and enough divergence between haplotypes both assemblers theoretically could.
Some things to try:
1. Align assemblies to one another to see what the difference actually is, for a quick visulization, http://assemblytics.com/
2. Look at the alternate contigs in the HGAP.4 assembly, <HGAP.4 job dir>/tasks/falcon_ns.tasks.task_falcon2_run_asm-0/a_ctg.fa. HGAP.4 is based on Falcon so does generate contigs for possible alternate haplotypes http://pb-falcon.readthedocs.io/en/latest/
Some more resources on diploid assembly: http://pb-falcon.readthedocs.io/en/l...PacBioHops.pdf
http://pb-falcon.readthedocs.io/en/l...idsPAG2018.pdf

[cement_head]

Originally Posted by Antony03

Hello,

I need to assemble a fungus genome for which no information is known. When doing the assembly with canu, I get an assembly of 4918 contigs with a total of 57 123 496 bp. However, with HGAP4, I get an assembly of 2617 contigs with a total of 27,258,314 bp. Why such a big difference? Is it because the genome is diploid and HGAP4 assembles only one haplotype while canu assembles both?

Thank you,

Antony

Curious as to how long the HGAP4 assembler took and what were the specs on your computer/cluster?